Something structural broke in software pricing over the last two years, and most vendors are still selling as if it did not.
For twenty years the logic was airtight. Building software required engineers, engineers were scarce and expensive, and therefore renting a finished product at a few hundred dollars a month was obviously cheaper than building it. That arithmetic held for so long it stopped being examined.
It no longer holds everywhere. And the places it stops holding are exactly the places where the fattest software margins live.
Not all software is equally at risk. The useful question is what a vendor is really selling you.
| What you are renting | Exposure | Why |
|---|---|---|
| A workflow on top of your own data | Severe | This is now a weekend of work if you know your workflow |
| A dashboard over an API you already pay for | Severe | The data was never theirs to begin with |
| A system of record | Low | Migration risk, audit history and inertia are real moats |
| Regulated or compliance-bearing software | Very low | Nobody sane builds their own payroll tax engine |
| A genuine network | Very low | You cannot rebuild other people’s participation |
Notice that the exposed row is where an enormous amount of modern SaaS actually sits. A great many products are a competent interface over data the customer already owns, sold per seat, per month, forever.
If your product is a nice interface over my data, and I can now describe that interface to a machine in an afternoon, you were never selling software. You were selling the fact that building it used to be hard.
Seat pricing assumes headcount and value move together. That assumption is dissolving. If one operator using good internal tooling covers what four used to, seat-based revenue falls even when the customer is thriving. The vendor’s revenue is indexed to precisely the thing its customers are now trying to reduce.
We looked at this inversion in more detail in Backwards — AI priced itself on consumption at the same moment legacy software ran the other way.
Not collapse. Compression, and a change in what gets paid for.
The long tail of thin tools gets absorbed into internal builds. Category leaders survive on data gravity and integration depth, but with far less pricing power on the second and third product in the bundle. And the winners are the ones that sell an outcome you cannot generate yourself — a network, a proprietary dataset, a regulatory shield.
The vendors who suffer most will be the ones whose renewal conversation has been, for years, essentially "you already have all your data in here."
It would be easy to read this as permission to build everything. That is a trap. Internal software has a real ongoing cost — it just does not arrive as an invoice, so it is easy to pretend it is free until it is not.
The correct posture is not "build everything." It is "stop paying rent on things that were only ever expensive because building was hard." That list is longer than it was two years ago, and it will be longer again next year.
Analysis reflects LAMPWORK’s operating view and our own experience replacing internal tooling. See our build log, "We Rebuilt Our CRM in a Weekend."
We talk to founders at every stage — long before they're ready to sell.
Start a Conversation →