How to make the business case for paying down tech debt
· 5 min read
“We need to pay down tech debt” is the least fundable sentence in engineering. It names no problem a non-engineer can see, puts no price on it, and asks for an open-ended amount of time. The work may be exactly right. The pitch guarantees it loses to a feature with a customer's name on it.
The fix is to stop talking about debt and start talking about what it costs.
Step 1: find the cost in one of three currencies
- Cost saved: engineer hours lost to slow builds, flaky tests, manual releases or repeated incidents.
- Revenue gained: features that are blocked or slowed, deals lost to missing capabilities, customers who churn over reliability.
- Risk reduced: the expected cost of an outage, a breach or a failed audit — probability times impact.
Pick the one you can measure. One solid number beats three vague ones.
Step 2: do the arithmetic
Suppose the deploy pipeline fails one release in five. With 250 releases a year, that is 50 failures. Each costs about 20 engineer-hours to diagnose, fix and rerun: 1,000 hours a year. At a fully loaded $100 an hour, the pipeline costs roughly $100,000 a year, every year.
Step 3: price the fix and show the payback
Two engineers for six weeks is about 480 hours, or $48,000. Payback is under six months, and the saving continues after that. Stated that way, the project is a clear win, where “tech debt” was a matter of opinion.
Step 4: say where the saved time goes
Finance will notice that salaries are paid either way, so hours saved are not cash saved. Answer before they ask: the 1,000 hours go to the integration work that sales has been waiting on. A saving only counts when the freed time is redeployed to something of value.
Step 5: name the risk and the early signal
Every proposal has a way of going wrong. Say what it is and how you will know: “The main risk is that the migration overruns. We will know by week three, and we can stop there having still removed the worst failure mode.” This raises confidence, because it shows you have thought past the happy path.
Step 6: price doing nothing
The default alternative is to carry on as now, and it is never free. Put its cost on the page: another $100,000 next year, rising as the team grows. Often that single line is the strongest argument in the document.
The one-page version
Problem, in one sentence a non-engineer understands. Annual cost, with the arithmetic. Proposed fix and its price. Payback period. Main risk and early signal. What you are asking for. If it does not fit on one page, the case is not yet clear enough.
