The most expensive decision in a Power Platform project is often made in the first five minutes: where the data lives.
SharePoint is already there, already paid for and already familiar. Dataverse is Microsoft's proper database for the Power Platform, and it needs a premium licence for every user. Both are the right answer somewhere. The expensive mistake is picking one without knowing where the line sits.
One of my clients came to me after a vendor built their entire field operations app on SharePoint lists. By month eight they had 12,000 records, screens that quietly showed only part of the data, and no way to stop one team seeing another team's rows. SharePoint was not the problem. Building on it as if it had no limits was.
What SharePoint does well
For a firm of 30 to 150 people, SharePoint lists are often the right start.
- No extra licence. Lists, Power Apps and Power Automate with standard connectors come with the Microsoft 365 business plans most firms already have.
- The data is visible. Anyone with access can open the list in a browser, filter it, and export it to Excel. Nothing is locked inside an app.
- Files sit beside the data. A certificate, a photo or a signed form lives with the record it belongs to, in the same site.
- It is enough for a register. An inspection register, a training matrix or a site diary for a firm this size is a few thousand rows a year, and most of them are closed and never touched again.
Where SharePoint runs out
The limits are real, and an app that ignores them fails quietly.
- Views slow down past 5,000 items. A list can hold millions of items, but a view or query that does not filter on an indexed column hits the list view threshold at 5,000. Index the columns you filter on, from the first day.
- Not every formula runs on the server. Filters and sorts on simple columns are passed to SharePoint. Sums, counts and some operators are not: Power Apps works them out on the first 500 rows it downloaded (2,000 at most) and says nothing about the rest.
- Permissions get messy. Item level permissions exist, but they do not scale well, and there is no column level security. "Subcontractors see only their own documents" is hard.
- Relations are flat. Lookups work. Many to many relations, cascading rules and a proper audit trail per field do not.
What Dataverse adds
- A relational database with real many to many relations and business rules enforced at the data layer, whichever app touches the data.
- Row level and column level security through roles.
- Formulas and aggregates that run on the server over the whole table.
- Auditing per field, and a structured base for Copilot and other AI features.
The cost is the licence. Power Apps Premium lists at around $20 per user per month in 2026, and storage beyond the included allowance is paid for. For a firm of 80 people where 30 use the app, that is a line in the budget every month, for as long as the app lives.
How to decide
SharePoint fits when the app:
- is a register or a log the firm keeps for itself: inspections, training, diary entries, timesheets
- is mostly about documents and files
Dataverse fits when the app:
- is used by outside parties who must see only their own records
- holds tens of thousands of live rows that a screen must search or total
- is part of a larger system with Dynamics 365, complex relations or AI over the data
If SharePoint is the answer, build it the way it needs to be built:
- Index every column a screen filters on, before the list grows.
- Read only open items. A screen lists what needs attention, not the whole history.
- Keep totals in a summary row that a scheduled flow writes, instead of counting the list in the app.
- One list per kind of thing (people, plant, jobs), linked by ID, so the next app can use the same lists.
The cost of getting it wrong
Moving from SharePoint to Dataverse later is not a button. Data has to be mapped field by field, attachments carried over by flows, security rebuilt and every flow reconnected. For a mid size app with fifteen lists that is easily three weeks of work. At £80 an hour, 120 hours of rework is £9,600, before the disruption to the people using it.
So the question is not which platform is better. It is what the app will be in two or three years. If it stays a register the firm keeps for itself, SharePoint is cheaper for the whole of its life. If it is heading towards outside users, heavy reporting or a larger system, start on Dataverse and do not migrate under pressure later.
Questions to ask before you approve a new app
- Where does the data live, and why there?
- How many rows will it have in two years, and which screens search all of them?
- Who must not see which records?
- Which columns are indexed, and which formulas run on the server?
- What does moving it look like if the answer changes?
Not sure which one your app needs? Send me a message. Half an hour on Teams, no charge, and you will know where your app sits.