Everything between the tap
and the receipt.
One SDK call in the app, one dashboard for everything the app shouldn't have to know: who is entitled, what they see, what it earned, and where that has to be forwarded.
Entitlements in one line
Ask the profile whether the customer is in. Grace periods, billing retries, refunds and Family Sharing are already folded into the answer before your code sees it.
A purchase finishes last
The StoreKit transaction is only finished once our server has recorded it. One that never landed stays unfinished, and the next launch retries it — a paid customer can't quietly end up with nothing.
Action keys, not releases
Put PREMIUM_LOCKED in your code once. Which paywall it opens, for which audience, in which experiment arm — every part of that answer lives server-side.
One event taxonomy
Apple's V2 notifications and Google's RTDN arrive signature-verified and land on one set of names, so a trial that converted reads the same on both stores.
Audiences that show their work
Build a rule, see the reach with its denominator attached — 24% of 5,000 profiles, never a bare percentage that could mean anything.
Your stack, still yours
Processed events go out to webhooks, Amplitude, AppsFlyer or Meta CAPI — or the store's own payload, byte-for-byte, if you'd rather parse it yourself.
Change the offer
while the app
sits in review.
Pick a template, edit it against a live phone preview, publish. The device gets flat, localized strings and renders — every decision was already made on the server.
- ✦Prices in the customer's own currency, read from StoreKit — including the struck-through "before" price a discount implies.
- ✦Exit offers, up to three hops. A close tap can lead to a better deal instead of an empty screen.
- ✦Dismiss surveys that answer back. Ask why they're leaving; a "too expensive" can show the annual plan and hand the paywall back.
- ✦Every string, any locale, translated in one click and reviewable before it ships.
unlocked.
- Unlimited sessions
- Offline mode
- No ads, ever
Design the exit,
not just the offer.
Most of the revenue is decided after the close tap. Draw what happens there once, on a canvas, and the SDK plays it out — survey, info screen, second offer, push — without another line of app code.
Four kinds of booster
Info screen, survey, message and push notification — each edited and previewed as the screen it will actually be, not as a row in a list.
Every branch reports
How many were leaving, how many stopped to answer, how many stayed, and what that was worth. Recovered revenue is the number that stands on its own.
The objection tally
One reason for leaving usually dominates — that's the headline. The sentences people typed sit underneath it, because those are what tell you how to fix it.
Numbers that agree
with the stores.
Built on the store's own notifications rather than on what the app remembered to send — so a refund is a refund, a grace period isn't churn, and an experiment never shows a lift without the confidence beside it.
week 1 → week 8
Cohorts, not vanity totals
Curves and triangles side by side, pooled or split, so "retention improved" has a week attached to it. Break any metric down by store, product or country.
Experiments with a verdict
The page answers one question — can this be shipped yet. Lift and confidence appear together, and neither appears at all until both arms have enough views to mean anything.
Delivered, or told
exactly why not.
Every outbound event is written down before anything is sent, so a crash mid-flight leaves a row to retry instead of a hole in your funnel. Connect each destination up to three times — one per environment, per consumer, per account.
Retry ladder
Six attempts over roughly eight hours. A timeout, a 429 or a 5xx is retried; a 4xx is your side and would fail identically forever, so it stops. And a customer who declined tracking is skipped, never failed — consent is not an error rate.
Day one shouldn't
look like a quiet week.
A fresh subscription dashboard normally starts empty and fills up one webhook at a time. Kebabty backfills instead, and hands the workspace to a team rather than to one login.
Import your Apple history
Pull 180 days of production transactions (30 in sandbox — that's Apple's window, not ours) into profiles, transactions and events. Preview the shape before you commit, watch it run, resume it if it stalls.
Names you already know
The event taxonomy is Adapty-compatible on purpose, so the feed reads the way your team already reads it and your downstream dashboards don't need rewriting on the day you switch.
A workspace, not an account
The organization owns the apps and the billing. Invite by link, hand out one of five roles, and keep a sensitive app visible to only the people on it.
Keys that can't wander
Store credentials are encrypted at rest, webhook URLs are per-connection and token-pinned, and the analyst who reads revenue never sees a device id.
Prove who's asking
The public SDK key proves which app is calling, never which person. Turn on signed identity and your backend signs every user id the app claims — a stolen key can't claim someone else's subscription.
Free until it's
worth paying for.
Every feature on this page is in every plan. You start paying when the revenue we track for you is real — not when you want a second app or an experiment.
- ✦ The whole platform, nothing gated
- ✦ Unlimited paywalls, flows and experiments
- ✦ All six integration destinations
- ✦ 3 team members
- ✦ Everything in Starter
- ✦ Unlimited apps and team members
- ✦ Historical import from the App Store
- ✦ Priority support
- ✦ Custom rate on tracked revenue
- ✦ Extended data retention
- ✦ Migration support and onboarding
- ✦ SLA and a named contact
Before you wire it up.
Is Android supported?＋
Do I need an App Store release to change a paywall?＋
What happens if a purchase never reaches your servers?＋
Can I keep the analytics stack I already have?＋
How do you know which customer is which?＋
We're already on another provider.＋
Put your subscriptions
on the skewer.
Create a workspace, connect the store, drop in the SDK. The first entitlement check works before you've written a paywall.