Tech debt cost: the one-page briefing for your CFO
The brief that lands. Annual wasted payroll on the line, compounding interest behind it, a specific investment with a 5 to 7 month payback in front of it. Engineering language stays out of the room.
The 90-Second Answer
A 25-engineer team at $160,000 fully-loaded cost with a 30% debt drag is burning $1.2 million per year on debt-related work. Untouched, it compounds at 15-18% per year (CAST CRASH benchmark). A targeted 12-week remediation typically returns 80-400% inside 12 months with break-even in 5-7 months (McKinsey 2023). That is the entire pitch.
Slide 01 / The Cost of Inaction
Start with payroll, not code
The CFO controls one of the largest opex lines in the business: engineering payroll. In a series-B SaaS company, fully-loaded engineering payroll is typically 35-55% of total opex. A CFO who hears “we have a refactoring problem” is being asked to fund work outside the budget; a CFO who hears “28% of the engineering payroll line is being wasted on rework that compounds” is being shown a leak inside the budget. The second framing converts. The first does not.
The McKinsey 2023 study put the drag at 25-42% of engineering capacity, with a midpoint near 30%. The Stripe Developer Coefficient research (2018, PDF) measured 33% lost to maintenance and bad code, equivalent to 17.3 hours per engineer per week. The CAST CRASH 2024 benchmark documents 15-25% annual compounding on untouched debt. These three numbers are enough to construct the cost-of-inaction figure without a bespoke measurement framework. Anchor to the conservative end (25% drag, 15% compound), itemise the payroll line, multiply.
The arithmetic the CFO actually needs is dollar-denominated and recurring. A 25-engineer team at a $160,000 fully-loaded cost is $4 million per year in payroll. Apply a 30% drag and the wasted portion is $1.2 million per year. Carry forward at a 15-18% annual compound for five years and the cumulative waste is $7.5-8.5 million if nothing is done. The recurring opex framing matters more than the headline figure: this is not a one-off, it is structural, and it is on the books today.
Slide 02 / Translation Layer
What engineers say, what wins CFO support
Every engineering metric needs a financial translation before it enters the room. The translation is not optional, and the engineer who runs the pitch should be the person who has practised the translation, not the senior engineer who has the deepest grasp of the codebase. A junior engineer who can recite the dollar figures is more persuasive than a principal engineer who keeps slipping back into “module coupling.”
The CFO is not hostile to engineering; the CFO is hostile to ambiguous opex. Each row converts an ambiguous fact into an unambiguous dollar amount.
Slide 03 / The Ask
A specific investment with a payback period
A CFO funds initiatives that look like investments, not initiatives that look like maintenance. The difference is whether the request has a payback period attached. “Spend 20% of engineering capacity on debt reduction” is a maintenance request. “Spend 12 engineer-weeks across Q3 on the auth module, projected annual saving of $340K with payback at month 7” is an investment request. Same work, different framing, completely different decision tree.
The McKinsey 2023 dataset gives the upper-bound payback expectation: median ROI of 80 to 400% inside the first 12 months, with break-even at 5 to 7 months for targeted initiatives. This is the figure to lean on in the room. A finance team that has spent the year evaluating SaaS vendor purchases at 3-year payback will look at a 5-7 month payback and consider it a high-confidence allocation, even if the engineering team is internally less certain. The published research carries the room.
The investment ask should always be scoped to a single module, system, or class of work. “The whole codebase” is not an investment, it is an open-ended commitment with no scope boundary. “The notifications service, the auth module, and the billing job runner, in that order, over the next two quarters” is a portfolio of three investments the CFO can evaluate independently and approve or defer each. Scope creates approvability.
Slide 04 / Objection Handling
The five objections a CFO will raise
CFOs reject most ad-hoc investment asks on the same five grounds. Pre-empt every one. The pitch fails when the engineering leader treats objection-handling as adversarial rather than expected. A finance leader who pushes back is doing the job; an engineering leader who is annoyed by the pushback is misreading the room.
“How do we know the 30% drag figure is true for us?” Anchor to a published benchmark, not internal measurement. McKinsey 25-42%, Stripe 33%, DORA elite-vs-low gap, all defensible against finance scrutiny. Internal measurement is a follow-up commitment, not a prerequisite.
“Can we defer it a quarter?” Yes, and the cost of deferring is the compounding figure. A $1.2M annual drag at 18% compound becomes $1.41M next year. The deferral is itself a decision with a cost.
“Why does the engineering team need leadership approval at all? Just allocate.” Because the work displaces features the product team has committed to. The brief is the formal mechanism to renegotiate the roadmap with finance and product in the room together.
“What is the consequence if I say no?” The 5-year cumulative cost-of-inaction figure, named, on a slide. Compounding to $7.5M+ on the working example, plus the second-order consequences (attrition, missed roadmap, incident escalation).
“Why now versus next year?” Because the longer the deferral, the deeper the workarounds, the harder the eventual remediation. Cost-of-cleanup grows faster than the cost-of-prevention curve documented in Boehm's defect-cost research.
Slide 05 / The Close
Convert to a budget motion, not a discussion
The brief ends with a budget motion: a specific dollar amount, a specific scope, a specific timeline, and a named owner. “Approve $480,000 of engineering capacity in Q3-Q4 against the notifications-service refactor, owned by the platform lead, with the projected $340K annual recurring saving reviewed at end of Q1 next year.” That is a motion finance can approve or defer in one sentence, with a measurable review point. Anything vaguer than this ends in a deferral.
The review point matters as much as the ask. CFOs are conditioned to expect engineering initiatives to slip both scope and payback. A measurable review point at month 6 or 12 is the credibility deposit that earns the next debt-reduction approval. Skipping the review point makes every subsequent pitch harder, even if the current initiative succeeded.
The companion pages on this site address adjacent decisions: how to brief the CEO (strategic framing, not financial), how to brief the board (fiduciary framing), and whether the work qualifies as capex under IRC §174 (it usually does not, since 2022 changed the rules). The COGS impact page covers when engineering payroll capitalizes into cost-of-revenue, which changes the CFO conversation entirely. For the engineering-deep version of the same arithmetic, see the sister site technicaldebtcost.com.
Field Notes
Frequently asked questions
What is the single number a CFO needs to see?+
Annual wasted payroll. Team size times fully-loaded cost per engineer times debt drag percentage. For a 25-engineer team at $160K fully-loaded with 30% drag, that is $1.2M per year, every year, recurring.
Why does engineering language fail with CFOs?+
Refactoring is a verb the CFO has heard 200 times without a return number attached. The conversation does not land until the request becomes a line item with a quantified cost of inaction and a payback period in months.
What payback period will a CFO accept for a tech debt initiative?+
Most CFOs in mid-stage SaaS will fund initiatives with payback inside 12 months, and many will fund inside 18 months when the cost-of-inaction is itemised. McKinsey 2023 reports median payback of 5-7 months for targeted debt-reduction work, which sits comfortably inside that window.
Should the pitch land in the annual budget or a quarterly reforecast?+
Annual budget for any initiative that consumes more than 10% of engineering capacity for a quarter or more. Quarterly reforecast for sub-10% allocations that can be absorbed within the existing engineering opex line without re-baseline.
How do I quantify the cost of inaction if I do not have a measurement framework yet?+
Anchor to the McKinsey 25-42% range, pick the conservative end (25%), apply to a verifiable engineering opex line. Add the 15-18% annual compounding figure from CAST CRASH benchmarks. A 12-month inaction cost is the result. This is defensible without a bespoke measurement.
What happens if the CFO declines the ask?+
Document the brief, file the request, and revisit at the next quarterly reforecast with one quarter of additional compounding interest added to the cost-of-inaction figure. The number gets larger by definition. Do not re-pitch on the same terms inside 90 days.
Adjacent Reading