Martech stack audit: decide what to keep, fix or retire
Quick answer
A martech stack audit should tell you which tools earn their keep, which need fixing and which you can retire. Follow the work through the systems, add up the full costs and check what would break if you switched one off.
Begin with workflows rather than vendor logos
An inventory is necessary, but it is not an audit. A list of subscriptions tells you what the business pays for; it does not explain how a campaign moves from idea to customer or where the process breaks.
Map a few important workflows: publishing, campaign launch, enquiry routing, audience management and reporting. For each, identify the applications, manual transfers, owners and points where work waits.
The 2025 chiefmartec landscape catalogued 15,384 solutions across 49 categories. That dated market snapshot illustrates the breadth of choice. It is not evidence that a business needs more tools, nor should it be mistaken for a current count of your viable suppliers.
Measure use in context
Low login frequency does not automatically mean low value. A tool used once a month to reconcile revenue may be essential. High login frequency may reveal a difficult interface rather than strong adoption.
Ask what completed work the tool enables, who relies on the output and how the work would happen without it. Inspect the last real examples. A licence owner who cannot identify a recent useful output may have a different problem from one who uses the tool rarely but critically.
Separate paid seats, active users and authorised users. Teams sometimes retain broad access because nobody owns offboarding or because a shared process has no documented substitute.
| Decision | Evidence | Next step |
|---|---|---|
| Keep | Clear value and accountable owner | Monitor outcomes and renewal terms |
| Fix | Necessary capability; weak implementation | Assign a bounded repair |
| Retire | No defensible value or unnecessary duplication | Test export and dependencies |
| Investigate | Insufficient evidence | Set a deadline and required proof |
Count the operating cost
Include subscriptions, usage fees, implementation, connectors, external support and internal maintenance. Add the time required to reconcile inconsistent definitions between tools. A cheap application with fragile integrations can be expensive to operate.
Our SaaS buying guide explains why charging units matter. Contacts, records, events, seats and AI credits can grow differently. Model a realistic operating scenario rather than comparing entry-level prices.
If the platform costs £12,000 a year, the integrations need £6,000 of support and your team spends £8,000 worth of time administering it, you’re paying £26,000 before any migration costs. That’s the number a replacement needs to improve on. A cheaper licence can still leave you with a bigger bill.
Inspect dependencies before retiring a tool
Document inbound data, outbound data, scheduled jobs, embedded forms, audience synchronisations and any contractual retention requirements. A tool that appears unused may still supply a field used by a critical report.
Export a sample and test whether the replacement can use it. Check identifiers, timestamps, attachments and relationships. An export button does not guarantee a usable migration.
For connected workflows, our low-code operations guide covers exception handling and retries. The payment-workflow guide shows why event states and duplicate handling matter when marketing depends on operational systems.
Make three kinds of decision
Keep a tool when its value and ownership are clear. Fix it when the capability is needed but the implementation, adoption or data flow is weak. Retire it when the workflow is unnecessary, duplicated or better served elsewhere and the exit has been tested.
Give every decision a date and owner. “Review later” without a trigger becomes another year of renewal. For a fix, specify the evidence that would justify keeping the tool at the next review.
Do not consolidate solely to reduce the number of logos. A broader suite can simplify contracts while increasing dependence on one supplier. Conversely, a carefully connected specialist tool can be worthwhile if it solves a critical problem well.
MarTech Logic’s view: audit the handoffs
The most revealing part of a stack review is often the movement between systems. That is where identifiers disappear, preferences become stale and staff create unofficial spreadsheets to keep work moving.
Ask each team where it compensates for the technology. The answer may justify training, a better integration or a simpler process rather than a new purchase.
End the audit with a short decision register and a dependency map. Those are more useful than a maturity score nobody can act on. When AI features are part of the renewal pitch, use cost per accepted result to test their contribution separately from the platform’s existing value.
Related reading
AI marketing ROI: measure the cost of work you can actually use · CDP vs CRM vs data warehouse: choose around the customer task · Marketing budget allocation: fund the next useful decision