"The Growth Guy" | Author, Growth Hacking for Dummies (Wiley & Sons) | Startup Mentor |

I’m publishing the canon for Anuj Adhiya’s Growth Operating System (Growth OS) — the 5 practices I use with early-stage teams.

• Weekly planning ritual (Mondays)
• Hypothesis-driven experiments (“If X then Y by Z% because [assumption]”)
• Decision-ready metrics (cohorts, activation, payback)
• Friday post-mortems → Monday planning
• Backlog hygiene (impact vs effort)

I’m also running weekly tests on how to make this machine-digestible so assistants cite the right page.

Canon + method + weekly results → https://www.notion.so/Growth-Operating-System-Format-Experiment-Index-268e586c41bf8091a9ffd01374c03bd1

DMs open if you’re testing similar things and want to compare notes.

Anuj Adhiya 📈

I was evaluating a role and realized I couldn't answer the only question that actually mattered: not "is this a good job," but "what are the odds this compounds into real career capital?"

Compensation I could model. Culture I could ask about. But the structural stuff - whether the mandate was real or advisory, whether the category was expanding or silently contracting, whether the equity story had any basis in the funding reality - required reading signals most candidates never look at.

I'd spent a lot of time doing exactly that for companies - evaluating startups by reading their public record rather than their pitch.

Funding structure. Hiring patterns. Org chart signals. Competitive positioning.

The same pattern recognition that tells you whether a company will hit the services trap also tells you whether a VP role has real authority or just the appearance of it.

So I ran the experiment.

I applied the framework to a job instead of a company.

It produced something useful enough that a friend asked me to do it for theirs. Then a few more people. Each one said a version of the same thing: "I wish I'd had this before my last role."

That was the signal.

I built Career Diligence.

I've spent a long time helping founders evaluate whether their bets are well-structured.

This is the same thing, for the other side of the table.

If you're a candidate evaluating a Director, Head of, VP, or C-level role right now - the company and role analysis is free. No account needed.

Check it out at 👉 careerdiligence.com

Building a Growth OS for startups | Testing how to “market to algorithms” | Documenting everything

Open article: Career Diligence by careerdiligence.com

Hit two small milestones over the last week:
• 7,000 LinkedIn followers
• 100 subscribers to my Operator Hypotheses newsletter

Both numbers feel simultaneously small and larger than expected.

I started posting and writing mostly to think in public and sharpen ideas.
Turns out a few people actually enjoy reading it.

Appreciate everyone who’s been following along.

Back to it.

I turned 300+ podcast transcripts from Lenny Rachitsky into a 60-second BuzzFeed-style quiz for product folks that went viral on Linkedin last week.

The original question was simple:
"What is an interesting way to repurpose a massive content archive without making it another answer engine?"

Instead of:
• “Top 10 lessons from Lenny’s podcast”
• “Summaries of growth frameworks”
• “Answer engine from all podcast guests”

I built:
“Which Lenny guest is your product soulmate?”

But it had strict rules:
- 5–7 questions
- Zero typing
- 60 seconds max
- Instant result
- Built to be shared

The north star wasn’t depth.
It was time-to-smile.

Click → match → share.

A pure vibes quiz would’ve been easy.
But shallow.
So instead of matching on personality, I matched on decision philosophy.

Across hundreds of transcripts, the same tensions kept appearing:
• Intuition vs data
• Speed vs craft
• Systems vs instincts
• Focus vs optionality
• Frontier bets vs compounding
• Short-term vs long-term

Every answer nudges you along those trade-offs.
Under the hood, it’s structured.
On the surface, it feels effortless.

Also the truth about virality is that people share things that make them look interesting.

If you match someone with:
“Guest #42 who's great but hardly anyone knows”
They won’t post it.

If you match them with:
Brian Chesky
Julie Zhuo
Elena Verna
Marty Cagan..now it travels.

Identity is social currency.
Being paired with a recognizable, high-status name makes the result feel worth sharing.
So yes, picking big names was intentional.
Recognition fuels distribution.

Then came the repetition problem.
If I only used 8–10 guests, most people would cluster around the same 2–3 names.
Feeds would look identical.
Novelty would die fast.

So I defined four philosophical clusters:
Human-centered / craft-driven
Systems / compounding thinkers
Frontier / velocity builders
Wartime / decisive operators

Each cluster needed multiple "anchor people".
That pushed it to ~18 guests.
Below ~15, clustering spikes.
Above ~20, extra variety barely improves the experience.
18 was the sweet spot for distribution.

Another deliberate choice:
No randomness.
Same answers = same result.
That allowed static share pages for each archetype so social previews rendered cleanly across platforms

This is important because most “fun” tools break at the share layer.
This one was built for share from day one.

What actually drove engagement? 
Not deep debate.
It was identity signaling.
“I got X.”
“That tracks.”
Tagging friends.
Light comparison.

It gave product/growth people a mirror they could post.
That’s it.
And that was enough.

Big takeaway:
Repurposing content doesn’t have to mean summarizing it or making it easier to search.
Sometimes you let people locate themselves inside it.

Same archive.
Different interface.
Completely different energy.

If you took it, who did you get?

If you haven't yet, try it out here: https://lnkd.in/eRxXdedt

Stuff I’d text you if we talked every week:

A few pieces this week that change how you think about building software, getting cited by AI, and doing analytics work that actually matters.

1️⃣ "How Google Picks Which Sentences to Cite in AI Mode: Reverse-Engineering 42,971 Citations" by Daniel Shashko
https://lnkd.in/enTWWMUq

→ A rare reverse-engineering study showing AI search systems favor short, declarative sentences near the top of structured pages - meaning how you structure content increasingly determines whether AI will cite you.

2️⃣ "The Interface Was the Product" by Ibrahim Bashir
https://lnkd.in/eb-PFsTj

→ The argument: traditional SaaS competed through UI, but when the interface collapses into conversation, the real product becomes the underlying capability and system design.

3️⃣ "Will Claude Code ruin our team?" by Justin Jackson
https://lnkd.in/egJf8pfE

→ As AI lowers the cost of implementation, the leverage inside product teams shifts away from who can build and toward who can decide what’s actually worth building.

4️⃣ "The Illusion of Building" by Marius Horatau
https://lnkd.in/e5xEMVPm

→ A great critique of how easy it has become to mistake activity for progress in the AI tooling era - shipping prototypes and demos instead of solving meaningful problems.

5️⃣ "It's about the strategy, stupid" by Timo Dechau
https://lnkd.in/e9Wbi4eQ

→ Most analytics work fails because it starts with data and dashboards instead of strategy - the real question is what the business is trying to accomplish and what needs to become visible to support that.

If any of these changed how you think or work, then it was worth the time.

taplio