^{git commit -m “uncomfortably constraining the thing, and picking my egg”}

Hey everyone, today’s letter comes from a simple to understand yet hard to execute observation: the consumer apps that succeed do so by doing one single thing really well.

I’ve worked on both B2C and B2B in my career, and consumer is much harder.
From the growth side, you need distribution cheap enough to reach a lot of people. From the product side, you need to get all those people you reached and convinced to try your thing out to use it for long enough to form a habit.

And your marketing message has to perfectly match what the user actually experiences in the first 30 seconds inside the app. If your homepage promises X and your product shows them A+B+C or just D, you've broken the contract before they've pressed a button.

All those hard things, and then some, are why consumer is just hard.

But there is something that makes the hard thing an easier thing, so today I’ll share one finding from my own data, blended with research from How They Grow and further digging, that’s driven an impactful change in my app.

TLDR: it’s about a single unambiguous hook, and a strategic constraint deep enough that the experience feels almost silly simple.

Let’s get to it

—Jaryd

^{git status / Little-Moments}

Right now the project I’m working on is Little Moments. In under 60s a day turn one photo from your camera roll into a saved memory using your voice, which become beautiful printed keepsakes.

It’s a react native IOS app that’s available on the app store, with 58 users, just 1 paying customer, $10MRR, three 5 star reviews and one 1 star review. Try it here

^{git checkout Todays-Partners}

Every founder I know has a list of things that need to get done this week—and won't. Tendem is a new AI × Human partner that actually closes the loop: it picks up the tedious work, starts the work with the best AI models, and the unique part—a real human finishes it and ships it back to you done. Finding contacts at target companies, investor updates, competitor tear-downs, market research—whatever real work you need finished accurately.

They’re giving you guys $50 in free credits. Use code HOWTHEYGROW10
No setup. No API key. No learning curve. Upfront pricing per task.

See how it works → Tendem, by Toloka

_{P.S - I started with contact-finding for sales— the human check made the output much better}

HubSpot and Marketing Against the Grain put out a pack called Claude Hacks for Marketers. Don't let the title throw you. It's ~60 Claude Skills, and at least half are pure operator work: AI Opportunity Matrix, MVP Feature Specification, Decision Framework Builder, 90-Day AI Transformation Roadmap, Stakeholder Influence Map. Each one's built as a full Skill with inputs, diagnostic process, and output format—not a one-liner.

Free to download: How to Use Claude Like a Marketing Power User

_{P.S. — the Mini Problem Solver skill alone is worth the download}

^{git diff / The main thing for you to review +1 constrained feature -3 open features}

A few weeks ago I got my hands on the new PostHog Code app, which marries my analytics and user signals with my coding agents in one place. I’d been wanting to try this for a while—having my live product data sitting one prompt away from the code, so questions I'd usually save for "I'll get to it" just become "ask, fix, ship."

I asked it to review my data around onboarding and early new user patterns. To find where people were dropping off for what types of prompts, and to come up with a list of ideas to improve retention based on data and session recordings.

The answer was odd, because it went against what my current hook was—single word-based prompts (e.g Candle) as a memory trigger.

Here’s what it told me: Photo as the trigger x Voice to capture it is a winning combination for engagement by far, despite photo being 25% of my prompt cycle.

Not only were people saving more regularly with photo and adding more depth to answers, but they were also using Dig Deeper to enhance moments at a much higher rate than any other prompt type.

With photo, people were actually telling stories.

This showed me that despite constraining the app in v1 → v2, there was room for more constraint. More simplicity.

In consumer apps D1→D7 engagement and retention is literally make or break. No second chances. So despite a small sample, directional data has to be enough to make a decision or I’ll just be burning the precious few new users that try it.

I'd built a Swiss Army knife when the data told me users wanted a butter knife.

So, I decided to cut the 3 prompt types and focus on making the one that was doing most of the work really good, and emphasize the voice combo.

Because that’s what people usually want; it’s what I want when I download an app. We want just one thing from it. And we want to…need to…feel something about it inside 30 seconds.

With the product constraint, I also white boarded in Claude Design and honed onboarding around one path: splash → auth → photo permission → see your first photo → capture your first moment → see it in your Capsule.

The pitch on getlittlemoments.com collapsed to this: Photo + voice, turned into a real keepsake in under 60s a day.

The app collapsed to the same thing. The hook on the marketing page is now identical to the hook in the app.

That last point is a horse worth beating once more.

The hook on your homepage and the hook inside your product have to be the same one thing. If your homepage promises to help people with their taxes and the user's first screen inside the app is about their credit score, you've broken the contract before they've even started. Marketing-hook ≠ product-hook is a very expensive bug in consumer.

Founders want to show off all the features. The consumer just wants the egg.

Grant's mantra at Gamma was that the first 30 seconds had to be dead simple. Not the whole product. The first 30 seconds. Get that magical, earn the next 30 seconds, then the next. That was their entire focus the year they figured out PMF.

I'd been throwing four eggs.

I love a good metaphor, and this one egg idea was an interesting one. So I looked for more proof before punting it…

I went hunting for the pattern in other consumer apps that did the same thing well, and the ones that tried and disappeared.

BeReal — one notification, two cameras, two minutes. That's the entire product surface. They went from 500K users in late 2021 to 20M DAUs and iPhone App of the Year by end of 2022. Then plateaued, and got bought by Voodoo for €500M in 2024. The product was famous for what it refused to be — not stories, not feeds, not filters, not on-demand. One window. Or nothing.

Wordle — One word a day. Same word for everyone. Six guesses. He shipped it in October 2021 with 90 daily players. Two months later: 300,000. Two weeks after that: 2 million. The New York Times bought it. There's a version called Wordle Unlimited that lets you play as much as you want with hints, categories, and a streak tab. It's also the one nobody talks about.

Instagram — the one that hit close to my decision. Instagram had to cut before they could grow. Burbn was Kevin Systrom's first app. A check-in app with photos, plans, points, a friends graph, the works. He looked at the data, saw people only used the photo bit, and ripped 80% of the product out. One screen. One filter wheel. One feed. Hit a million users in 10 weeks.

Calm — the "Do Nothing for 2 Minutes" web page validated demand existed. The product didn't take off when they bundled more meditations, more teachers, more music, more sleep stories. It took off when they shipped Daily Calm — one 10-minute meditation a day, with a quote at the end. The rest is what they did after they had a habit.

And the smoking gun comes when looking at the many-egg attempts…

Quibi — $1.75 billion. Failed in six months. They tried to be a phone-first streaming service with horizontal and vertical video, original content, ten-minute episodes, celebrity casts, prestige drama and reality and news, with no sharing, no clipping, no embedding. They could've been "10-minute shows you watch on the subway." They were instead "streaming reinvented." Streaming did not need reinvention. It needed an egg. I wrote a full piece on this here

Color Labs — $41M raised pre-launch. The pitch was "see what people near you are taking photos of right now." On day one the app shipped with multiple modes, group photos, broadcasting, listening, an explore tab, and a "people near you" surface. The reviews were 2 stars. They shut down inside 18 months.

Path — Path 1.0 shipped in 2010 as a quieter Snapchat for your inner circle. Tabs. Explore. Friends button. Profile. Activity feed. Dave Morin himself called v1 "70% failure." They rebuilt v2 to be simpler, then over the years piled in messages, songs, fitness, location, life milestones. By 2018, they shut down. Every eulogy I've read uses the same phrase: a jack of all trades that didn't have an identity.

Stack the wins next to the losses and the pattern is clear:

Every winner had an almost silly constraint. One photo. One word. Two minutes. One meditation a day. Silly. Restrictive. Almost insulting if you described it aloud in a planning meeting. And that's the entire product.

I don't have a clean framework nor think we need one, just pause and ask yourself when facing a "should I add this?" decision. Even now writing these out here I feel like I still have too much.

Rather be loved and hated for your constraint by polar sides, than lukewarm in the middle. Nobody wins in the middle.

^{git push —force / The one thing to remember}

The hook on your homepage and the hook inside your app have to be the same one thing.

Not three things. Not "the platform for X." One thing. One egg. With a constraint silly enough that a fool can repeat it back to you.

Then go do this:

I cut 60% of my app and signup conversion increase by 40%

The thing I said recently but I keep telling myself: when more is easy, less is so much more. The cost of adding has gone to zero. The cost of carrying it hasn't. That gap is where most consumer apps go to die.

Cut until it hurts. Then cut one more. Until nobody is dropping eggs.

^{git add —a few extra things}

Every week, I highlight tools I love and those I actually use as full time staff product manager and part time solo builder. Today, I’m highlighting beehiiv—the platform I use to write, grow, and monetize my newsletter. Moved from Substack, and it’s been making newsletter ops much easier.

See the full set of vetted tools I use all the time here → Jaryd’s Stack

ICYMI, here are some recent entries from The Diff

If you’re looking for something else to sink your teeth into right now…

Loved this breakdown by Tom from Strategy Breakdowns on Perplexity Computer. Instead of just writing about it, he built a real working tool (Reddit Radar) inside Computer to prove his thesis—5 conversations, 2 evenings, 0 code—and lands a sharp read on why orchestration, not the model, is becoming the moat.

Whenever you’re ready, here are a few ways we could work together