Skip to main content

8 min readPublished November 7, 2026Talavishiraaj Groups
Technology & Engineering

Vendor lock-in is usually described as a danger to escape, almost always by a vendor who would like you to escape towards them. It is more useful as a number: the cost of leaving, estimated before you sign rather than discovered later. Costed that way, lock-in is frequently the right trade, and the mistake is not accepting it but accepting it without knowing what it costs.

In short
  1. 01Lock-in is not using a platform. It is the cost and time of leaving one, which is a number you can estimate in advance.
  2. 02Four things create it: your data model, proprietary extensions, integrations built around it, and the operational knowledge your team accumulates.
  3. 03Operational knowledge is the component nobody budgets for and it is often the largest.
  4. 04Accepting lock-in is frequently correct. Avoiding it entirely means building more yourself, which carries its own permanent cost.
  5. 05Ask for an export of your real data during evaluation, not a promise of one in the contract.

What is vendor lock-in?

Vendor lock-in is the situation where moving off a product costs enough that you stay on it regardless of whether it is still the right product. The lock is the switching cost, not the dependency itself.

That distinction gets lost in most discussions of the term, and losing it leads to bad decisions. Every platform you adopt creates some dependency. That is what adopting a platform means and it is not a failure. The question is how much leaving would cost, and whether that figure is acceptable given what you get in return.

Framed as a number rather than a risk, it becomes something you can weigh against the alternative instead of something to be feared.

What actually creates it?

Your data model is the first and most underestimated. Once your records live in a vendor’s structure, with its field names, its relationships and its particular way of representing an exception, moving means a translation project rather than a copy.

Proprietary extensions are the second. Workflow rules, custom objects, scripts and automations written in a vendor specific language do not travel. The more of your business logic lives there, the more of it you rewrite to leave.

Integrations are the third. Every other system wired to this one has to be rewired, and those connections are usually built once, by someone who has since moved on, with no record of why a particular field was mapped the way it was.

The fourth is the one nobody budgets for. Your team knows this platform. They know the workarounds, the reporting quirks, the thing that breaks at month end. That knowledge took years to accumulate and leaving resets it to zero, which shows up as months of reduced output rather than as a line in a migration plan.

What does leaving really cost?

More than the migration quote, because the quote covers moving the data and the data is the easy part. The expensive items are rebuilding the integrations, re-encoding the business logic, running both systems in parallel while you prove the new one, and the productivity dip while people relearn their jobs.

Published figures put integration work at a third to a half of initial development cost in ordinary projects, and a platform migration is largely integration work. That is a reasonable starting point for an estimate when nothing better exists.

There is also a cost to not leaving, and it is the one that compounds. A platform you have outgrown shapes every subsequent decision: features get designed around what it can do, and the gap between what the business needs and what the system permits widens quietly for years.

Is vendor lock-in always bad?

No, and treating it as always bad produces worse outcomes than accepting it deliberately. The alternative to depending on a vendor is building and maintaining the capability yourself, which is not freedom from cost, it is a different and usually larger cost that never ends.

Lock-in is a reasonable trade when the platform does something genuinely hard that is not your differentiator, when the alternative is eighteen months of your own engineers building a worse version, and when the switching cost is proportionate to the value you get while you stay.

It is a bad trade when the thing being locked in is the part of your business that makes you different from your competitors. Outsourcing your differentiator to a platform every competitor can also buy is the version of this decision that is genuinely hard to recover from.

Worth noting who tends to write about this. Most of what is published on vendor lock-in comes from platform vendors, and the advice reliably concludes that you should avoid being locked into someone else. We sell no platform, which is why this piece can say that lock-in is often correct.

How do you price lock-in before you sign?

Ask for an export of your own data during evaluation, in a format you have to work with, and have someone technical actually open it. A contractual right to export is not the same as a usable export, and the difference only appears when you try.

Write down which parts of your business logic would live inside the platform, and get an estimate for rebuilding them elsewhere. You will not get a precise number. You will find out whether it is weeks or quarters, which is the decision relevant distinction.

Count the integrations the platform will sit behind, and treat each one as work to be redone. Then ask who would do that work and whether they still work here.

Put the resulting figure in the evaluation alongside the licence cost. A platform that is cheaper to run and far more expensive to leave is not cheaper, and that comparison is almost never made because the two numbers arrive at different times.

When should you walk away from a platform?

When the switching cost is rising faster than the value you get. That is the signal that matters, and it is measurable: if each year adds more integrations and more embedded logic without adding capability, the trade is deteriorating even though nothing has broken.

When you are designing around its limits rather than your customers’ needs. The moment a roadmap conversation turns on what the system will allow, the system has started making commercial decisions.

And when the vendor’s direction has diverged from yours. A platform being acquired, repositioned upmarket or quietly deprioritised is a change to your switching cost whether or not anyone tells you, and it is worth re-estimating rather than assuming the figure you calculated at signing still holds.

Where this work sits
Technology & Engineering →

We build and integrate the production systems a business runs on: internal platforms, customer facing applications, integrations, and the data layer underneath them.

Weighing a decision like this one?

Describe the situation and we will tell you how we would approach it.