A martech stack rarely grows to plan. It grows because every year one more tool arrives that solves a real problem, and because almost nobody asks what gets switched off in return. After a few years you are looking at a landscape no single person can describe in full, and at contracts that quietly keep renewing.
- The most expensive item in a martech stack is rarely the licence. It is the time nobody budgeted for the rollout.
- Unused tools do not disappear on their own, because switching them off belongs to nobody.
- Cleaning up starts with an honest usage count, not with a vendor decision.
- Fewer tools does not mean less capability. It means the data sits in fewer places.
Table of contents
Why the martech stack keeps growing
The growth has a structural cause. Tools get bought locally, usually by the person who has the problem right now, and almost always on a budget small enough to stay below the line where somebody looks closely. Buying is therefore fast. Switching something off requires a conversation nobody wants to open.
The figures have been uncomfortably stable for years. Gartner finds that marketing teams use only around half of the capability they have paid for, and that a large enterprise runs well over eighty applications on average. Read through the research on marketing technology and you will not find an outlier. You will find the normal case.
What the unused tools actually cost
The licence fee is the visible part, and usually the smaller one. What costs more is everything the martech stack drags along with it in daily operation.
- Every additional system needs an integration, and every integration needs somebody to repair it when it breaks.
- Customer records exist several times over, in slightly different states, and in a dispute nobody knows which one is right.
- Each privacy review and each processing agreement costs time whether the tool is used or not.
- New joiners learn tools they will never open again after their third month.
None of this appears in a contract. It shows up as friction, as a campaign that slips, as one more approval round that eats a week. The question of the martech stack is therefore never only a procurement question. It is a question of how fast a team can move and how much its numbers can be trusted.
How I go about clearing it out
I do not start with the contracts. I start with how the martech stack is actually used. For every tool I ask for three numbers: how many people logged in during the last quarter, which process would stop without it, and where its data flows. That takes about two weeks and answers half the questions by itself.
Then I sort into three groups. What stays, because a process depends on it. What gets replaced, because another system already does the same job. And what gets cancelled without a replacement, because the job itself no longer exists. The third group is regularly the largest and regularly the most awkward, because behind each of those tools stands a person who once needed it.
One rule has proven itself: nothing gets switched off until it is named who takes the task on instead. Otherwise it returns within a year under a different product name.
What a lean martech stack makes possible
The real gain is not the saved licence. It is that the data sits in fewer places. Only then can you have a serious conversation about a strategy for your own data, because a strategy spread across seven separate stores is not a strategy. It is a statement of intent.
The second gain is speed. A smaller martech stack means fewer handovers, fewer integrations and fewer people who have to agree before a campaign goes live. I have never seen a team that could do less after the clear-out. I have seen several that were faster.
So my measure is not the number of tools. It is whether I can say, for every single one, the sentence that explains why it is there. With a stack that has grown over years, that rarely works on the first attempt. Which is exactly the reason to walk through it once a year.