Legacy replacement
What does it cost to replace an Access database with a web app?
Search this and you get programmatically generated pages that swap the city name in the title, forum threads from 2010, and American conversion vendors. Here is the arithmetic instead, which is unusually tractable for this kind of project.
It prices from things you can count
Most software estimates are hard because the scope is a conversation. An Access replacement is different: the thing already exists, and it tells you its own size. Open the database and look at the objects list.
| What to count | Why it drives cost |
|---|---|
| Tables | Each becomes a data model and a migration. The relationships matter more than the count. |
| Forms | Each is a screen to rebuild, with validation and permissions it never had. |
| Reports | Often the most-loved and least-documented part. Frequently half the work. |
| Queries and macros | Where the business rules actually live. The archaeology happens here. |
| Concurrent users | One user is a different product from thirty. This is usually why you are leaving. |
| Scale | Developer days | At £525/day |
|---|---|---|
| Small: a handful of tables, under 10 forms, one or two users | 15–30 | £8,000–16,000 |
| Departmental: 15–30 tables, 20–40 forms, a dozen users | 35–70 | £18,000–37,000 |
| Business-critical: dozens of forms and reports, integrations, 30+ users | 60–120 | £31,000–63,000 |
The expensive part is not the rebuild
Rebuilding the screens is the predictable bit. The cost and the risk sit somewhere else: these databases have been in use for fifteen or twenty years, and in that time they have accumulated exceptions. A field that means something different for one customer. A report nobody can explain but finance refuses to lose. A macro that fires on a Tuesday.
Every one of those encodes a real business rule, and almost none of them is written down anywhere but in the database. The work is deciding which rules still matter, which were workarounds for an Access limitation that no longer applies, and which were a mistake in 2011 that everyone has since adapted to.
That is not a technical exercise, and it cannot be done by the developer alone. It needs the person who actually uses the thing, in the room, for a day or two. Budget for that explicitly; the projects that go wrong are the ones that skipped it.
What makes the data migration hard
Not volume. Access databases are small by modern standards. What costs time is everything that was allowed because nothing stopped it: dates stored as text, duplicate records that differ by a trailing space, a “Notes” field used for three unrelated purposes, and relationships enforced socially rather than by the database. Cleaning that is often a third of the project, and you only find out how bad it is by looking.
Hosting it is not replacing it
There are services that will put an Access database on a remote desktop and call it cloud. It works, and it keeps every single thing that made you want to leave — the concurrency limits, the single point of failure, the file that one person has open. It can be a sensible stopgap while a replacement is built. It is not a replacement.
The cheap first step
Do not commission a rebuild from a paragraph. Pay for two to three days of discovery: someone opens the database, inventories the objects, interviews the people who use it, and comes back with a scope and a range. On this kind of project discovery pays for itself more reliably than on any other, because the existing system is a specification that nobody has read.
Sources. Day rate: ITJobsWatch, UK contract software developer, median £525/day, six months to 6 October 2026, retrieved 7 October 2026. The day ranges, the counting method and the migration observations are my own from doing these, not a published benchmark.
Still running the business on Access?
Send me a screenshot of the objects list — tables, forms, reports. That is genuinely enough for a real estimate, and you get one back the same day.
Start here →