no barrier reef
letter nº 50 2026 · 10 · 08 / ≈ 6 min read

Your app is working. That's why they'll cancel.

Hey,

Talked to a founder a couple of weeks back whose app runs behind the scenes for Shopify merchants.

His niche isn’t very competitive. That’s the good news. It’s also the bad news.

Harder to get users. It’s a back-end app, not an upsell app. There’s no obvious “here’s the money I made you yesterday” number to show a merchant, and new users are harder to get. And a change on Shopify’s side means more apps will be going after the same thing, so the quiet won’t last.

But retention is much better. If you’re the solution they need, they just use you and aren’t tempted to hop from one upsell app to the next. You do something mostly once and it keeps going.

That only works if merchants notice.

The trap is connect and forget. On the call, the founder told me his real challenge. “It’s more getting people to actually utilize it and use it.” He figures the paid part will come “once they actually see the value, but I just don’t think they see the value.”

Then he said, “They kind of connect it and that’s about it.”

A merchant who connects an app and never sees it do anything has no reason to keep paying for it. Better retention doesn’t help you if nobody can tell the app is working.

Show one number that proves it’s working. Make it the hero of your home screen. Ask what would make a merchant feel foolish for canceling. Then run it through the three-second test: a busy merchant gets it at a glance, with no tooltip, and it traces back to why they installed your app.

It’s usually one of these:

  • Money made
  • Money saved
  • Time saved
  • Problems prevented

Put this month next to last month so the trend is obvious. Everything else goes one tab deeper.

If your app works behind the scenes, the number won’t be obvious. Paste this into your AI of choice and fill in the brackets:

I run a Shopify app that [what it does, one sentence]. Merchants install it to [the main reason they install it]. Here’s what my app tracks: [list the data points you have].

Help me find the one number my app’s home screen should lead with.

Rules:

  • It must trace back to why the merchant installed the app.
  • A busy merchant must understand it in three seconds, with no tooltip.
  • It answers one question: “Is this app still worth what I’m paying?”

Start from this question: what number would make a merchant feel foolish for canceling?

Give me 3 candidates (money made, money saved, time saved, or problems prevented). For each, show how to calculate it from the data I listed. Then pick one, explain why, and suggest how to show its trend against last month.

I laid out the full method in the post on turning your home screen into a retention tool.

Keep them through the slow season. Pricing has a job here too. Merchants should upgrade as your app helps them grow. From free to paid, paid to growth, growth to scale.

But it also has to work the other way.

Say a merchant gets a boom in Q4, Black Friday, whatever. Lots of orders. Then fewer orders. They might uninstall your app after Q4. If they can move back and forth between tiers, they don’t churn.

And take a hard look at the free tier. Are you giving your best features away for free?

Back to connect and forget. If merchants connect your app and then forget it’s there, that’s one of the first things I check in an audit: what they see the second they open it.

Book a call and we’ll go through yours together.

Ohad

Gate: “PASS (content-editor review 2026-10-07, sources: call transcript 2026-09-22_11-00 and blog post 2026-10-04_shopify-app-home-screen-retention only). Fabrication gate clear. Edits: dropped the causal ‘so’ that the call does not state (no-number -> harder users; once-and-keeps-going -> not tempted to hop), cut ‘the line that stuck with me’, replaced the muddy ‘invisible and sticky’ closer (never set up in the body) with a concrete question tied to the three-second test, retitled the first bold lead-in. ‘No OBVIOUS … number’ tweak judged OK: the call (L518-520) says a back-end app cannot show ‘I made you $100 yesterday’, and ‘obvious’ keeps that true while staying consistent with the find-your-one-number fix. Prompt block untouched. No founder/app/category names, COGS/inventory terms, em dash, ‘lands’, percentage, or death metaphor. Two body dividers plus this Gate block. Subject and send exclusion untouched. Not pushed to NitroSend.”

Claim-to-source list (transcript 2026-09-22_11-00 in Reference/Transcripts/, filename withheld, it names the client; blog = ohadmichaeli/src/content/blog/2026-10-04_shopify-app-home-screen-retention.md):

  • Founder’s app runs behind the scenes for Shopify merchants: L518 “I call this a back-end app.” Category withheld.
  • “His niche isn’t very competitive”: L302 and L520.
  • “A change on Shopify’s side means more apps will be going after the same thing”: L294, L302. Tool name withheld.
  • Back-end vs upsell, no “made you $100 yesterday” number, harder to get new users (listed in parallel on the call, not stated as cause): L518-522.
  • Retention much better, you are the solution they need, not tempted to move between upsell apps: L522. “You do something mostly once and it keeps going”: L526 (separate point, not the stated reason).
  • “It’s more getting people to actually utilize it and use it”: L120.
  • “once they actually see the value, but I just don’t think they see the value”: L122 (original: “The paid thing will come once they actually see the value, but I just don’t think they see the value.”).
  • “They kind of connect it and that’s about it”: L124 (original continues “from the analytics that I’m seeing”).
  • “A merchant who connects an app and never sees it do anything has no reason to keep paying”: inference from L120-124, stated as analysis, not attributed to the founder.
  • Ask what would make a merchant feel foolish for canceling; number must trace back to why they installed it; no tooltip: blog “How to Find the One Number That Proves Value” and FAQ “How do I know which number to put on the home screen?”
  • Three-second test (merchant wants the answer in the first three seconds): blog “What Do Founders Put on the Home Screen, and What Do Merchants Actually Look For?” and “Why Do Merchants Check In Instead of Analyzing the Data?”
  • Money made / money saved / time saved: blog “How to Find the One Number That Proves Value” list. Problems prevented: blog FAQ “Does this apply to apps that don’t generate or save money directly?” (“issues prevented”) and the approved prompt wording.
  • Trend against last month: blog “What This Looks Like in Practice” (green arrow next to last month’s number) and the list item “A trend arrow”.
  • Everything else one tab deeper: blog “Where Should the Rest of Your Data Live?” and Key Takeaways.
  • AI prompt: verbatim, wording approved by Ohad in the brief (not in the blog post).
  • Upgrade as the app helps them grow, free to paid to growth to scale: L476.
  • “It also needs to lead the other way”: L478.
  • Q4 / Black Friday boom then fewer orders, might uninstall after Q4: L480-482.
  • Move back and forth between tiers so they don’t churn: L484 (Ohad’s “turn” = churn).
  • “What are your best features? Are you giving them away for free?”: L490-492.
  • Closing question (“what a merchant sees the second they open your app”): blog “What This Looks Like in Practice” (what a merchant sees the second they log in) and three-second test.
  • CTA close claim “one of the first things I check in an audit: what they see the second they open it”: blog 2026-10-04_shopify-app-home-screen-retention.md L74 (“This is usually one of the first things I check in a Shopify app audit: what a merchant sees the second they log in.”). Verdict 2026-10-07 (CTA change): PASS. Claim traces to Ohad’s published post; audit framed as diagnosis, no fix promised; added bridge “Back to connect and forget.” because the preceding free-tier paragraph made the close a non-sequitur; call link tidycal.com/ohadm/discover untouched; no em dash.
  • Consistent with LinkedIn post tue-1006 (same L302/L518-526 facts). Founder identity anonymized throughout.