Nobody complains about internal tools. That is the problem.
Internal systems have a captive audience. The people using them cannot go elsewhere, so they work around the problems instead of reporting them, and nothing generates the pressure that makes public-facing software improve.
A decade that did not move
Across 42 organisations’ intranets, employee task success was 74%. Ten years earlier it was 75%, while public websites reached around 80%. Nielsen Norman Group, 2012. The date is the point.
Where the budget already goes
Legacy systems typically consume 70 to 75% of total IT budget and around 80% of total cost of ownership. CISQ, 2022 — an estimate synthesised from 88 sources rather than a measurement, and worth treating as such.
Knowing is not fixing
Even on the flaws organisations worked hardest to patch, only about 54% were fully remediated, at a median of 32 days. Verizon, 2025. A system nobody owns does not get looked after by default.
When a web app is the answer, and when it is not.
The test is whether somebody is currently doing work the system will not let them do properly.
Worth doing when
- People keep a parallel spreadsheet because the system will not do something.
- Clients or suppliers need to see or submit something and currently email it.
- The same work is done differently by different people, and both are defensible.
- It has to work on a phone, on site, with no training.
Worth postponing when
- Nobody has watched somebody do the task end to end.
- The request is for a portal, with no first job named.
- An existing tool does it and the objection is the monthly fee.
- The people who would use it have not been asked.
Four things move the cost, and the interface is rarely the biggest of them.