A martech stack audit takes about 30 days: build a full inventory, measure real use against the seats you pay for, find where tools overlap, work out what each tool costs per opportunity it touches, then give every tool one of four labels: keep, fix, merge or cut. Make each decision before the notice period on that contract, not on the renewal date.
The renewal invoice arrives for a tool nobody in the room can describe. Someone thinks the events team bought it. Someone else thinks it was the agency. It renews anyway, because cancelling feels riskier than paying.
Multiply that by every tool in the stack and you have the most expensive habit in modern marketing: buying capability and never using it.
This guide is the method we use to get from a list of tools to a decision on every one within 30 days. It works for a stack of 15 tools or 150.
The numbers behind the habit
In August 2023, Gartner reported that marketing leaders were using just 33% of their martech stack’s capabilities, down from 42% in 2022 and 58% in 2020. Two years of decline in a row, from a survey of 405 marketing leaders. The stack kept growing; the use of it kept shrinking.
How much of the martech stack is in use
Share of martech capabilities marketing leaders say they use
The same pattern shows up across all software. Zylo, which tracks more than 40 million software licences, reported in January 2025 that companies spend an average of $4,830 per employee on SaaS, up 21.9% in a year. It put the average waste on unused licences at about $21 million a year.
Those are averages across very different companies, so treat them as a direction, not a forecast for yours. The direction is clear enough.
Martech stack audit, assessment or evaluation: what is the difference?
Mostly the words. A martech stack assessment or evaluation usually scores the stack against a goal, such as readiness for AI or a move to a new CRM. An audit goes tool by tool and ends in a decision for each one.
The method below covers both: the inventory and usage steps are the assessment, and the decision table is the audit.
How do you build the inventory in a week?
The procurement list is never complete. Tools arrive on company cards, through agencies, as free plans that quietly became paid ones, and as add-ons inside larger contracts. Pull from five places in the same week:
- Finance. Twelve months of card statements, expense claims and supplier invoices. Search for software, SaaS and subscription.
- Sign-on logs. Every app people reach through Google, Microsoft or Okta sign-in.
- The CRM and marketing automation platform. Every connected app and integration, including the ones nobody remembers adding.
- Your agencies. The tools they run on your behalf, and who pays for them.
- Team leads. A five-minute form: what do you log in to every week that is not on this list?
For each tool, capture eight fields: name, owner by name, team, annual cost, seats paid for, contract end date, notice period, and the funnel stage it serves (attract, convert, close, keep or grow).
The notice period matters more than the renewal date. Many contracts ask for 30 to 90 days of notice, so a tool you decide to cut in its renewal month has often already renewed.
How do you measure real usage, not licences?
A licence tells you what you bought. Usage tells you what you got. Measure three things for each tool:
- Active users. People who logged in during the last 30 days, against the seats you pay for. Most admin consoles show this; for the rest, the sign-on logs will.
- Depth. Whether people use the features you bought the higher tier for. A marketing automation platform used only to send newsletters is a newsletter tool at an automation price.
- Output. What the tool produced last quarter: campaigns sent, forms live, reports read, records enriched.
Our rules of thumb: under 50% of seats active means right-size the contract at renewal. Under 25% puts the tool on the shortlist to fix or cut. A tool with active users but no output is a training problem, not a tool problem.
Low numbers here are normal. Gartner reported in August 2023 that marketing leaders use 33% of their martech stack’s capabilities. Across all business software, Zylo’s 2026 SaaS Management Index found organisations leave an average of 36% of their licences unused.
Where do tools overlap, and which one should win?
Sort the inventory by funnel stage and by job. Two tools on the same job is a question. Three is an answer.
The usual overlaps are predictable: two or three tools sending email; a CRM, a separate sales engagement tool and spreadsheets all holding pipeline; several analytics and reporting tools; and point tools for webinars, chat, forms and enrichment that the main platform already covers.
When two tools do the same job, the winner is usually not the better product. It is the one that:
- keeps its data in the CRM, or sends it there without manual work;
- has the most active users today, because moving people costs more than moving data;
- has a named owner who will answer for it;
- costs less per active user, once implementation and admin time are counted.
Features come after those four. A slightly weaker tool that everyone uses and that feeds the CRM beats a stronger one that sits in its own corner.
What does each tool cost per pipeline dollar?
Cost per seat is the wrong measure for a revenue engine. Use cost per outcome. For each tool, add the annual licence, the share of people’s time spent running it and any agency or integration fees. Then divide by what it touched in the same year: opportunities influenced and pipeline value, or renewals and expansion for retention tools.
An illustration, not a benchmark: a webinar platform costs $18,000 a year once you count the time to run it, and the CRM shows it touched 40 opportunities. That is $450 per opportunity touched. If paid search costs $300 per opportunity touched, the question is not whether webinars work. It is whether this one earns its premium.
Two cautions. Attribution is never perfect, so compare tools on the same job with the same method rather than chasing a precise number. And a tool that cannot show any outcome because its data never reaches the CRM fails this test by default. That alone is a fix-or-cut signal.
The keep, fix, merge or cut decision table
| Label | When it applies | What happens next |
|---|---|---|
| Keep | A clear job on the funnel, more than half the seats active, data reaching the CRM, and a cost per outcome in line with the alternatives. | Confirm the owner, right-size seats at renewal and review in a year. |
| Fix | The job is right and the cost is fair, but use is low or the data is not flowing. | Train the users, connect the data and set a 90-day usage target. If it misses, it moves to cut. |
| Merge | Another tool in the stack does the same job and wins on the four tests above. | Plan the data move and the switch-over, then give notice in time. |
| Cut | No clear job, under a quarter of seats active, or no outcome you can see. | Export the data, check what depends on it, give notice before the deadline and remove access. |
The rule behind the table: a tool earns its place by doing a job on the funnel, being used by the people who need it and putting its data where the rest of the business can see it. Miss two of the three, and it goes.
A 30-day plan
- Week 1: the inventory from all five sources, with owners and notice periods.
- Week 2: usage and output for every tool, from admin consoles and sign-on logs.
- Week 3: the overlap map and cost per outcome, tool by tool.
- Week 4: labels agreed with each owner, a renewal calendar for the year and notices sent for anything due to be cut.
Then make it a habit. Check the stack every quarter against the renewal calendar, and add one rule for new purchases: no tool without a named owner, a job on the funnel and a plan for getting its data into the CRM.
Mistakes that undo the audit
- Cutting a tool before exporting its data and checking what else depends on it.
- Counting the licence and ignoring the people time that keeps each tool running.
- Missing the notice period and paying for another year.
- Treating fix as failure. Most stacks are under-adopted rather than too big, and the money is already spent.
What it looks like when a company acts
In June 2023, Mi3 reported that Lendlease, the Australian property group, had moved its CRM and marketing cloud from Salesforce to HubSpot, its digital experience platform from Sitecore to Optimizely, and most content creation from Adobe to Canva. The earlier Salesforce implementation was reportedly more than $15 million. Insiders said the company would have faced several million more in project costs had it stayed the course.
A Lendlease spokesperson said the platform migration had delivered substantial cost savings without losing features or functionality.
The lesson is not that one vendor is better than another. It is that a stack built for every possible need ends up serving none of them well, and nobody notices until someone adds up the bill.
Cutting is the easy part. The value is in the second label, fix: most stacks are not too big; they are under-adopted. The money is already spent. The return is waiting on training, ownership and a clear job for each tool.
What is your stack really costing you? The Revenue Engine Health Check scores margin and efficiency, including how much of your stack is used every week, alongside four other areas. Five minutes, free.
Frequently asked questions
How do I audit my martech stack and decide what to cut?
Build a full inventory from finance records, sign-on logs, CRM integrations, agencies and team leads. Measure active users against seats paid for, map overlaps by funnel stage, work out each tool’s cost per opportunity touched, then label every tool keep, fix, merge or cut. A tool that misses two of three tests (a clear job on the funnel, real use, data in the CRM) goes. The whole audit takes about 30 days.
What is the difference between a martech stack audit and a martech stack assessment?
An assessment or evaluation usually scores the stack against a goal, such as AI readiness or a CRM change. An audit goes tool by tool and ends in a keep, fix, merge or cut decision for each one. In practice the first half of an audit is the assessment.
How much of their martech stack do marketing teams actually use?
About a third. Gartner reported in August 2023 that marketing leaders use 33% of their martech stack’s capabilities, down from 42% in 2022 and 58% in 2020.
What is a good martech utilisation rate?
There is no universal benchmark. We treat more than half of seats active in the last 30 days as healthy and under a quarter as a reason to fix or cut.
Should we cut under-used tools or fix them?
Fix first where the tool does a real job but nobody was trained. Most stacks are under-adopted rather than too big, so training, a named owner and a clear job for each tool usually return more than cutting.
How often should we audit our martech stack?
Run a full audit once a year and a short check every quarter against the renewal calendar, so every cut or merge decision is made before the notice period on that contract.
Who should own the martech stack audit?
One person with authority across marketing, sales and revenue operations, usually the marketing operations or RevOps lead, with finance supplying the cost data and each tool’s owner agreeing its label.
