The first time I ran a serious tech stack audit for a multi-unit restaurant group, I expected to find $15,000 to $20,000 in annual savings. I found $84,000. Two tools solving the same problem. Three tools nobody had opened in six months. One vendor with a 22 percent renewal escalator that had compounded across three years unnoticed. A "critical" analytics platform that was still being paid for six months after the person who set it up had left.
None of that was anyone's fault. It was the natural drift of an operating stack that had never been audited. This is the discipline that stops the drift.
Why annual, and why a day
You run the audit annually because the drift is annual. Tools get adopted for a specific initiative. The initiative ends. The tool stays. Vendor contracts auto-renew with escalators. Integrations break when a POS updates and nobody notices until reporting looks wrong. A year is about the timeframe over which the cost of that drift becomes meaningful money.
You run it in a day because it fails otherwise. A two-week audit turns into a two-month audit turns into a "we'll do it next quarter" audit. A one-day audit ends. It produces a decision list. The decision list gets acted on the next week.
The day-one output is not a strategy. It is three lists. Kill. Consolidate. Keep. Anything past those three lists is decoration.
Fig. 1 · The three-question test.
Step one: pull every recurring charge from the last 12 months
Start with the money, because the money never lies. Pull every credit card statement, every accounts payable ledger entry, every auto-debit from operating accounts for the last 12 months. Filter for recurring charges. Categorize by tool.
You will find things you did not know you were paying for. A domain renewal from three years ago on a subdomain that was never launched. A design tool one manager bought and never canceled. A per-location fee for a service that only ever ran in two units. A subscription that increased 40 percent at last renewal because a "premium" flag got flipped by the vendor.
Do not skip this step even if your finance team says they know the recurring spend. They know what they see in the accounting system. They do not know what got charged to a manager's card and expensed as "office supplies." A serious audit finds the shadow spend, and the shadow spend is usually 10 to 20 percent of the total.
Step two: score each tool on three questions
Every tool gets three questions asked about it, in this order.
Is anyone actually using it?
Verified by usage data, not by the department head who assures you it is still important. Pull login history. Check active seats. Look at when the last file was modified, when the last report was run, when the last user session happened. If nobody has opened the tool in 90 days, the answer to question one is no.
Every unaudited stack has at least two tools that fail this question. Usually they are tools that were bought for a specific initiative (a menu redesign, a compliance audit, a rollout project) and stayed on the credit card after the initiative ended.
Is it integrated with the rest of the stack?
A tool that produces data nobody else can consume is not worth what you pay for it. If the tool is standalone, if you have to log in specifically to see the data, if the data cannot flow into your dashboards or your operating rhythm, the answer to question two is no.
Test the integrations you assume are running. This is where the biggest silent waste hides. Send a test transaction end to end. Watch the record move through the systems. In the last two audits I ran, over 40 percent of integrations that were "definitely working" had broken silently. Nobody had noticed because nobody was checking the downstream data hard enough to catch it.
Would we buy this tool again today?
Knowing the current price. Knowing the current alternatives. Knowing what the workflow actually looks like now, not what it was when the tool was adopted. This is the honest question, and it is the one that surfaces the tools that were right two years ago and are wrong now.
Legacy vendor contracts fail this question often. So do enterprise deals signed at scale that have not been renegotiated as the business changed. So do tools that have been overtaken by cheaper, better competitors.
Two nos means kill. Three nos means kill yesterday. One no means look harder, but the tool is on notice.
Step three: test every integration you assume works
This is the step most audits skip and it is the step that saves the most money and the most trust in the dashboards.
Make a list of every integration in the stack. POS to accounting. POS to labor. Labor to payroll. Accounting to reporting. Reservation system to CRM. Every one of them. For each integration, run a test transaction end to end and verify it lands correctly in the downstream system.
Fig. 2 · The integration map, with silent failures marked.
You will find broken links. Silent ones. In a recent audit I ran, the labor-to-payroll integration had been misfiring by 0.4 percent for eight months because a new pay code had not been mapped. Nobody caught it because the total looked right. The gross was quietly wrong across every payroll, across every unit, for two-thirds of a year. Finding that one cost less than an hour and paid for the audit ten times over.
Step four: build the kill, consolidate, and keep lists
At the end of the day, every tool sits on one of three lists.
Kill. Fails multiple questions. Cancel at the next contract exit or immediately if month-to-month. Communicate the change to the small number of people who used it and provide the replacement path.
Consolidate. Overlaps materially with another tool the operation already pays for. Migrate off the redundant one. This is where the biggest savings usually live.
Keep. Passes all three questions. Note the renewal date and any escalator language on the contract. Set a calendar reminder 60 days before renewal to negotiate.
Any tool that survives the audit gets an owner named on it. The owner is responsible for the tool's continued value, for using it, and for surfacing it in next year's audit with usage data ready.
The tools that always survive, and the ones that rarely do
Across the audits I have run, some patterns hold.
Almost always survive: The POS (Toast, Square, Aloha). The scheduling tool once it has cleared its own selection audit. The accounting system (QuickBooks Enterprise, NetSuite, Sage). One or two dashboard tools (Power BI, Looker Studio). The reservation and guest tool if it is used (Resy, OpenTable). One communication tool for the field (Slack, Teams, or MS Teams).
Frequently killed or consolidated: Standalone analytics tools that duplicate what the POS or Power BI already shows. Legacy inventory tools that are not integrated with the POS. Marketing tools bought for a specific campaign. Compliance tracking tools that were bought before the current record system existed. Multiple project management tools running in parallel across departments.
Sacred cows worth touching: Any tool whose defense is "the CFO likes it" or "we have always used it." That is not a business case. Ask the three questions anyway.
What I got wrong the first time
Two mistakes worth sharing.
I let IT run the first audit. IT protects tools that generate tickets. IT also does not see the accounts payable side, so shadow spend never surfaced. The audit found 40 percent less waste than it should have. The next year the head of operations ran it, with IT supporting on the integration testing. That audit found the real number.
I ran the audit without a decision date. The output list sat around for six weeks. Half the tools that should have been killed made it to their next auto-renewal because nobody had authority to act on the list. Now I set the decision meeting for two business days after the audit ends. Every tool on the kill list gets canceled that week or has a written case for staying.
The point
A tech stack audit is not a strategic exercise. It is an operating hygiene practice. Once a year, one day, three questions per tool. The output is a spreadsheet with three columns and a set of decisions.
The savings are real, usually 15 to 30 percent of the SaaS budget. The bigger value is trust: after the audit, the operator knows exactly what is running in the stack, what it costs, and what it delivers. That knowledge is worth more than the money you save. It changes how the next tool decision gets made.
Pick the date. Put it on the calendar. Run the audit. Then do it again in twelve months.