Skip to content
AppDues
Log in
Contact salesSign up

Indie app launch playbook: beta polish, one press pitch, three emails

Two very different apps, one shared lesson: small teams win launches by making a few high-leverage moves well. A link-saving app called Plinky spent months in beta with real infrastructure behind it. A separate side project, an early AI chat wrapper, shipped in 19 days and earned most of its early money from a single email to a writer. This guide turns those two stories into a decision sequence you can run on your own app: what to build before beta, how to price the first paid tier, how to pitch press, and how to structure a discount campaign so a quiet first email does not read as failure.

Run the beta like the real product

Plinky's founder did not build in secret. He ran a beta for months and treated it as a live product, which meant real infrastructure rather than a throwaway build. The clearest example: when the underlying data structure had to change, he did not wipe the database. He built a full export and import path so the small share of testers who had invested real links would keep them. He later judged that the time cost did not pay back on its own terms, yet the work still returned value: when paying users asked for export after launch, the feature already existed and had been exercised by real data.

Use this as a checklist before you invite testers:

  • Decide now whether testers' data must survive a schema change. If yes, build export and import before the change, not after.
  • Treat beta feedback as production feedback. Bugs found by testers are cheaper than bugs found by buyers.
  • Accept that some beta work will not pay for itself directly. Its return shows up as launch speed and user trust.
  • Keep a written record of what you built for beta, so launch-week requests map to existing components.

One well-aimed press email beats a broad campaign

The AI wrapper's launch had no paid ads and no elaborate funnel. Six months after ChatGPT arrived, OpenAI released the GPT-3.5 API, and there was still no official ChatGPT iOS app. Two people built a barebones wrapper in 19 days and pitched it to John Gruber as a specific answer for Apple users, not as another generic AI tool. Gruber posted it on Daring Fireball. Payment alerts started arriving on the founders' phones: about $5,000 in the first two hours, and $10,000 within a day and a half.

The transferable move is the framing. The pitch named an audience and a gap, and it went to one writer whose readers matched that audience. A useful pre-send check:

  • Can you state the audience in one sentence, and the gap it fills in another?
  • Is the recipient's readership actually that audience?
  • Does the pitch describe a use case rather than a feature list?
  • Have you made the app easy to try immediately after reading?

One caveat worth holding: a single post produced an unusually sharp spike. Treat that as a captured example, not a forecast for your own launch.

Freemium versus trial: what the free tier hides

Plinky launched with a generous free tier: 50 saved links at no cost. A staff product marketer warned that these users would churn, and that is what happened. The founder's own verdict was that he had chosen freemium to be generous and came to regret it.

The revenue loss was only part of the problem. Free users who left did so invisibly, which removed chances to learn exactly when someone decided the app was worth paying for. The free limit was later cut to 10 links, and the plan moved toward a standard free trial, on the reasoning that charging sooner is the only way to measure real value accurately.

A practical way to choose:

  • If your product's value grows with stored data (saved links, notes, projects), a large free allowance can hold users in a state where they never confront the paywall.
  • If you need to learn why people convert, a trial gives every user a dated decision point; a free tier gives you silence.
  • If you keep a free tier, set the limit low enough that a typical active user reaches it, and instrument what happens at the boundary.
  • Watch churn among non-payers, not just conversion among payers. Invisible exits are the expensive ones.

The three-touch discount sequence

Plinky is priced at $39.99 a year. Its founder found that structured email to free users was the strongest revenue lever available, and he collects addresses up front through Sign in with Apple, which gives a direct channel that does not depend on App Store release notes.

A rule from his wife, a staff product marketer, shapes the offer: a discount below 50% is not worth announcing. So sales run at 50% off, and each one uses three touches:

  1. Announcement email at the start of the sale.
  2. Reminder a few days later, sent to people who did not open the first message.
  3. Final warning in the last 24 hours.

The last email typically moves the most people. That detail changes how you read early results: if the announcement email does not convert right away, the sequence is behaving as designed, not failing. Judge the campaign after the final warning, and compare the three touches against each other rather than against a single-email benchmark.

A short pre-campaign check:

  • Is the discount at or above 50%?
  • Does the reminder target non-openers rather than the whole list?
  • Is the final message clearly time-boxed to 24 hours?
  • Do you have a way to tell which touch produced each purchase?

Reading the signals after launch

Two failure modes look alike from the outside and need different responses.

A pitch that produced no coverage is a targeting or framing problem: revisit the audience sentence and the recipient. A campaign whose first email produced no sales may simply be mid-sequence: wait for the reminder and the final 24-hour message before changing the offer. Mixing these up leads teams to rewrite a working sequence or to re-pitch a writer who was never the right fit.

The founder also described organizing launches with a lookback calendar, open-sourcing a database library called Boutique, and watching an AI agent rewrite the wrapper's codebase in about three and a half hours. Those are planning and tooling notes rather than launch rules; the durable pattern is the sequence above: harden the beta, aim one pitch precisely, choose a pricing gate that produces visible decisions, and run discounts as a timed series.

Frequently asked questions

How long did the AI wrapper take to build, and what did it earn?

It was built in 19 days by two people after the GPT-3.5 API became available and before an official ChatGPT iOS app existed. After a Daring Fireball post, it brought in roughly $5,000 in the first two hours and $10,000 within a day and a half. That is a captured result from one launch, not a typical outcome.

Why did the founder regret offering freemium?

The free tier allowed 50 saved links. Users who churned left without leaving a trace, so there were fewer opportunities to see when someone decided the app was worth paying for. The limit was later reduced to 10 links and the plan shifted toward a standard free trial.

What are the three emails in the sales sequence?

An announcement at the start of the sale, a reminder a few days later aimed at people who did not open the first message, and a final warning in the last 24 hours. The final message usually drives the most purchases.

How is the email list collected?

Through Sign in with Apple at sign-up, which gives a direct channel to users that does not rely on App Store release notes.

What price is the link-saving app sold at?

It is priced at $39.99 a year, with periodic 50% off sales announced by email.