Taking over a stalled Power Apps app

Someone in the business built it, which is the point of Power Apps, and it worked. Then it grew. The list it reads went past the row limits, so the app silently shows the first two thousand records. A screen that was quick with ten users is slow with a hundred. The premium connector it needed moved everyone onto a licence the finance director didn't budget for. The person who built it left, and the formulas are theirs.

None of that is a failure of Power Apps. It's a tool for building small, quick applications close to Microsoft 365, and the trouble starts when a small, quick application becomes the system.

• The logic, readable. A canvas app's source is stored as YAML with Power Fx formulas, and Microsoft's Git integration puts those files in a repository. That's a full description of every screen, control and rule. It runs nowhere but Power Apps, but it can be read, and it's the specification for whatever comes next.
• The data. Dataverse tables, SharePoint lists or a SQL database, all exportable. Dataverse and SharePoint data moves to a database you own; SQL data usually stays where it is.
• The flows. Power Automate flows are the app's background logic: approvals, emails, scheduled jobs. Exportable as definitions, readable as rules, rebuilt on the other side.
• The users and roles, as records, from Microsoft 365.

Fix it in place when the app is small, close to Microsoft 365, and stalled by something the platform can solve: a query rewritten to delegate properly, a screen restructured, a flow that fails quietly made to fail loudly. Power Apps is good at what it's good at, and we'll say so.

Rebuild on technology you own when the app has become the system. Row limits that can't be delegated around, performance that doesn't recover, a per-user licence bill that grows with the business, integrations the connectors can't reach, or the need to give customers access. The rebuild starts from the YAML and the flows as a specification, keeps the data, and runs alongside the Power Apps app until each process has proven itself.

Often both. Stabilise it in place so it stops hurting, then move the processes that outgrew it. The general keep, finish or replace reasoning applies, and the assessment makes the call from the source and the data.

1. Get the source into Git, using Microsoft's integration, so the app's definition exists somewhere you control and changes are visible.
2. Read it. The assessment: every screen, every formula, every data source, every flow, every connector, and the licence each one needs.
3. Read the data. What's in Dataverse, SharePoint or SQL, how clean it is, and how it comes out.
4. Decide, from evidence, per process: fix in place, or rebuild.
5. Rebuild alongside, where that's the answer. The new system runs in staging on the migrated data, checked by the people who use it, reconciled, then switched process by process. The Power Apps app stays until the last one has moved.
6. Or fix in place, priced as its own stage, with the same test-then-live discipline.

It comes out of Dataverse or SharePoint once, checked, and never meets a change until the change has proven itself somewhere else. The migration is tested against a copy in a test environment, then run in staging with every table and record reconciled, before anything connects to live. Where the data is already in SQL Server, it usually stays there and both systems share it.

The assessment is from £395 + VAT, sized on a free one-hour consultation, with an exact price before you commit. For a Power Apps app it reads the source, the flows and the data, and prices fixing in place and rebuilding as fixed numbers, alongside what the current licensing costs a year. What drives the price and the payment terms each have a page.

Read the source before deciding anything, and be willing to say the platform is fine. Marc Allington has worked in Power Apps and in the web and .NET stacks apps move to when they outgrow it. Everything we build runs on open technology in your own accounts, with full source code and ownership transferring to you on final payment, and no per-user licence.