Subscribe now

Two AI Giants, Two Very Different Healthcare Bets

Last week, OpenAI and Anthropic both announced major healthcare integrations within days of each other. On the surface, they look pretty similar. Both let you connect medical records to an AI chatbot, both promise HIPAA compliance.

But look closer and you see something more interesting: these companies are making fundamentally different bets about where AI actually creates value in healthcare.

The Two Strategies

OpenAI’s bet: become your health assistant. They’re building on their massive consumer base (230 million people already ask ChatGPT health questions weekly). Now you can upload lab results and get research-backed explanations with citations. They’re also rolling out enterprise tools for hospitals like Cedars-Sinai. The strategy? Own both ends of the market.

Anthropic’s bet: solve hospital paperwork. Prior authorization. Insurance appeals. Medical coding. They’re plugging Claude directly into the filing systems hospitals already use (CMS databases, ICD-10 codes, provider registries). It’s about speed and operational efficiency, not research synthesis.

Why I Think OpenAI’s Approach Matters More

So here’s the thing. Anthropic is doing valuable work, but it’s kinda incremental optimization of an existing broken system. OpenAI is addressing something more fundamental: broad access to healthcare information. In the US, that information exists online but requires sophistication to access and understand. Better accessibility is a crucial first step.

The Real Vision: AI as Your Health Documentation Partner

But the real vision isn’t just ChatGPT answering health questions. It’s AI becoming a documentation partner. Think of it like having a nurse following you around all day, logging symptoms, tracking patterns through your phone and smartwatch data.

Think about what doctor visits actually are right now: mostly information gathering. You describing symptoms, forgetting details, trying to remember when things started. With AI documentation, those visits shift to strategy and game planning. That’s genuinely powerful.

What Version Two Looks Like

And version two isn’t far off. Continuous monitoring detecting anomalies, active communication asking for context. AI notices elevated heart rate and asks what you were doing. You provide context. That full history goes to your doctor with pattern analysis already done.

The Access Question

Will there be liability issues? Sure. Access inequality concerns? Probably at first. But technology always starts expensive and concentrated, then gets subsidized and distributed broadly. Internet. Smartphones. Solar panels. Wealthy people willing to pay for the first version fund the innovation that eventually makes it cheap for everyone.

You have to start somewhere.

The Bottom Line

The current healthcare system can’t stay as is. Something has to break or change. This technology could be the catalyst that forces innovation across the entire ecosystem.

Neither company’s strategy is obviously wrong. The hospital of the future might use both. But I’m betting on the approach that reimagines the relationship between patients and their health information, not the one that makes existing paperwork slightly less painful.

Subscribe now

Subscribe now

Hello and welcome to week 2. Last week was busy and Claude Code has opened up a whole new world for me as far as building apps goes. Here’s what I got up to last week:

Progress updates

Current projects

Nova Music

Nova got a ton of love this week. The big addition was converting the student portal to a family-based system. Now families log in once and see all their kids’ schedules, assignments, and practice logs in one place with color-coded events. I also built out a full billing portal where families can view and pay invoices through Stripe, complete with payment confirmation emails.

Other additions include an assignment checklist system for tracking practice goals, a teacher resources page with drag-and-drop file uploads and in-page previews, and magic link emails for easy family onboarding.

Experimentation Labs Site

I ended up splitting out the newsletter.experimentationlabs.com and root experimentationlabs.com websites so that I can use the root one as more of a portfolio website. I added two new tools to the site: a Test Plan Generator with AI-powered suggestions and an A/B Test Analyzer with statistical significance calculations. Also revamped the navigation with a megamenu and added SEO basics (robots.txt, sitemap, Open Graph tags).

Onewheel Utah

Bringing an old project back to life! Years ago I ran a small Onewheel website I built with WordPress, but life got busy and it’s been hard to want to keep managing or paying to manage this. So this week with Claude I built a full custom WordPress theme with Tailwind and DaisyUI for a revamped website and marketplace. So that’s back to some of it’s former glory, but with a whole new architecture. I don’t know how much love this one will get going forward, but its a fun side project. I’d love to build some AI Agents to automate content management and stuff like that, so we’ll see what time permits.

What’s coming up

I’m going to be out of town for most of this week, so I doubt there will be too much that gets done here, my major focus is on building a GTM plan for Nova so that I can start to onboard beta testers and be ready for launch at the end of Q1, so probably not too many visible updates.

• Upcoming posts

  • Tech News Tuesday - OpenAI and Anthropic step into healthcare

  • AI Wednesday - Vibe code to live guide

  • Growth Thursday - When is friction in your product a good thing?

  • Fail Friday - The losing loop of carnival games

Last week’s posts

• Using Claude Opus 4.5 to build Nova Music

• A Difficult Tech Job Market + 2026 Predictions

• How I’m Using Cursor and Granola to Get Better at Interviewing

• Friday Fail: On Sale Now!

If you’re building a product and experimenting with AI as a force multiplier, follow along. I’ll keep sharing concrete, real examples.

Subscribe now

Subscribe now

He ran over and found a bunch of individual Pokemon cards. He looked up and saw the sign that said “on sale $0.99”. Right below that he saw a sign that said “Used to be $1”.

He laughed and said “Dad that’s not a sale, it’s only $0.01 cheaper!”

Sorry Gamestop, even my 7 year old didn’t fall for it.

In other news…

Happy Friday!

Subscribe now

No one tells you if you’re bad at interviewing.

The recruiter won’t call back and say, “Hey, you bombed that one.” The hiring manager won’t explain what they were looking for or whether you had it. You’re left guessing. And that’s a problem, because I might have the skills but not know how to talk about them.

So I built a system to get feedback despite that.

The Setup

I use two tools: Granola and Cursor.

Granola records and transcribes my interviews. Once I have the transcript, I bring it into Cursor along with a bunch of other context: my resume, the job description, notes on the company, and something I call my Career Source of Truth.

Why Cursor?

Because it lets me give Claude Code all the context it needs. The job. The company. The interviewer. My full career history. The actual conversation. With all of that in one place, Claude can give me real feedback on how the interview went.

The Career Source of Truth

Here’s the thing about my career: it hasn’t always been pretty.

• I’ve had bosses I liked and bosses I hated

• I’ve had roles I loved and roles I tolerated

• I’ve worked at companies I’m proud of and companies I’m not

All of that matters. But I’m not going to say it in an interview. In an interview, I’m the product. I’m selling myself as the solution to their problem. I don’t lead with the flaws. Nobody selling a phone tells you about the bad battery life. I focus on the highlights. There’s nothing wrong with that.

Here’s how I built my Career Source of Truth:

1. I talk through my whole career out loud using a tool like Wispr Flow. Every job. Every boss. Every win and every failure.

2. I dump all of that into an LLM and ask it to ask me more questions until we have a full picture.

3. I save that document and upload it to Cursor.

I’ve also added sections with prepared stories for common interview questions. I know what types of questions I’ll get for product and growth roles. Having those stories ready makes a huge difference.

The File Structure

Here’s the folder structure I use. Copy it and make it your own:

📁 job-search/
├── 📁 .claude/                     # Claude Code settings (optional)
├── 📁 interview-transcripts/
│   ├── 📁 company-a/
│   │   ├── job-description.txt
│   │   ├── company-research.txt
│   │   ├── interviewer-profile.txt
│   │   ├── recruiter-screen.txt
│   │   ├── hiring-manager-interview.txt
│   │   └── technical-interview.txt
│   ├── 📁 company-b/
│   │   └── ...
│   └── 📁 company-c/
│       └── ...
├── 📁 resumes/
│   ├── base-resume.txt
│   ├── growth-pm-version.txt
│   ├── core-pm-version.txt
│   └── hybrid-version.txt
├── career-source-of-truth.txt
├── drivers-and-drainers.txt
└── personality-assessments.pdf      # CliftonStrengths, MBTI, etc.

What goes where:

• interview-transcripts/ : One subfolder per company. Job descriptions, research, interviewer profiles, and transcripts from each conversation.

• resumes/ : Different versions of your resume for different role types.

• career-source-of-truth.txt : Your full career history, including the stuff you wouldn’t say out loud.

• drivers-and-drainers.txt : What gives you energy vs. what drains you.

• personality-assessments.pdf : Optional, but useful for feedback on role fit.

Career Source of Truth Template

Here’s a template you can copy. The goal is to create an honest record of your career that you can pull from when prepping.

markdown

CAREER SOURCE OF TRUTH
======================
Last updated: [Date]

Purpose: Unbiased record of career history, accomplishments, and context.
Use this to pull details for resumes, cover letters, and interview prep.

========================================================================
[COMPANY NAME] ([Start Date] - [End Date])
========================================================================

COMPANY CONTEXT
- What does the company do?
- Stage/size (Series A, public, etc.)
- Approximate headcount
- How did you leave? (quit, laid off, etc.)

REPORTING STRUCTURE
- Who did you report to? (title and name)
- What was the relationship like? (good, bad, micromanager, hands-off)

TEAM
- Direct reports? How many?
- Cross-functional collaborators?
- What functions did you work with most?

TOOLS
- Analytics, experimentation, product tools
- Other relevant tech

KEY PROJECTS

[Project Name]:
- Problem: What was broken or missing?
- Solution: What did you do?
- Your role: What did YOU own?
- Result: Quantified outcome (%, $, time saved, etc.)

KEY METRICS
- List your most impressive numbers
- Tie them to business outcomes where possible

STAKEHOLDER CHALLENGES
- Any notable conflicts or alignment issues?
- How did you resolve them?

WHY YOU LEFT
- Be honest, this is for your reference

HOW YOU FELT ABOUT THE ROLE
- What did you like? What did you hate? Would you go back?

========================================================================
[REPEAT FOR EACH ROLE]
========================================================================

========================================================================
EDUCATION & CERTIFICATIONS
========================================================================
- School, degree, dates
- Relevant coursework or skills learned
- Online courses, bootcamps, certifications

========================================================================
SKILLS & TOOLS
========================================================================
- Technical (languages, platforms)
- Analytics tools
- Product tools
- AI/prototyping tools

========================================================================
PERSONAL CONTEXT
========================================================================
- Location, relocation preferences
- Family context relevant to career decisions
- Interests and side projects that demonstrate skills

========================================================================
CAREER NARRATIVE THEMES
========================================================================

Use this section to identify patterns across your career:

PATTERN: [Theme name]
- Evidence from Role 1
- Evidence from Role 2
- Why this matters

Examples of patterns to look for:
- Promoted fast
- Self-directed learner
- Types of work you gravitate toward vs. avoid
- Recurring strengths and challenges

========================================================================
INTERVIEW Q&A - PREPARED ANSWERS
========================================================================

TOP STORIES (have 5-6 ready):
1. [Story name] > [Key metric] ([Category])
2. [Story name] > [Key metric] ([Category])
3. [Story name] > [Key metric] ([Category])

WORDS/PHRASES TO STOP SAYING:
- [Filler words you overuse]
- [Weak phrases]

WORDS/PHRASES TO START SAYING:
- "I owned..."
- "I drove..."
- "The result was..."

========================================================================
COMMON QUESTIONS CHEAT SHEET
========================================================================

Q: Tell me about yourself
> [Your 30-second pitch]

Q: Why this role?
> [Your answer]

Q: Tell me about a project you led
> [Your go-to story]

Q: Tell me about a failure
> [Your story, own it, show learning]

[Add more questions specific to your field]

========================================================================
STORIES BY THEME - QUICK REFERENCE
========================================================================

[THEME 1: e.g., Cross-functional collaboration]:
> [Story + key metric]

[THEME 2: e.g., Data-driven decisions]:
> [Story + key metric]

[YOUR WEAK SPOT, have stories ready]:
> [Story that addresses the gap]

========================================================================
THINGS TO AVOID SAYING
========================================================================
- [Topics or phrases that don't land well]
- [Anything that raises red flags]

How I Use It

Once I have all this context in place, I ask Claude: “Here’s my most recent interview. Based on the job description, company info, interviewer background, my resume, and my Career Source of Truth: how did this go? What went well? What didn’t land?”

Claude gives me real feedback. What I did well. What fell flat. Moments where I could have said something different.

Where This Shines

This works for single interviews. But it gets powerful when I stack up multiple interviews, either multiple rounds at one company or similar roles across different companies.

Claude can:

• Spot trends across interviews (”When you say X, people respond well. When you say Y, it falls flat.”)

• Help me prepare for likely questions based on patterns

• Dig into my Career Source of Truth and say, “Someone asked you this question and you didn’t have a good answer. But you have a perfect story for it. Here’s how to frame it.”

I’ve also uploaded my CliftonStrengths assessment. Claude can read it and tell me my top strengths and blind spots. I have a separate doc with my drivers and drainers: what gives me energy and what depletes it. With all that context, Claude can tell me which roles I’ll thrive in and which ones will drain me.

Evaluating Them, Not Just Selling Yourself

Here’s an underrated benefit: Claude helps me evaluate companies, not just pitch myself.

It might say, “Company X looks great on paper, but there are red flags in these transcripts. In your next interview, dig into this topic that seems unresolved.”

That’s huge. It helps me ask better questions. It helps me judge whether a company is a good fit for me, not just whether I can convince them I’m a fit for them.

Bringing It Home

This system has been a game changer: a file structure with all my context, an LLM that can analyze multiple interviews, and feedback that’s (hopefully) unbiased.

This matters to me because I get imposter syndrome. I think, “I can do this job. But I don’t know if I speak the language. I don’t know if I can sell myself for this role, even though I have the experience.”

This system helps me frame my skills the right way. I know what I’ve done. Claude helps me shape it for the person I’m talking to. Because I have to tell different stories to different people. Design, marketing, product, engineering, executives: they all want to hear the same story, but with the right angle. Good PMs are persuasive. You tell the same story in a way that lands with each audience.

I’m not pushing fraud or deception. But in interviews, I often have the experience. I just don’t know how to talk about it in a way that clicks with the person across the table.

This system fixes that.

Subscribe now

Subscribe now

Between really quick advancements in AI, a very strange economy and tech companies doubling down on profits and AI Agents, being a tech worker kind of sucked for the last few months.

So where does that leave tech workers coming in to 2026?

Here are my predictions:

The tech market will bounce back, at least a little.

Companies doubled down hard on AI last year and used “AI” as a way to lay people off, but end of year lay offs have been happening in tech for many many years, AI was just the flavor of the quarter in 2025. And I should point out, I’m a big believer in AI, it’s changing a lot about how I work in some really cool ways, but I think lots of founders had big visions of how it would allow them to automate everything and that just didn’t happen, at least not last year.

So I think tech will hire more, but probably not as much as in previous years.

AI isn’t taking your job, but someone fluent with AI might take it

While I do think there will be more hiring in 2026, I think teams are trending towards flatter and leaner empowered by AI to do more. And I get it, as a product manager I can use AI to write code, query databases, research and more. It doesn’t solve all my problems right now, but it does allow me to operate faster across a wider breadth than ever before, so while I think companies will hire, I think AI is a requirement, in fact in every single interview I’ve had recently hiring managers and teams are asking me how I’m using AI in my work. Thankfully I have good answers, but I can imagine that if someone walked in and said they don’t like AI or aren’t on the hype train, that person might struggle to land a gig.

The product management career will have a serious change in scope

I can really only speak to the product and growth sides of things, but from what I can see, engineering and design are not the bottlenecks anymore, thanks to AI, good ideas and planning are. I think we’ll see a switch from 1 PM and 2-10 Eng per squad to 0.5 PMs to 1-2 Eng. I think there will need to be a change to a business focus, more of a general manager mentality. Products and features being built can happen in record time, so the thinking and ensuring user experience is the real bottleneck.

Engineering roles will change dramatically

Though I’m not an engineer, if I were a new or junior software engineer just getting into the game, I would start to feel nervous and would want to ramp up quick. AI is changing engineering in some serious ways that can’t be denied anymore. With Opus 4.5 I’m building fast a full web app that I plan to launch this year. Writing code isn’t the crucial skill anymore, but understanding how to architect systems and build for security are things that I still need. I imagine engineers will change their work to write less code, review a lot more AI written code and have to master a new abstraction layer that doesn’t require writing everything yourself, but requires understanding systems and integrations and security in a new deeper way.

What are your thoughts? What are your predictions for 2026? Is AI taking all our jobs? Or just a passing fad?

Subscribe now

Subscribe now

I’ve had this idea since my wife started teaching piano when we moved to California. An app to help her with admin tasks, primarily billing, scheduling and lesson management.

I started by looking at competitors, there are a few, and trying to understand what the solutions to this problem looked like for them. Then I tried for months with various LLM coding models and vibe coding tools to build a version of this that works.

I began building with Bolt and Lovable, and with Sonnet 4.5 and Codex was able to make decent progress, but always bumped into a few of the same issues, namely creating a semi-separate system for student login that kept them connected to the teacher, basically relatively complicated database relationships, and then other small little features or bugs, usually inconsistent. But always present.

So I spent months in Bolt and Lovable trying different iterations of prompting to get around the problems, but could never get the LLMs to cross the finish line.

Enter Opus 4.5 + Bolt. Once Bolt had these enhanced capabilities, there were absolutely 0 problems. Tough database structure? No problem. Change the CSS from Tailwind to Shadcn? Barely 30 minutes of work. Want a Stripe integration? Easy, I’m the biggest bottleneck, not the AI.

It opened a whole new world of code to me.

So in about a week I went from this idea that I knew well, I knew the problem, the solution, what I wanted it to look like, how I wanted it to feel and work, and finally I could bring it all to life with minimal effort and cost.

The alpha is live now at trynovamusic.com if you’re interested and want to take a look, I’d love your feedback.

There are still plenty of features, bugs and integrations that need clean up and fixing, but I feel pretty comfortable launching this as an alpha.

Big takeaways:

1. Code isn’t the bottleneck anymore, I am.

2. Ideas are becoming more valuable than execution very quickly, execution is still a bottleneck, but it’s mostly about what I know or don’t know, AI can solve most of the problems I throw at it from a code perspective.

3. The barrier to entry for digital products is getting lower and lower, I’m not sure what that means going forward, but its great for someone like me who has lots of ideas on what I’d like to build. It’s also a bit of a forcing function, now that I can build anything, what should I actually build?

Subscribe now

Hello and welcome to my first weekly update of 2026!

One of my goals for 2026 is to write, publish and build more in public and posting a weekly update is one of the ways I plan to hold myself accountable to those goals.

I’m pulling some inspiration from one of my favorite authors, Brandon Sanderson, who posts a weekly update regularly that I end up watching mostly every week and I really enjoy.

In my weekly updates you can expect to get updates on projects I’m working on, upcoming projects, posts, etc for the coming week, a brief report on what I did last week and occasionally other updates. Should be short and sweet and the plan is for it to come out every Monday.

Expect a lot more posts from me this year.

Progress updates

As far as updates go on projects, there are 2 that I am currently working on:

1. Nova Music

2. Experimentation Labs

Nova Music is a music instructor CRM and studio management tool that I’ve been working on for my wife who teaches Piano. It started as a fun way to play with AI and vibe coding tools and has turned into something real that I’m very excited about.

It’s still currently in Alpha, but I’m excited that the core functionality of the product itself is working! There are still a few bugs to clean up and lots of feature enhancements I want to make, but it’s in a place where my wife is going to start using it in her lessons starting next week.

I expect she’ll have lots of valuable feedback that I’ll incorporate as quickly as I can.

My goal is to get Nova to some Beta users in February and to try to go to market by March.

There’s still plenty to do before we get there, but I think it’s doable.

Experimentation Labs is my Substack newsletter.

I’ve written on and off for a few years, it’s something I enjoy and a skill I want to improve, but it’s tough to be consistent, so one of my goals for 2026 is to post more often.

I’m hoping to post quite a bit more. My posts will include Weekly Updates, AI news and my takes on AI tools, breakdowns of growth strategies, go to market strategies and interesting products.

I don’t want to over commit to specific topics yet, those are still up in the air, but I plan to try several things and we’ll see what sticks with the audience and what gets me excited to study and write about.

I’m excited about this goal, I hope it will stretch me in some uncomfortable ways.

What’s coming up

This week I’m excited to write about :

1. How I’m using Claude Code Opus 4.5 to build Nova

2. How I’m using Cursor and Granola to improve my job interview performance

3. A difficult tech job market

4. Vibe code to live quick guide

Happy New Year

Here’s to a happy new year and doing our best to stick to our goals!

As someone who’s been working in tech for about a decade, I’ve seen my fair share of layoffs and as a result I’ve developed a pretty unique skill of landing jobs at some of the most competitive companies in the world.

Today I want to share 4 ways that I’m using AI to help me land new roles:

Subscribe now

Use ChatGPT Agent to develop a company profile

Before I start working at any company I want to try to get a read on how the company is doing financially, who’s running the show and if there are any other red flags, or green flags, I should be aware of. So I run a prompt like this and read the report.

1. Prompt:

   1. I’m interviewing with a company called [company name] today, I’m chatting with a person named [name of person] for a [name of role] role. Can you do some prep for me? I want to know about the company, the products, the job description and expectations, the funding, the valuation, the revenue, the Glassdoor employee reviews, the company trajectory, any recent changes in employee headcount, any notable former employees who were only there for a few months or other red or green flags. I want a critical but fair review of their everything (include sources) because I want to know if this is a place I want to work and if I can succeed there.

Create a ChatGPT project, upload relevant info then have it run a report

One of my favorite ChatGPT features is the ability to create a project. You can give the project custom prompt instructions and files for it to reference, you can even contain it’s memory to the project, which is super helpful. Give this a try:

1. Create a project, add the job description, the hiring manager info, the recruiter info and my resume and then get ChatGPT agent to research them

2. Prompt:

   1. Your task is to learn everything you can about [person name] and compile it into a prep doc for me. I’ll be interviewing with them for a [role] at [company] and want to be well informed. Help me understand how they think, write, communicate and the kinds of questions they’re likely to ask me in the interview so I can prepare. Here’s their LinkedIn [linkedin link]

Interview ChatGPT voice mode

Once I have all these reports on the company and the individuals I’m interviewing with, I upload those to my project and then have a voice conversation with it. It’s helpful to do this from 2 perspectives, me as the interviewee and as the interviewer, I think both help me wrap my head around how the conversation is going to go. Try this:

1. Prompt (I read this explicitly to make sure I get it right):

   1. For this conversation, assume the role of [person name], [job title] at [company]. You’re screening me for the role. Ask me exactly five questions to determine my fit for the position based on [company]’s needs. I want you to then, after you ask these five questions, pause the conversation and provide [person name]’s perspective, highlighting clearly what went well in my responses, what was acceptable but could be sharper, and what specifically needs improvement to best position me as a competent [role] at [company]. Do not factor conciseness into your feedback. Focus solely on the clarity of competence, role alignment, and suitability for the [role] at [company].

      1. I add the note about conciseness because every time I’ve tried this it tells me I talk too much, and I find that ridiculous. I think it thinks my answers should be like 1 sentence, but I find that doesn’t allow me to actually articulate what I need to, but who knows, maybe I talk to much and AI is right.

Record the conversation and review it

I try to record the conversation if possible, but worst case scenario, I just record my side of the conversation and ask it to fill in the gaps, I can help if it needs. Once I have that it’s really helpful to debrief and see how I did in the real interview. Try this:

1. Prompt:

   1. <upload the transcript of the conversation> Just finished my interview with [company], this was with [person and role], how did I do? Give me a critical but fair opinion on what went well, what was ok and what I totally failed on.

AI Powered Interviewing Is Awesome

I find leveraging AI to help me prep and debrief interviews to be incredibly helpful. It’s really helpful to understand both perspectives, and to get specific feedback on what I can be doing better and how to improve.

Let me know if you give these a try and tell me what you think!

Every Product Manager has built something that should have worked. The spreadsheet math was bulletproof. The execution was flawless. You shipped innovation after innovation.

But sometimes great metrics hide a dying business model.

Subscribe now

Day 1: The Impossible Brief

June 24, 2025. My first day at my new job. Most people get a laptop and a buddy. I got a 0-1 project with a 90-day deadline to generate $1M in revenue.

“We’re building a lending marketplace for SMBs,” my new boss explained. “You start today.”

No onboarding. No ramp time. Just immediate ownership of what would become one of our most ambitious experiments.

The vision was compelling: Small businesses waste weeks on loan applications, getting rejected by banks, starting over. We’d build an intelligent marketplace that compressed weeks into minutes, matching businesses with the right lender instantly.

The constraints were... interesting:

- Small growth pod (including me, still learning everyone’s names)

- Targeting 90 days to revenue

- 1 confirmed lending partner

Day 1. Full ownership. Impossible timeline. Perfect.

Week 1-2: From 0 to Launch

While most teams would spend months researching, we had a different philosophy: ship immediately, learn from reality. In 14 days, we went from whiteboard to production.

The MVP was deliberately minimal, but the infrastructure wasn’t.

We built the whole analytics stack (Segment → Amplitude), comprehensive error tracking that let us watch sessions like reality TV (thanks for the T-shirt LogRocket), a hilariously simple but effective attribution system we called ‘caveman attribution’, just different landing pages for different platforms and even integrated with business registries because nobody should have to type their business address in 2025.

The early data was a masterclass in humility.

Of every 100 people who hit our landing page, 3 started the application. Of those brave souls, 75% rage quit when we asked them to connect their bank account.

The ones who did complete it? They were filling out loan applications squinting at our desktop optimized form on their phones (because of course 78% were on mobile).

That insight changed everything. These weren’t calm business decisions, they were desperate searches for capital when anxiety peaks. We weren’t building B2B software; we were offering SMBs a cashflow lifeline.

Week 3: The Strategic Pivot

With 1% conversion, I made a tough decision: complete redesign mid-flight. Not iteration, reconstruction.

The new architecture I proposed was radically different:

• Progressive application: 1 question per screen with auto-advance

• Mobile-first design: Optimized for thumbs, not desktop pointers

• Smart sequencing: Pre-qualification by the end of the 4th question, scary stuff (bank connection) later in the app, after we’ve built some trust

• Reduced cognitive load: 27 questions → 16, all multiple choice where possible

Results within 72 hours of launch:

• Landing page → application start: 39% → 65% (67% relative improvement)

• Application completion rate: 30% → 50%

• End-to-end conversion: 1% → 4% (4x improvement)

• Cost per signup: $60 → $43 (28% reduction)

The form completion rate of 50% wasn’t just good, it was industry-leading. Most financial applications struggle to break 30%.

Week 4-5: The Innovation Sprint

This is where we built something the company had never tried before, a lead scoring system:

The Lead Scoring Innovation:

• Real-time qualification tiers: S, A, B, C and D (auto-decline) based on self reported revenue, credit score, and time in business

• Meta Conversion API integration: a feedback loop sending lead quality data back to Meta

• Automated routing: S-tier leads got white-glove treatment, D-tier were auto rejected (lending partners wouldn’t approve any D under any circumstances)

• Self-improving acquisition: Each scored lead taught Meta what “good” looked like and the targeting improved

The technical implementation was complex but elegant:

• Tier S: $50k-$100K+ revenue, 720+ credit, 12+ months → Priority queue

• Tier A: $25-$50k revenue, 680-719 credit → Standard queue

• Tier B: $25k-$50k revenue, 600-649 credit → Secondary partners

• Tier C: $10-25k+ revenue, 550-559 credit, 12+ months → Priority queue

• Tier D: <$10k revenue OR <6 months OR <550 credit → Helpful resources

I actual built a simple prototype to help visualize it:

https://zippy-clafoutis-f2aaa0.netlify.app/

This wasn’t just lead scoring, it was building a self-optimizing marketplace. Our CAC started dropping as the system learned, eventually reaching $181 per qualified lead (down from $400+).

Marketing Innovation & Attribution Mastery

Working with limited marketing channels (Meta’s financial services restrictions), I pioneered several approaches:

Attribution Innovation:

• Created platform-specific landing pages for simplified tracking

• Implemented “caveman attribution” that solved the attribution problem

• Built custom conversion events that taught Meta about lead quality, not just volume

Messaging Breakthroughs:

• “Approved before your bank calls you back” - our top performer

• Mobile-first creative strategy (vertical video outperformed static by 3x)

• Speed-focused positioning that resonated with anxious SMB owners

The result: marketing efficiency improved weekly, with cost per qualified lead dropping from $400+ to $181 while maintaining quality.

Week 6: The Hard Truth

By week 6, we’d achieved something remarkable, and remarkably challenging.

Our metrics were genuinely exceptional: applications up more than 50%, a 50% completion rate that made other fintechs jealous, and our customer acquisition cost had dropped by over 60%.

But funded deals, the metric that really mattered?

Basically zero.

Revenue?

A rounding error.

The unit economics exercise took 6 minutes and reset expectations from 6 weeks of intense effort.

Simple math: we’d need to spend nearly $2 on acquisition for every $1 in revenue.

And that was with our already-optimized funnel. I modeled every scenario.

Cut CAC in half again? Still losing money.

Double approval rates? Still underwater.

Triple both? Congrats, you’re at negative margins. There was no combination of improvements that made this work.

Decision Time

Six weeks in, I became the messenger for inconvenient math. The project I’d been hired to run on day one was improving every week but would probably never work.

My role wasn’t to decide its fate, but to ensure leadership had crystal-clear visibility into reality. I presented the unit economics and proposed options for how we could keep pushing with some timeline updates.

Eventually it was decided that we would reallocate resources and frame this not as failure but as strategic reallocation. We weren’t ending a project, we were reallocating resources strategically, to apply the learnings to our core application.

Leadership made the call, but they made it with conviction because the data was undeniable. That’s the real job of a PM: ensuring the right decision gets made.

The Platform We Actually Built

While the project only drove a small amount of direct revenue, we built lasting infrastructure:

Technical Assets:

• Analytics infrastructure: Became company standard for all new products

• Lead scoring engine: Adapted across entire product application experience

• Conversion API integration: Now standard practice

Strategic Insights:

• Discovered exact CAC thresholds for profitability

• Mapped SMB lending landscape

• Built relationships with 4+ lending partners

• Created playbook for rapid marketplace development

Five Principles for 0-1 Products

1. Day 1 Ownership Beats Long Onboarding

Starting with full ownership on day 1 forced rapid learning and decisive action. No time for analysis paralysis.

2. Ship to Learn, Don’t Learn to Ship

Our ugly MVP taught us more in 2 weeks than 6 months of research would have. Real users with real money make real decisions.

3. Build Systems, Not Just Products

Every component was built for reusability. The product died; the infrastructure lives on.

4. Innovation Comes From Constraints

Limited time and resources forced creative solutions like the lead scoring system.

5. Strategic Courage Beats Stubborn Persistence

Killing a project with improving metrics takes courage. But recognizing fundamental flaws early saves resources for winning bets.

So that’s how 6 weeks into a new job, I proved to myself and the company that I could take ownership, drive innovation, and make hard decisions based on data, not ego.

Subscribe now

Most A/B testing is broken and you’ve probably felt it. Teams run test after test, chase promising metrics, and still end up with inconclusive results or worse: decisions based on noise rather than signal. The solution isn’t more testing; it’s better testing. Let me show you exactly how the best product teams run A/B tests to consistently achieve meaningful, measurable results.

Start With the Right Foundation

Your testing platform isn’t just another tool in your stack; it’s the backbone of every experiment you’ll run.

You need something that makes test setup straightforward, provides clear and reliable data, and highlights statistical significance without leaving you guessing.

The wrong platform will create friction at every step. The right one becomes invisible, letting you focus on insights rather than infrastructure.

Define Your Metrics Before You Start

Here’s where most teams stumble: they launch tests without clearly defining what success looks like. By the time results come in, they’re left interpreting ambiguous data and debating what actually matters. Set your standards upfront and save yourself the headache.

Start with confidence levels, aim for at least 95% confidence, which translates to a p-value of 0.05. This means you’re accepting only a 5% chance that your results are due to random variation.

Then establish your Minimum Detectable Effect, typically targeting a 10-20% improvement depending on your market dynamics, traffic volume, and the importance of what you’re testing.

Finally, set your statistical power around 80% to ensure your tests can reliably detect real changes when they exist.

Calculate Your Timeline and Commit to It

One of the most common mistakes in A/B testing is leaving experiments open-ended, checking them periodically, and calling them “done” whenever results look favorable. This is a recipe for false positives and wasted effort.

Instead, calculate your timeline based on historical data. At Keep, we’d pull a 90-day average of the action we were testing (say, button clicks), break it down to a weekly average, and target approximately 1,000 conversions per variant.

This gave us a clear endpoint before we even launched.

If a test didn’t achieve statistical significance by that timeline, we documented it clearly, concluded the test, made a decision and moved forward. Sometimes that meant adjusting our approach and planning a retest with a longer timeline or different design.

Choose One North Star Metric

When you track everything, you understand nothing.

Too many metrics create confusion and make it nearly impossible to make clean decisions. The solution is ruthlessly simple: choose one primary metric as your Overall Evaluation Criterion (OEC).

Airbnb does this brilliantly. Their primary OEC is the number of nights booked.

Not clicks, not searches, not wishlist additions. Just nights booked.

This single metric simplifies every experimental decision across the entire company. You can supplement this with two or three secondary metrics for additional context, but your primary metric should be sacred. Everything else is just commentary.

Establish Guardrails Before Launch

The best teams prepare for both success and failure before they launch a single test.

Start by clearly articulating your hypothesis in a simple if-then statement: “If we change X, we expect Y result because of Z reasoning.” This forces clarity and prevents retroactive justification of random results.

Then set up guardrails that will pause your test if something goes wrong. If your core conversion metric drops by 10-20%, you want the test stopped immediately, not discovered days later when you finally check the dashboard. These guardrails protect your business while you’re experimenting. This is also something that I monitor manually when possible.

As testing programs scale, it gets harder to do that, but it’s advisable to automate this process.

Run Your Tests With Discipline

Once your test is live, monitor it carefully but resist the urge to make decisions based on early results. You should be checking your metrics regularly to ensure nothing is going catastrophically wrong, but checking to see if you’re winning is different from checking to make sure you’re not breaking things.

This is the trap of what we call “p-value peeking.” When you see your variant pulling ahead early, it’s incredibly tempting to call it a winner and ship it. But premature analysis based on incomplete data dramatically increases your chance of false positives. Early patterns often don’t hold as more data comes in.

So watch your test, absolutely. Set guardrails that will alert you if key metrics drop significantly. If your core conversion metric declines by 10-20%, pause the test immediately to investigate. You can always restart a test, but you can’t undo real damage to your business.

But don’t make positive decisions early. Set your timeline and trust it. Let the test run its full course before you declare a winner.

While you’re waiting, complement your quantitative data with qualitative feedback. Run user interviews, watch session recordings, gather context that numbers alone can’t provide. When your test concludes and you have results, this qualitative insight helps you understand not just what happened, but why it happened and what the actual user experience was like.

And if your results seem too good to be true, they probably are. Use Twyman’s Law as your gut check: any figure that looks interesting or different is usually wrong. When something seems unrealistic, run a replication test before you roll it out company-wide.

Document Everything

Yes, documentation feels tedious. Yes, it’s time consuming. And yes, it’s absolutely crucial. Record exactly what you tested, which metrics you tracked, what the outcomes were, and what decisions you made based on those outcomes.

This documentation becomes your team’s institutional memory. At Lendio, we maintained a central repository of experiments that anyone in the company could access. New team members could quickly learn from past tests. Product managers could avoid redundant experiments. Executives could understand the empirical foundation behind major product decisions. Without documentation, you’re doomed to repeat the same experiments over and over, learning nothing.

Communicate Constantly

The best testing programs aren’t run in isolation, they’re transparent and collaborative. Establish a clear communication cadence: pre-launch updates, launch announcements, kill updates when you pause tests, and results summaries when experiments conclude.

At Keep, Lendio and Forge we used dedicated Slack channels for experiments, ensuring stakeholders always knew what was running, why we were running it, and what we were learning. This transparency builds trust, prevents surprises, and helps everyone understand that good testing includes both wins and losses.

Structured A/B testing isn’t just another process to follow, it’s your strongest tool for confident decision-making and measurable growth. When you move from ad-hoc experimentation to a disciplined testing framework, you transform uncertainty into actionable insights.

Stop spinning your wheels. Start making real, measurable progress today.

Subscribe now

Ryan is currently 35 days into a challenge to visit all 50 states in 50 days, posting daily videos and raising millions for St. Jude Children's Research Hospital. To date, he's generated over 100 million views and raised more than $6.1 million in donations. Here's why that matters for brands and creators:

Brands Spending Smarter

Brands traditionally allocate large budgets to marketing campaigns with uncertain returns. Ryan’s series offers an engaging and cost-effective alternative. For instance, if Lectric Bikes spends $600,000 sponsoring Ryan’s series and achieves 200 million impressions, their cost per impression would be approximately $0.003 (a CPM of $3). Compared to the industry standard CPM of around $7, this represents exceptional value and highlights how engaging content partnerships can efficiently reach large audiences.

Joyride and St. Jude: Creating Meaningful Impact

Ryan smartly integrated his candy brand, Joyride, with the charitable cause of St. Jude Children's Research Hospital. This thoughtful integration allows Joyride to build lasting emotional connections with viewers. If just 5% of viewers (around 200,000 people) try Joyride, that is significant exposure. With an average spend of $4 per customer and an estimated four repeat purchases, just 10,000 loyal customers would translate into roughly $160,000 in customer lifetime value (LTV). Not bad for quick napkin math; the actual outcomes could be even stronger.

Spotlighting Local Airbnbs

Ryan’s journey across America naturally highlights local Airbnbs, showcasing unique stays in ways traditional advertising rarely matches. Each Airbnb featured receives organic exposure, and the winning location earns additional publicity, creating tangible value for hosts.

Ryan’s Financial Reality (Quick Math)

Beyond donations and brand partnerships, Ryan’s series also generates significant YouTube ad revenue. With a conservative CPM estimate of $7, each view generates around $0.007. At 200 million total views, Ryan’s series is projected to generate approximately $1.4 million from ads alone.

Sheesh.

Who’s Winning from This Series?

Ryan’s YouTube Channel: Enjoying substantial growth, increased views, and ad revenue.

Joyride (Ryan’s Candy Brand): Achieving meaningful brand exposure and measurable customer lifetime value.

Donors and Viewers: Engaging with a compelling cause and feeling part of something impactful.

Brands and Sponsors: Benefiting from authentic goodwill and positive visibility at a highly competitive cost per impression. Talk about bang for your buck.

St. Jude Children’s Research Hospital: Receiving essential financial support and increased public awareness.

Local Airbnbs: Gaining valuable organic visibility and enhanced recognition.

Ryan’s success stems from creating an interactive, authentic experience that genuinely engages communities. It goes beyond mere entertainment, showcasing marketing's true potential to drive meaningful impact.

I’m a sucker for productivity advice.

Show me a YouTube guy in a fitted tee talking about cold plunges and 5am sprints, and I’m in. I’ll even take notes.

But there’s a problem.

Those guys? They don’t live the same life I do.

They’re single. No kids. Living in a high-rise with no furniture. Their biggest morning obstacle is whether to drink mushroom coffee or butter coffee.

Meanwhile, I’m up at 2:33am cleaning up someone else’s barf (which, if you think about it, technically makes me more productive than the 5am crew) and making sure it doesn’t fall onto another kid's face in the bottom bunk.

I’m in the ultra-elite, toddler-induced pre-dawn grind set.

Honestly, I should probably sell a course about it or start a podcast called 'Barf & Brilliance’ or ‘Diaper Deadlifts’.

Whatever, you get the point.

By 7:30am, I’m not journaling, I’m negotiating a ceasefire over frozen waffles or making sure the insurgents don’t burn the house to the ground with toast cooked at the toaster’s 9 setting. Then I get to fight over why you can’t wear your doggy shirt for the 4th day in a row.

Bonus points if I manage to make my own breakfast without stepping on a Lego or accidentally backing into a kid. Oh, and also you or your significant other woke up sick, again, for the 3rd time this month. Thanks Kindergarten.

By 9am, the Hustle Bro has crushed his deep work session, posted twice on LinkedIn, answered emails, and is halfway through a kettlebell workout. I’ve survived breakfast combat, refereed an argument about keeping oatmeal in your bowls, and maybe put on pants.

They say to “grind in the evenings” too. Which is hilarious.

Evenings in my house are less “grind” and more “boss level.” You ever try to get three overtired kids to brush their teeth at the same time?

I’d rather launch a startup.

At least in a startup, no one cries because their toothpaste is the wrong color or doesn’t have a picture of a Pokémon on it.

Working out? I want to work out. I really do.

Some mornings I sneak in a workout with my three tiny training partners climbing on me like I’m a jungle gym. I do squats, they do pull-ups on my thighs. I do pushups, they climb dad hurdles.

This is the part hustle culture doesn’t account for.

The variable of real life.

The part where optimizing your calendar doesn’t help when someone throws up on it.

Where the 5am club is involuntary, and includes a toddler demanding “tucky in.”

Look, I admire the hustle. I genuinely do. I love the idea of pushing yourself.

This isn’t some passive-aggressive excuse for not trying hard, it’s just a different kind of hard. It’s not a lack of ambition, it’s a redistribution of energy. And sometimes that energy goes to wiping a nose or managing a meltdown instead of launching a SaaS product.

But let’s be honest: the hustle bro's life is an edge case. Optimized to the point of unreality. It’s like trying to benchmark your daily routine against a robot that doesn’t eat, sleep, or ever forget where they put their keys.

It’s what happens when you remove friction: no kids, no community obligations, no commute, no birthday parties at trampoline parks.

And that's fine. But it’s not replicable for the majority of us. Not if you’re trying to raise actual humans. Or show up for your neighborhood. Or just be the kind of person who remembers to take the trash out before the truck comes.

I’ve seen the posts. “Work 12 hours a day for 10 years so you never have to work again.”

Cool.

But also: who’s making dinner?

Who’s coaching little league?

Who’s helping their neighbor move a couch?

Because here’s the part you don’t hear on those podcasts:

Sometimes the highest-leverage thing you can do is sit on the floor and play Legos with a four-year-old.

Not because it scales.

But because it doesn’t.

Harvard Business Review never ran that case study.

And I get it, those things don’t make great LinkedIn content.

“Skipped deep work today to clean crayon off the walls. Again.”

But that’s the work, too. It’s just quieter. Stickier. Occasionally smells like feet.

So yeah, I’m envious of the hustle bros sometimes.

Their lives look clean.

Efficient.

Bathed in morning light.

But I also wonder:

What’s all that optimization for?

What’s the endgame?

Are you building something people can be part of—or just something people can look at?

Maybe the real flex isn’t getting up at 5am.

Maybe it’s showing up, again and again, for the stuff that doesn’t scale.

And let’s be clear, that doesn’t mean it’s easier.

It means it’s messier. Less efficient. More emotionally exhausting.

It’s not opting out of hard work, it’s choosing the version with more crying, more cleaning, and way fewer Instagrammable moments.

And if you’re doing that, if you’re juggling a job, a family, a community, and still finding moments to chase your ambition, congrats.

You’re not behind.

You’re playing a different game entirely.

One where the prize isn’t productivity.

It’s people and a life that’s lived in, not looked at.

And yeah, it’s exhausting.

It’s demanding in ways that don’t show up on charts or dashboards.

But that doesn’t make it any less of a grind, just a different one.

Share

Welcome to a new series I’m calling “The Lab Report”. The Lab Report is a deep dive into an interesting topic, so if you have some ideas shoot them my way!

This weeks post is focused on comparing CRO to Growth Product Management roles, something I’ve been thinking a lot lately. Enjoy and please leave any feedback!

Without further adieu…

Growth PMs vs. CROs - What Is The Actual Difference?

What’s the actual difference between a CRO and a Growth PM and why do so many companies conflate the two when the work, the scope, and the outcomes couldn’t be more different?

They both run experiments. They both chase conversion metrics. They both speak fluent A/B.

But beneath the surface, these roles operate on very different levels. One is focused on squeezing value out of existing traffic. The other is designing systems that fundamentally change how the product grows. And yes, that difference shows up in the org chart, the scope of influence, and ultimately, the paycheck.

Studying compensation for CROs and Growth PMs led me down a rabbit hole that turned into this post. If you saw my earlier breakdown on Conversion Rate Optimization vs. Growth, that was the overview. This one zooms in on the roles and the org charts.

Day-to-Day Work: Zoomed-In vs. Zoomed-Out Optimization

Let’s just say this: if your job feels like interior decorating for landing pages, you’re probably doing CRO. If it feels like you’re quietly rearranging how the entire company thinks about user retention, that’s more Growth PM energy.

CROs typically operate in the marketing department, where they are responsible for squeezing performance improvements out of existing touch points. This can include optimizing landing pages, tweaking form designs, running headline tests, and analyzing bounce rates. The work is tactical and often campaign-driven, focused on how to turn a specific ad click or site visit into a measurable action like a sign up or purchase. Success is usually tracked on a short-term basis: Did this test improve conversion by 3%? If yes, implement. If not, move on.

In contrast, Growth PMs work within the product team, where their mandate spans the full user lifecycle from acquisition and activation to retention and monetization. They own growth at a systems level. That means they’re responsible not just for whether someone signs up, but whether that person sticks around, finds value, and contributes to long-term business outcomes. Their work might include rethinking onboarding, launching in-product referrals, designing monetization features, or identifying drop-offs in the activation flow. The time horizon is longer, and the stakes are higher.

Experimentation Approaches: Surface Tweaks vs. Infrastructure Changes

CROs live in the world of fast wins and faster tests. It’s the digital equivalent of, “Let’s see if people click more when it’s blue.” Growth PMs, on the other hand, are the ones asking, “What if the whole onboarding flow is the problem?” One runs sprint drills. The other is playing a long game with a whiteboard and way too many sticky notes.

Both CROs and Growth PMs rely heavily on experimentation. But the experiments they run, and the implications of those experiments, differ significantly.

CROs tend to focus on rapid testing cycles. These are experiments with low implementation cost and minimal downstream risk like comparing different CTA styles, button placements, or hero images. The tools used (Google Optimize, Optimizely, etc.) are generally plug-and-play and designed for fast, visual iteration. CROs look for statistical significance on relatively high-traffic pages and aim for steady, incremental gains.

Growth PMs, meanwhile, often lead experiments that require code changes and coordination across engineering, design, and analytics. These aren’t just visual tweaks they’re infrastructure-level bets. A Growth PM might redesign a mobile onboarding sequence, test different pricing models behind a feature flag, or roll out a new retention mechanism. The feedback loops are longer, and the interpretation of results often requires a more nuanced understanding of behavioral data over time.

Metrics and Measurement: Snapshot KPIs vs. Lifecycle Health

You can tell what someone cares about by what they measure. CROs are watching the front door like hawks. Growth PMs are the ones pacing the hallway, wondering why guests came, stayed for five minutes, and left without saying goodbye.

CROs track metrics like:

• Page-level conversion rate

• Bounce rate and exit rate

• CTRs and cost-per-conversion

• Funnel drop-off at specific marketing stages

Their success is often defined by how effectively they can convert traffic into leads or customers, using a known set of acquisition levers.

Growth PMs, on the other hand, are held accountable for:

• Activation rates (e.g., % of users reaching first value)

• Retention cohorts and churn curves

• Revenue per user or LTV

• Overall growth model performance (e.g., virality, monetization lift)

Their job is to make sure users don’t just show up but stick around and generate value over time. They’re optimizing for compound growth, not campaign ROI.

Org Design and Influence

CROs often report to marketing leadership and collaborate most closely with performance marketers, content writers, and designers. Their work tends to align with campaign timelines and short-term goals. When site changes require engineering support, they often have to lobby for prioritization or workaround using third-party tools.

Growth PMs typically sit in a cross-functional product pod and have dedicated engineering and design support. They report into product leadership (sometimes directly into the CPO) and are expected to contribute to strategic planning. Because they touch revenue, retention, and scalability, they often present directly to the executive team and are expected to show impact at a business level.

In mature orgs, Growth PMs help define how the product grows. CROs help ensure that the acquisition efforts are converting as efficiently as possible.

Compensation Disparities: Scope, Strategy, Scarcity

The reason Growth PMs generally get paid more comes down to scope and leverage. Growth PMs are expected to:

• Think across time horizons

• Make decisions that affect product and business strategy

• Orchestrate large, cross-functional initiatives

CROs, while critical to performance marketing, tend to operate within narrower boundaries. Their experiments are more reversible, and the impact is typically constrained to top-of-funnel outcomes.

U.S. salary ranges:

• CRO Specialist or Manager: $70K–$130K

• Growth Product Manager: $130K–$200K+

• Head of Growth: $180K–$250K+ (plus equity, depending on company stage)

Context Matters: Startup vs. Enterprise

At a startup, job titles are more like rough guesses based on whoever last edited the job posting. One day you’re optimizing landing pages, the next you’re trying to convince the CEO that a 10-minute onboarding is maybe not the move. In an enterprise, things get a little more orderly or at least the chaos is hidden behind a few layers of process.

In startups, roles tend to be fluid. A Growth PM might be running Facebook ads one day and scoping onboarding UX the next. A CRO expert might find themselves owning the entire growth stack until the team matures. Titles are often misleading; execution trumps specialization.

In enterprises, the roles are more clearly defined:

• CROs focus on conversion optimization across digital campaigns and site assets

• Growth PMs work within product to increase engagement, retention, and LTV

Some companies spin up centralized growth teams with full-stack support. Others embed Growth PMs across squads. But the general trend is toward product-led growth owning retention and revenue, while marketing continues to drive top-of-funnel acquisition.

Summary: Different Roles, Shared Goals

If you're still not sure which role is which, just ask: is this person rearranging traffic or rethinking systems? One tunes the engine. The other rebuilds it in traffic with a Slack thread blowing up.

CROs and Growth PMs both operate within an experimentation culture but the nature of what they optimize and how they drive growth is fundamentally different. CROs maximize the performance of the systems that already exist. Growth PMs design and iterate on the systems themselves.

If your site traffic is healthy but conversions are weak, hire a CRO. If users sign up but don’t stick around, hire a Growth PM. And if you’re early stage and trying to decide which role to prioritize? Start with someone who can think like a PM but execute like a marketer until the complexity demands you split the role.

Sources and References

1. Glassdoor – Conversion Rate Optimization Specialist salary: https://www.glassdoor.com/Salaries/conversion-rate-optimization-specialist-salary-SRCH_KO0%2C39.htm

2. ZipRecruiter – CRO Specialist salary: https://www.ziprecruiter.com/Salaries/Conversion-Rate-Optimization-Specialist-Salary

3. Built In – CRO Manager (Prodege LLC): https://builtin.com/job/conversion-rate-optimization-manager/2428372

4. Chime – CRO Manager, Web: https://careers.chime.com/en/jobs/7958856002/conversion-rate-optimization-manager-web/

5. Greenhouse – Growth PM (Robinhood): https://job-boards.greenhouse.io/robinhood/jobs/6566145

6. Medium – Growth PM vs Marketing vs Manager (Omamuzo Samson): https://oma1da.medium.com/growth-product-manager-vs-growth-marketing-manager-vs-growth-manager-whats-the-difference-b78fd6a60676

7. Productboard Blog – Core vs Growth PM: https://www.productboard.com/blog/what-core-and-growth-product-managers-can-teach-each-other/

8. Lenny’s Newsletter – Hiring for Growth (Elena Verna):

Lenny's Newsletter

Six rules of hiring for growth

👋 Hey, Lenny here! Welcome to the monthly ✨ free edition ✨ of my weekly newsletter. If you’re not a subscriber, here’s what you missed this month…

Read more

4 years ago · 75 likes · 2 comments · Elena Verna

1. Reddit – WTF is a Growth PM: https://www.reddit.com/r/ProductManagement/comments/su0izr/wtf_is_a_growth_pm/

2. Medium – Growth vs Marketing (Daphne Tideman): https://medium.com/better-marketing/growth-isnt-marketing-f0bf614ec0ce

3. Built In – Senior Growth PM (ActiveCampaign): https://builtin.com/job/senior-growth-product-manager/3872503

4. Built In – Product Growth Manager (LiveFlow): https://builtin.com/job/product-growth-manager/4774653

5. Built In – Growth PM, Experimentation (Quince): https://builtin.com/job/conversion-rate-optimization-cro-manager/2299070

6. BlueSteps – CRO Manager (search): https://www.bluesteps.com/Opportunity/SearchResults?keywords=Conversion%20Rate%20Optimization%20Manager

7. VWO – How to Master eCommerce CRO: https://vwo.com/blog/how-to-master-ecommerce-cro-interview-with-tomasz-mazur/

8. PayScale – CRO Skill Salary: https://www.payscale.com/research/US/Skill=Conversion_Rate_Optimization_(CRO)/Salary

9. Omniconvert – Guide to Hiring a CRO Specialist: https://www.omniconvert.com/guide-to-hiring-a-cro-specialist/

10. SmartWorkersHome – Sr. Growth PM (Cro Metrics): https://smartworkershome.com/jobs/senior-growth-product-manager-cro-metrics

I wanted to figure out which of the main vibe coding tools, v0, Bolt, Lovable, and Replit, work best in different scenarios. To find out, I ran a test.

I gave each tool the same prompt: Build a web app that allows me to input a URL or upload a screenshot and then make variations to it for an A/B test.

After that initial prompt, I tailored follow up instructions based on each tool’s output to see how far you can get with targeted prompting and a little familiarity with how each tool behaves.

The Task

The prototype I wanted was simple in concept:

• Paste a URL or upload a screenshot

• Visually select elements (like buttons or headlines)

• Change styles or text

• Save those variations for testing

The goal was to streamline internal experimentation with a lightweight A/B testing editor.

How Each Tool Performed

v0

• It finished first, producing a barebones design quickly. But the app didn't work.

• It removed the URL feature entirely and focused on image upload, which also didn't work.

• When prompted about the broken functionality, it responded with a six-month roadmap, which didn't help fix the immediate issues.

• Eventually, it got stuck trying to write code and couldn’t recover without additional direction.

Takeaway: It's good for quick UI sketches or mockups, but it tends to stall when the task requires thinking beyond the screen.

Bolt

• The first version had a well-structured, multi-screen interface, and the UI looked polished.

• Neither the URL nor image uploader worked, and both attempts led to a broken screen.

• In later prompts using the “Discuss” feature instead of “Build,” it came closer to understanding what was needed.

• It appeared confident, but ultimately built the wrong things and couldn't get past the dummy image. It did better after restarting with more detailed instructions.

Takeaway: It's solid for prototyping user flows, but it needs more direction to deliver working features.

Lovable

• It broke early in the build, but eventually fixed itself and continued working.

• It didn’t build a working version, but got close after two prompts. The app remained mostly a shell with no real functionality.

• Despite that, it correctly inferred the need for a CORS solution and even asked to be prompted for refactoring, which showed a surprising level of self awareness.

• After two hours of testing, the visual editor still wasn’t working, but Lovable had tackled more technical nuance than the others.

Takeaway: It's more intuitive than the others and shows potential when paired with clearer prompts and more iteration.

Replit

• Neither the URL input nor image upload worked initially.

• When prompted to explain what was wrong, it offered a clear explanation of what needed to change and how to move forward.

• It got to a working version fairly quickly after that, successfully pulling in the page with an iframe.

• It still didn’t create variations, but the iframe integration and editor loading were functional and stable.

Takeaway: It's slower but more thorough, with stronger technical grounding. It's especially useful if you can write or interpret specs.

Observations

Prompting Strategy Matters

The tools worked better when:

• Prompts included clear technical details

• Instructions were step by step

Using ChatGPT as a Senior Dev

One approach that worked especially well was treating the AI’s output like it came from a junior developer, and then using ChatGPT as the senior engineer in the room.

After getting a partial or broken response from an editor, I’d summarize what it was trying to do, then give those notes to ChatGPT with this prompt:

"Assume these are notes from a junior dev. Act like a senior dev and write instructions back to them."

ChatGPT would respond with more structured, technically grounded directions. This process helped clarify the architecture, identify missing pieces, and improve the quality of my next prompt to the editor. It also helped me catch when I was being too vague or assuming the tool would fill in the gaps. In short, it created a loop where I could use one AI to upgrade the output of another.

ChatGPT was useful for translating fuzzy goals into specific tech specs, which made a big difference, until the details overwhelmed the editors.

Quick Comparison

Final Notes

None of these tools just work. But they're each useful in the right context. What matters is how well you can guide them, and how well you understand the problem you're trying to solve.

If your task is well structured and you're clear with your inputs, these editors can accelerate your workflow and save time on early stage builds. They're not a replacement for engineering yet, but they're a great extension of your thinking process.

If you're experimenting with vibe coding tools, or want to build internal prototypes without spinning up a full dev team, this kind of testing can help you figure out where each tool fits.

This isn’t one of those posts where I tell you to quit your job and go become a PM. But I do want to talk about what most people don’t realize: there’s a serious pay gap between two roles that look really similar on the surface.

I pulled some salary data. It isn't perfect, but I do think it tracks with what I’ve seen in the wild and what a lot of other folks have seen too.

Here's a side-by-side comparison of median base salaries across levels:

But before we get too far, let’s answer the obvious question:

Is it fair to compare these roles?

Kind of.

CRO and Growth Product Manager roles do share real overlap: Both are obsessed with user behavior and performance. Both rely on experimentation, iteration, and testing. Both require analytical chops and an eye for user experience.

But they also diverge in meaningful ways: Product roles tend to have broader ownership, roadmaps, cross functional alignment, long term feature development. CRO roles can be more focused, often tied closely to marketing campaigns or specific site flows.

So no, they’re not the same job. But they are close enough that the salary difference is worth noticing, especially when the work and mindset can look so similar on the ground.

Quick note: we’re focusing on CRO and Growth PMs here, but there are other experiment heavy paths worth mentioning, like data science. Plenty of data scientists live in the world of A/B tests and behavioral insights too. That’s a separate (and fascinating) rabbit hole, and we’ll save that comparison for another time.

Now, here’s what jumped out:

A Real-World Example: Same Vibes, Very Different Salaries

It’s one thing to look at charts. It’s another to see the gap in actual job postings.

Take this CRO Manager role at Chime:

• Chime CRO Job Posting

• Salary listed: $103,680 to $144,000

Now compare that to this Growth Product Manager role at Turno:

• Turno GPM Job Posting

• Salary listed: $135,000 to $165,000

Side note: this Turno role was even highlighted by Elena Verna as a strong example of a well-defined growth position, so this isn’t some weird outlier.

Same experimentation mindset. Very different compensation. Worth keeping in mind if you're navigating your next move.

CRO and Growth PM roles both live in the test and learn world. Behavior driven. Conversion obsessed. Iterative. But one of those roles tends to pay a lot more. Like, tens of thousands more.

You don’t have to be in love with money to find that interesting. It’s not just about the paycheck. It’s about responsibility, cognitive overhead, and what you’re signing up for.

And just to drive it home, here’s the percentage difference in pay by level:

Because here’s the thing:

Product roles usually ask more of you. Higher stakes. Broader business knowledge. Tighter engineering collaboration. There’s no tidy entry path into product from most universities, so getting in is more about who you know and what you’ve done.

Marketing is a little more glamorized (thanks, Don Draper). Product is more specialized. And structurally, product roles go higher. Ever heard of a VP of CRO? Yea me either.

If you want to see how it plays out over time, here’s how salary growth compares by experience:

And if you're wondering where this all comes from, here's the raw data side-by-side:

None of this means you should bail on marketing. If you love it, stay. Seriously. Some people just have the instinct for it. And that’s not nothing.

But if you’ve been deep in CRO, if you love testing and building better experiences, it might be worth asking: what hill are you climbing?

It's also worth noting that titles can vary wildly from company to company, especially on marketing teams. A 'Manager' at one org might be a 'Senior' at another. These aren’t perfect, standardized ladders, they’re messy approximations. So take all this as directional, not gospel.

That said, if you are looking to move toward product, here’s the good news: You’re not starting from zero. Experience running experiments, understanding user behavior, and optimizing flows is highly transferable. With a little coaching or mentorship, you can start to pick up the product-specific parts like managing a roadmap or working closely with engineering. Some roles (like Growth PMs or Product Analysts) can be great stepping stones.

So no, you don’t need to pivot your whole career overnight. But you might be closer to product than you think.

And if you’re thinking about making the leap… well, you’re not alone.

(Just don’t expect a VP of CRO role to be waiting for you.)

Sources and Credits

A big thanks to the folks and platforms that provided this data that made this comparison possible:

• Experiment Nation for collecting and publishing CRO salary data.

• Lenny Rachitsky for the breakdown of Product Manager compensation.

• Payscale, Salary.com, and Built In for additional benchmarks and ranges.

• Elena Verna for her insight on quality growth job descriptions and for spotlighting the Turno posting.

Want to watch/listen on YouTube?

Diagnosing the Funnel Breakdown

I’ve been thinking a lot about intent lately specifically, what happens when the signals I think I'm getting turn out to be noise.

One example that sticks with me is from my time at Route, where a surprisingly high volume of consumers were submitting the merchant demo request form.

The form was built for prospective merchants interested in my B2B offering.

But week after week, the sales team was buried in low-quality leads.

Meanwhile, actual consumers people trying to track packages or resolve issues were hitting a wall. No one was winning. Not the sales team. Not the user.

I needed a fix. But more than that, I needed to understand what was really happening at that critical moment of intent.

The Experiment

So, I put together a scrappy landing page. Instead of letting users go straight to the demo form, I added a simple decision point: “Are you a merchant or a consumer?”

• If they said merchant, they got the original form.

• If they said consumer, I showed them four clear options: track a package, download the app, file a claim, or read the FAQ.

The build was quick and intentionally rough. The goal wasn’t polish, it was clarity.

What I Expected

I figured a majority of these form submissions weren’t real leads.

They were people confused about where to go or what the form was for.

If I could route them correctly, I could reduce noise in the pipeline and improve the experience for consumers.

What Happened

The experiment ran for just a few days. Here’s what I saw:

• 11% drop in consumer submissions

• 18% increase in merchant submissions

Before: 143 total submissions, 89 from consumers (62%)
After: 101 total submissions, 56 from consumers (55%)

Cleaner data. Better intent match. And for the first time in months, the sales team could focus on the right leads. Admittedly not the biggest sample size, but I had to move fast and leading indicators was enough to take action.

Consumer Click Behavior:

• 46% “Track a package”

• 32% “Download the app”

• 10% “File a claim”

• 3% “FAQ”

• 7% “Other” (spam)

It was clear with these results, these users weren't trying to talk to sales. They just didn’t know where else to go.

A Second Lens: Brand Ad Traffic

I ran the same flow on brand campaign traffic. That gave me a fresh look at ad intent:

• 82% clicked “Track a package”

• 7% clicked “File a claim”

• 10% clicked “Merchant demo”

This told me two things:

1. I was attracting the wrong audience.

2. Even when people clicked a demo ad, many of them were just confused consumers.

Where I Took It From There

Armed with these insights, I made three key moves:

• Working with our Paid team, we adjusted targeting and messaging to filter better-qualified traffic.

• I partnered with the Consumer Marketing team to start building the resources consumers were actually looking for.

• I rolled out a permanent experience that asked users if they were merchants or consumers, effectively implementing the winning version of the test.

Why It Mattered

This test didn’t just clean up a form. It revealed a deeper truth about intent misalignment. Sales was wasting time. Consumers were getting ignored. And both issues stemmed from the same problema lack of clarity at the top of the funnel.

What worked wasn’t fancy. Just a quick page, a clear question, and a willingness to follow the data.

If you’re seeing conversion but not quality, check intent. That’s where the story usually starts.

Dan

Maybe you’ve heard people talk about tools like Bolt, Cursor, or v0. Or you’ve seen someone spin up an app in five minutes flat and thought, "Wait, how did they do that?"

That’s vibe coding. The term, as far as I know, was coined by Andrej Karpathy. It kind of perfectly captures what it feels like to build this way.

It's this strange, kind of magical experience where you open up a tool like Bolt, v0, Cursor, Replit, Lovable, or Windsurf and instead of writing code, you just start typing. Natural language. Prompts. Descriptions. Whatever comes to mind.

Then suddenly you’ve got an app. It’s not always perfect, but it works. And it’s fast. Way faster than opening up VS Code, setting up a repo, building a UI from scratch, debugging, and all the other stuff that used to feel like step zero.

I Built Some Stuff Too

I’ve been deep in it lately. Just over the past few weeks, I’ve built:

• a Pokémon of the Day app

• an A/B Test Analyzer

• a Prompt Library

• a Test Plan Generator

• an Emoji GIF Maker

• a memory game for my kids

• a weird side scroller called "Toad and Kirby" that my 6 year old made up (Nintendo, please don’t sue)

• A piano website for my wife

• and a RICE calculator

Some of these took more than a day. Some were done in under an hour. The point isn’t that they’re polished, they’re not. The point is that they exist. And more importantly, they solve a problem or make someone laugh or let a 6 year old build something just by telling his dad what he wants.

This isn’t just no code. It’s barely code. And I think it’s how a lot of MVPs are going to get built going forward.

You don’t need to be an engineer to build something anymore. You just need a decent idea, a little patience, and the willingness to let the vibes take the wheel.

How I'm using vibe coding as a Growth PM

As a Growth Product Manager, this approach has been a game changer. Going from idea to prototype in an afternoon isn't just fun, it's productive. I've been using vibe coding to spin up rough MVPs, test tools and interactive ideas before involving engineering. It's an incredibly empowering way to explore what's possible without needing to scope tickets or wait for sprint cycles.

Being able to go from natural language to a working prototype (or even a simple live app) helps me get feedback earlier, validate faster, and stay in the creative zone longer.

The future of AI in product work

As these tools get better, they're going to reshape how product teams operate. I don’t think it means fewer jobs but I do think it means leaner teams, faster cycles, and a lot more output per head. That’s exciting.

This kind of shift can feel scary big tech changes always do. But history shows we adapt. Jobs evolve, roles change, new industries appear. The key is to stay adaptable. Stay curious. Keep experimenting.

Right now I’m spending a lot of my time reading, studying, and actually using these tools. I want to be on the bleeding edge of this stuff. I want to know what they can do, and be the one pushing their boundaries.

Want to talk more about vibe coding or learn how to do it? Hit me up.

Dan

Why Test Plans Matter (Even If They Can Be a Pain)

Writing test plans is important, but also kinda sucks.

Good Growth PMs and Experimenters need them to keep everyone aligned, ensure clear goals and help nail down the necessary stats like sample size, test duration, minimum detectable effect (MDE), and conversion rates (CVR).

So I spent the past couple of months building a tool to make writing test plans less tedious, vibe coding with Bolt.new and the OpenAI API.

It took a few iterations and a new feature roll out from Bolt but, I got it working (mostly) eventually.

Try it here:

https://planner.experimentationlabs.com/

Introducing the AI Powered Test Plan Generator

The tool guides you through setting up your experiment: context, hypothesis, expected impact, experiment setup, success metrics, and variants.

Each section has an "Enhance with AI" button that taps GPT 4o to improve your draft, making your plan clearer and more professional, like having an experimentation expert edit your inputs and flesh them out.

Highlights and Favorite Features:

• Test duration calculator: Enter your traffic and statistical targets to instantly see how long your test should run. This attempts to cut out all the guesswork. Mostly all you need is data about your channel/source, current CVR and ideal testing stats (confidence, MDE, power, etc.)

• AI Enhance feature that makes fleshes out your inputs like context, hypothesis. Makes it easier to sound like you know what you're talking about. Also occasionally brings in something I hadn't considered, some good reason to do the thing.

Stuff I Learned Along The Way

• I'm still pretty new to databases . Connecting everything together proved tricky, but working through the problems gave me a deeper understanding of databases in general. I'm still not an expert, but I'm understanding it a lot more, thanks Supabase.

• I ran into some initial problems integrating the OpenAI API. A timely Bolt update simplified the deployment process significantly. It's probably not bulletproof from a security standpoint yet, but good enough for personal and internal use.

• Throughout this project, I realized the value of clearly phrasing questions to GPT models. Asking for explanations in simple terms was key to solving the unexpected issues that popped up.

• The tool is built with React, TypeScript, and Tailwind (the Bolt stack) and allows you to export your completed test plans directly to PDF or PowerPoint, making sharing and presenting straightforward. There are some things I'd like to update in the future with the export feature, but I'm honestly not sure what's even possible there, TBD.

Results in Real Life

I've used this for several recent test plans, here are a few examples of the AI enhanced output:

Context

We want to test which type of imagery works best for paid landing pages targeting the buyer audience. We plan to test 3 different versions, but in an unorthodox way. We will roll out a new version as a control and then launch 2 different versions, so control vs variant a vs variant b.

Notes

This was actually a pretty good output, I basically just brain dumped an unorthodox test idea into the input and it was able to take it and make sense of it, output wasn't bad.

Context

We're running an A/B test on the homepage hero section, comparing the current header with the wave design against the same copy paired with product imagery inspired by the new design direction. The goal is to assess which version drives better engagement and conversions.

Notes

The input here was pretty loose, in fact I just copy and pasted a request from a stakeholder, then it took it and turned it into something usable. I was pretty happy with that.

Next Steps (And a Question for You)

I plan to write more about this process soon, but I'd really like your input:

• What frustrates you most about test planning?

• Are you currently using AI tools in your workflow?

• What insights would you find valuable about building similar AI driven tools?

Shoot me a message, I'd love to hear from you.

Dan

That’s not just preference. It’s a pattern. And the sooner you figure out how you actually like to work, the better your choices get.

I wanted to see what my own patterns looked like, so I used AI to help me find them.

Here’s what I did and what it showed me.

Subscribe now

Step 1: Ask better questions

There’s a book called Job Moves by Ethan Bernstein, Michael B. Horn, and Bob Moesta. It’s full of questions designed to help you get clear on how you work. Not what looks good on a resume. What actually feels good to do.

I picked a few and answered them honestly. Then I asked ChatGPT to answer the same questions about me, based on what it’s picked up from our past conversations.

Not "pretend to be me." Just this:

"You’ve seen how I think. How would you answer these questions about me?"

Here’s what I asked:

• What kinds of tasks give me energy

• What kinds of tasks drain me

• When have I felt the most fulfilled doing something

• What exactly about micromanagement bothers me

• Why do I prefer autonomy

• What problems do I like solving, and how do I approach them

• How do I respond to unclear expectations

• When am I most productive, and why

• When do I feel most engaged

Step 2: Let AI reflect it back

I dropped in my answers. Then I added ChatGPT’s version. Finally, I asked:

"What patterns do you notice?"

It came back with something surprisingly structured. It didn’t just spit out vague traits or personality types. It mapped out how I work best, what shuts me down, and what actually matters.

Here’s what it gave me.

Energy Drivers (What I Need Most)

1. I thrive when I have autonomy to choose how I work. This looks like being given outcomes and trust to find the best path. This should not look like someone assigning detailed steps or watching over my shoulder.

2. I need to work on problems that are interesting and meaningful to me. This looks like projects that challenge my thinking and have real impact. This should not look like repetitive tasks or vague assignments with no clear value.

3. I do my best work when I have deep, focused time to build, research, or think. This looks like large blocks of uninterrupted time and protection from Slack chaos. This should not look like constant meetings, pings, or shifting priorities.

4. I feel most engaged when there is clarity about what matters and why. This looks like strong leadership, clear roadmaps, and well communicated goals. This should not look like vague direction, unclear priorities, or daily pivots.

5. I’m energized by building things that help people or generate real value. This looks like working on tools, experiments, or insights that solve real problems. This should not look like busywork or optics focused tasks that don’t matter.

6. I enjoy working with a small team of smart, curious people. This looks like 2–4 thoughtful collaborators who challenge and support each other. This should not look like large, political teams or low-trust group work.

7. I need to be trusted and not micromanaged. This looks like being given ownership, not oversight. This should not look like needing to justify every decision or follow rigid processes.

8. I’m motivated when I’m learning something new, like systems, tools, or methods. This looks like solving new problems or exploring new tech with real purpose. This should not look like being forced to learn tools that don’t interest me or aren’t useful.

9. I appreciate being recognized for good judgment and smart work. This looks like people acknowledging when my thinking led to a smart outcome. This should not look like shallow praise or my contributions being overlooked.

10. I work best when I can follow a personal rhythm, like slow mornings and a thoughtful pace. This looks like flexible hours, async work, and trust in how I manage my day. This should not look like performative hustle culture or rigid 9-to-5 expectations.

What matters most

Must-haves
If these are missing, I shut down.

1. Autonomy to choose how I work

2. Problems that are interesting and meaningful

3. Deep, focused time

4. Clarity about priorities and goals

5. Being trusted, not managed

Nice-to-haves
These make the experience better, but I can flex if the core stuff’s there.

1. Building things that help people

2. Small teams with smart collaborators

3. Freedom to work at my own rhythm

4. Recognition for thoughtful work

Lower priority
Still nice, but not dealbreakers.

1. Learning new tools or methods, as long as they’re useful

Why this works

I didn’t just get a list of tasks I like. I got clarity on why they matter and what happens when they’re missing. It’s not about avoiding bad jobs. It’s about understanding what helps me do my best work in the first place.

That kind of insight changes everything. Most people never pause to get it. They just react.

This gave me language to make better decisions. That’s the real win.

Try it for yourself

1. Copy the questions above

2. Answer them honestly

3. Ask ChatGPT to answer them too, based on your previous chats

4. Then ask, "What patterns do you notice"

You don’t need a special prompt. Just a little curiosity.

Worst case, you reflect for 20 minutes
Best case, you rethink your whole approach to work

Imagine a soccer team obsessed with pass completion. Crisp footwork, great team work, looks impressive. But what if they never score? Feels ridiculous.

That’s exactly how most product, growth, and small business teams operate. They track tasks, close tickets, ship features, and celebrate output. Meanwhile, impact? Nowhere to be found.

John Cutler calls this the “feels like progress” trap. More work doesn’t mean more results. And “getting things done” is pointless if those things don’t move the needle. Here’s how to stop playing the wrong game:

1. Stop Measuring the Work. Start Measuring the Win.

Your success metric isn’t:
🚫 Features shipped
🚫 A/B tests run
🚫 Jira tickets closed

Your success metric is:
✅ Revenue
✅ Retention
✅ Customers who actually care

Everything else? Just passing the ball around.

2. Kill Fake Productivity.

John Cutler’s visual says it best:

• Starting feels productive. Finishing is productive.

• Doing more things at once feels fast. Focusing on fewer things gets results.

• Individual effort looks efficient. Working together creates impact.

Busy ≠ effective. Let’s move on.

3. Shift from Work to Outcomes.

Your team isn’t here to check boxes. They’re here to win. That means:

• Impact > Activity. Ten features that change nothing are worse than one that does.

• Speed = Faster Learning. Not doing more. Figuring out what works, sooner.

• Deep Work Matters. The best break throughs happen when you give them room.

4. Measure What Matters.

Running hard in the wrong direction doesn’t get you bonus points. Stop tracking internal vanity metrics (tasks completed, hours logged). Shift to external ones (customer conversion, retention, expansion). If it doesn’t tie back to business goals, it’s just noise.

5. Fix Problems, Not Just Backlogs.

When your team hits a problem, don’t just go around it, fix it. Solving blockers means pulling the right people together to remove obstacles now, instead of pretending they don’t exist for weeks.

You Don’t Need to Do More. You Need to Win More.

Your team’s success isn’t measured by how busy they look. It’s measured by outcomes. Stop celebrating passes. Start scoring goals.

Want to talk more about this? Message me.