Consumer cloud app developer: 10 checks before you sign

Business and ScaleSep 26, 20268 min read

Ten checks before you hire a studio for a consumer cloud app: sync model, store billing while the US rate sits at zero, Texas age signals, cost per user.

In this piece

A consumer cloud app is a product where the phone is the interface and the cloud holds the truth: accounts, sync across devices, media, push, store billing. The screens are the half you can judge in a demo. The cloud half decides whether the product still works at 50,000 users, and it is where most pre-hire conversations stay vague.

Vagueness costs more than it used to. RevenueCat's 2026 report, built on more than 115,000 apps, found the top quartile growing 80% year over year while the bottom quartile shrank by a third (RevenueCat, State of Subscription Apps 2026). US store payment rules changed twice in eighteen months and now sit with the Supreme Court. Since 1 January 2026, app stores hand developers an age category for new Texas accounts. A studio that ships consumer cloud apps tracks all three. A studio that ships dashboards will learn them on your budget.

Ten checks below. Six to insist on, four to refuse.

The scope a consumer cloud app actually has

Agree the surface before the checks. A consumer cloud app is six systems: the client on two platforms, an identity and account service, a sync path with a local store, media handling behind a CDN, store billing with entitlement state, and the operational side (push, analytics, crash reporting, support tooling). A studio that quotes on screens has quoted on one of the six. Ask for a line per system, with the name of whoever owns it after launch.

Do: get the sync model in writing before the first screen

Sync is the architecture. Ask which model they intend to use: server-authoritative writes with a local SQLite mirror, read-path replication out of Postgres, or CRDTs for records two people edit offline at the same time. Each choice carries its own conflict rules and its own worst day.

Then ask the vendor question. On 11 August 2026 Electric, one of the better-known sync engines, announced it was joining Databricks (Electric). Acquisitions are normal and this one may end well. The point is that your data layer should survive one. A studio that answers "we would swap the transport and keep the mutation contract" has thought about it. A studio that names one SDK and stops has not.

Three answers to get in writing: which fields are last-write-wins, which need a real merge, and what the user sees when a merge is impossible.

Do: map the billing path store by store and region by region

In the US, external purchase links currently carry no Apple commission. The rate has been frozen at zero since the April 2025 contempt ruling, the Supreme Court granted review on 30 June 2026, and Apple filed its opening merits brief on 14 September 2026 (SCOTUSblog, Apple v. Epic Games). In the EU the same link needs Apple's External Purchase Link entitlement and the fees attached to it. Argument is unlikely before 2027, so anyone quoting you a settled number is guessing.

The useful question is who does the work when the number moves, and whether that work sits inside the contract or needs a new one. Then ask where subscription state lives: the store receipt, your server, or a billing layer. Our comparison of RevenueCat and Stripe Billing covers that trade. If the answer is that the client decides what a user has paid for, stop there.

Do: treat age signals and account deletion as table stakes

Since 1 January 2026, Apple passes developers an age category for new Texas accounts (under 13, 13 to 15, 16 to 17, 18 and over), the assurance method used, and a parental consent signal, and the developer has to delete that data once the check is done (Apple Developer). It applies to every app on the store, not only apps aimed at minors, and other states have passed their own version of the same act (Wiley).

In-app account deletion has been required since 2022 under App Review guideline 5.1.1(v), including revoking Sign in with Apple tokens through the REST API (Apple Developer), and every privacy-impacting SDK ships a privacy manifest. None of this is exotic. A studio shipping consumer apps this year lists it before you ask.

Do: ask for cost per monthly active user, not just a project price

The project price tells you what the build costs. Cost per monthly active user tells you whether the product survives its own success. Ask for the sheet: object storage and egress per user, database reads and writes at the sync frequency they picked, push volume, transcoding if there is media, model tokens if there is AI. Then ask for the same sheet at ten times the users.

Consumer apps get killed on the egress line more often than on the salary line. Build ranges sit in our breakdown of what a SaaS costs to build. The recurring number is the one a consumer app lives inside.

Do: set the release cadence against real review times

Apple states that 90% of submissions are reviewed in under 24 hours (Apple, App Review). First submissions, apps handling sensitive data, and anything touching minors run longer. Plan for the slow case and ask for four things: feature flags so code can ship dark, staged rollout on Play, a forced-upgrade path for the day a server contract changes, and a policy for users who never update.

Without a remote switch, every mistake is a review cycle old before anyone can fix it.

Do: require paywall and retention instrumentation in version one

RevenueCat's 2026 benchmarks give you the yardstick. Hard paywalls convert at 10.7% by day 35 against 2.1% for freemium. Trials of 17 days or more convert at 42.5% against 25.5% for short trials. Median day-35 download-to-paid is 2.6% in North America against 1.4% in India and Southeast Asia (RevenueCat). Those figures are only useful next to your own.

So the event schema is a launch deliverable: paywall views, trial starts, restore failures, day 1, 7 and 30 retention, cancellation reason where the store supplies it. Analytics added in month three make month one unmeasurable.

Don't accept a web portfolio as proof

Marketing sites and admin panels prove a studio can ship a frontend. A consumer cloud app adds two review teams, offline states, background refresh, the timing of a push permission prompt, migration to a new phone, restore purchases, the screenshot pipeline, and crash triage across OS versions. Ask for two apps live in the stores, their current ratings, and the story of their worst outage. The warning signs show up in how that story gets told.

Don't let "offline support" stay a bullet

Offline is a set of decisions and the user sees every one of them. What ships inside the app bundle. What an optimistic write looks like on screen. What happens to a queued write when the token expired an hour ago. How two conflicting edits resolve. What the interface says when they cannot. Get those five in the proposal. Otherwise offline means the app caches the last screen, and your users find out on a train.

Don't leave stores, keys and cloud accounts in the studio's name

App Store Connect, Play Console, signing certificates, the cloud project, the domain, the analytics property: all in your legal entity from day one, with the studio added as a member. This is the most common lock-in we see in consumer work and it costs nothing to prevent. Put ownership in the SOW and verify it before the first build is uploaded, rather than during the argument.

Don't schedule compliance for launch week

Privacy labels, the deletion flow, age-signal handling, push consent, retention windows: each one touches the data model. Bolted on in the final sprint they produce rejections and rework, both on the critical path. Ask which sprint each one sits in. "Before submission" means the schedule already has a bug.

The ten checks, in order

Sync model in writing, with the conflict rules spelled out.
A vendor-exit answer for the sync engine they picked.
Billing path per store and per region, and who absorbs a rule change.
Age signals and in-app deletion already in their process.
Cost per monthly active user, today and at ten times the users.
Release cadence built on slow reviews, feature flags and staged rollout.
Event schema for paywall and retention shipped in version one.
Two live consumer apps, with ratings and an honest outage story.
Store accounts, signing keys and cloud project in your entity.
Compliance work placed in named sprints, not the last one.

The pattern under all ten: a consumer cloud app gets judged on the phone and decided in the cloud. Each check is a way to find out, before money moves, whether the studio has lived in the second half.

Sources

Photo by Markus Kammermann ↗ on Unsplash ↗

Frequently asked questions

What does a consumer cloud app cost to run per user?+

Two variables decide it: how much state syncs offline, and how much media each user stores. An app that syncs a few thousand rows per person and stores no video sits near zero marginal cost on most managed backends. An app with user video, thumbnails and a CDN pays storage plus egress every month, for every account, whether the person opens the app or not. Ask the studio for the per-user sheet at current scale and at ten times that, and ask which line they would cut first.

Do we still need an in-house engineer if we hire a studio?+

One technical owner on your side, yes. Not to write code: to hold the store accounts, approve the data model, decide during an incident, and read the cost sheet every month. A consumer app generates decisions faster than a weekly call can absorb them. With no owner inside, the studio makes those calls by default and you inherit the result.

Can one studio handle iOS, Android and the backend?+

Yes, and on a consumer cloud app that is usually the safer arrangement, because the sync contract touches all three at the same time. The risk is not headcount. It is a split where one vendor owns the client and another owns the data model, so conflict rules get decided twice. If you do split the work, name one side as the owner of the schema and the conflict rules, and put that in the contract.

What happens to our pricing if the Supreme Court restores Apple's commission on external links?+

Your margin on web checkout changes, and so does the case for having built it. The US rate is frozen at zero while the case runs, briefing continues into December 2026, and argument is unlikely before 2027. Ask for two clauses now: who implements the change when it lands, and whether the entitlement and price-display work is billable. Both are cheap to agree today and expensive to negotiate the week the ruling arrives.

Related services

Start a project.

We write about what we build. Tell us what you want to build.

Talk to the studio Read more articles