no barrier reef
letter nº 54 2026 · 10 · 29 / ≈ 9 min read

You're about to fix the wrong thing

[3 minute read]

Hey,

“What should I fix first?”

15 people asked me some version of that on calls this year.

A founder asked me, “Like, would you want to attack ads or would you want to attack the listing?”

It came up in other words too.

  • “What should we focus on?”
  • “What should we build?”
  • “What are we doing wrong?”

Same question underneath. You can’t do everything, so what goes first?

Everything looks broken at once. Ads, listing, features, reviews, onboarding. All of it looks like it needs work, and you can’t staff all of it.

So the gut picks. And the gut picks whatever is most visible or most fun.

Ads feel proactive. Shipping a feature feels like progress. Either one works only if it sits where your app is losing people.

The trap is fixing things in order of annoyance. Often the loudest number is reviews or rankings. Those sit at the end of the chain. They’re built on installs, onboarding and retention.

If the break is in onboarding, your review count is the symptom. A better review-request email doesn’t reach people who never became users.

The fix is a map. Write down your funnel stage by stage:

  • impressions
  • listing views
  • installs
  • onboarding completion
  • first meaningful action
  • ongoing use

Then find the stage with the steepest drop from the one before it. Not the lowest absolute number.

Put that drop into one of three buckets, and each one points to a different first move:

  • Not seen. Impressions are thin. Check category placement, and the words in your title and subtitle against what merchants type.
  • Seen, but not installing. Views climb, installs stay flat. That’s a conversion problem, and it lives on the listing or in the positioning behind it.
  • Installing, but not staying. Installs look healthy, onboarding completion is weak. Fix onboarding first. Listing work won’t move it.

To run it on your own app:

  1. Write last month’s number next to each stage.
  2. Starting at listing views, divide each stage by the one before it. The smallest result is your steepest drop.
  3. Name its bucket and pick one fix from that bucket.
  4. Write down what you’ll leave alone while that fix runs.

If you can’t fill in the onboarding line, that’s your first fix. Set up the tracking before you change anything, or every decision after it is a guess.

Then change one thing, wait, and read it. Change five at once and you won’t know which one moved the number.

I wrote the full method, including what to skip even when it feels urgent, in the post on what to fix first in a Shopify app.

If you’ve got the funnel numbers and you’re still not sure which break to open first, Book a call and we’ll look at where yours leaks and what goes first.

Ohad