Consolidated learnings · updated 2026-11-10

The Micro-app Playbook

What we've learned from founders about building small apps that make money, consolidated across videos and checked against the web. Each learning links to the video evidence behind it. This is a working document: statuses and 'My thoughts' get refined as we go.

16learnings
6strong evidence
0adopted
1source videos
Evidence: ✓ confirmed · ≈ plausible · ? unverified · ✕ contradicted Status: proposed → adopted / rejected→ opens the video page at the evidence

Finding ideas

Where good micro-app ideas come from

L01 proposed strong

Shop for ideas on verified-revenue boards

TrustMRR lists real, Stripe-verified revenue for small products. Browse it to find niches where people already pay, and to sanity-check any founder's claims.

Our move

Before shortlisting any idea, pull 3-5 comparable products from TrustMRR and note their MRR, price and channel.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

L02 proposed strong

Build inside a marketplace that already has the users

Google Workspace Marketplace and the Chrome Web Store hand you distribution and trust. His add-ons got installs a solo founder could never buy: MagicSlides 1M+, SheetAI 201K, MagicForm 106K, SlackNotify 73K.

Our move

List under-served marketplaces (Workspace, Chrome, Shopify, Notion, Figma, Slack, Microsoft 365, new AI app stores) and look for jobs with few good add-ons.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

L03 proposed moderate

Reuse a winner on the next surface, but expect most to stay small

He took one idea across Sheets, Slides and Forms. Only Slides became big ($9.8k MRR); the others sit at $100-400 MRR. Adjacent expansion is cheap, not a guarantee.

Our move

Treat each adjacent surface as a separate bet with its own buy-button test.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

Validating

Proving people will pay before building

L04 proposed moderate

Put up the buy button before building

A 5-minute pre-order page promising delivery in 10-30 days gives a real payment signal. Start the build clock only when the first sale lands.

Our move

Every shortlisted idea gets a pre-order page and a small traffic test before any code.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

L05 proposed strong

Plan for a power law: one app will carry the portfolio

30+ launches produced one real business. MagicSlides is ~84% of his $868k verified revenue; the other 8 earners total ~$1.1-1.3k MRR. Many small bets are how you find the hit, not how you earn.

Our move

Set a kill rule (e.g. no paying users in 30 days → stop) and double down hard on the first app that shows pull.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

Building

Scope, stack and speed

L06 proposed strong

One app, one job

Single-feature apps are easier to explain, list and sell. 'When you add too many features, you don't even know yourself what you're selling.'

Our move

Write the one-line job before building; cut anything that doesn't serve it.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

L07 proposed moderate

Keep the stack boring and nearly free

Next.js, Supabase ($25), Cloudflare, Resend ($35) and Apps Script: about $60/month for the whole portfolio, with Claude doing most of the coding.

Our move

Standardise one starter template so each new app ships in days.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

Getting customers

Channels that actually bring users

L08 proposed moderate

Give each product its own YouTube channel

Search-friendly tutorial videos keep selling for years, and a channel per product keeps each audience focused.

Our move

For each app, publish 5 how-to videos targeting exact search phrases on launch week.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

L09 proposed moderate

Use lifetime deals for early cash and first reviews

AppSumo and Gumroad lifetime deals brought verified buyers and a review base (BlurWeb: 158 reviews, 4.58). The $85k AppSumo total he cites is not publicly verifiable.

Our move

Consider an LTD only after the core product is stable; LTD buyers are vocal about support.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

L10 proposed early

Run short-form video as cheap experiments

Ten phone-recorded Reels in an hour, then double down on what lands. The host's main takeaway too.

Our move

Batch-record 10 clips per launch; track views → site visits → signups.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

Charging

Pricing models and conversion

L11 proposed strong

Installs are not revenue: conversion and billing are the product

SheetAI has 201K installs and 36 paying subscribers; MagicForm 106K and 8. The top complaint on both: 'I paid and still can't access it' with no support reply.

Our move

Make paywall, billing and access bulletproof and answer support within a day. That alone beats his add-ons.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

L12 proposed moderate

Usage-capped tiers plus a done-for-you upsell

MagicSlides charges $8 / $16 / $29 a month with decks-per-month caps, plus a $99 designer-made deck. It is the only one of his apps with real recurring revenue (747 subscribers).

Our move

Default pricing template: free taste → 3 usage tiers → one premium human service.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

L13 proposed moderate

A bring-your-own-API-key free tier caps revenue

SheetAI lets users plug in their own OpenAI key for free. Power users never need to pay, and MRR fell from ~$916 to $388 in 2026.

Our move

Offer BYO-key only as a paid 'pro' option, not as the free tier.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

Trust and reputation

Reviews, ratings and social proof

L14 proposed moderate

Earn reviews honestly; inflated social proof is a liability

Signs of credits-for-5-star reviews, a founder self-review and site ratings higher than the stores show. All are against Google and Trustpilot policy and get called out in reviews.

Our move

Ask for reviews in-app after a success moment, never for a reward, and show only store-verified ratings.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

Traps to avoid

Patterns that look good and don't pay

L15 proposed strong

Generic 'chat with X' wrappers don't monetize

AskVideo made $4.2k lifetime, TalkingPDF has no verified revenue, MagicChat is at $0. Built-in AI in ChatGPT, YouTube and Gemini absorbs these.

Our move

Only build an AI wrapper if it produces a specific deliverable for a specific audience inside a tool they already use.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

L16 proposed moderate

Don't polish one product nobody buys

His biggest mistake was a year on one product, adding features and blaming users instead of moving on.

Our move

Review each live app monthly against its kill rule.

My thoughts

Nothing yet. Tell Claude in the thread and it lands here.

Open questions

Things to settle as more videos come in.

  • Which marketplaces are under-served right now, the way Workspace was for him in 2022?
  • Is a Workspace add-on still a good wedge in 2026 now that Gemini is built into Sheets, Slides and Forms?
  • How much of MagicSlides' success was timing (early 2023 AI wave) versus repeatable tactics?

Sources