The most expensive vendor relationship is the one you can't measure

I sat with a portfolio CEO last month and watched her almost renew a vendor contract that should have been killed two quarters ago.

The contract had auto-renewed twice. The vendor’s product had been silently degrading. The team using it had built three workarounds to keep it functional. And the renewal was about to go through again, not because anyone had decided the vendor was worth keeping, but because nobody had decided anything at all.

The reason it kept surviving wasn’t economics. It was inertia with an accounting problem underneath it: switching costs were known and concentrated, a migration project with a number attached. Status-quo costs were diffuse, spread across engineering hours nobody was counting. Known number beats invisible number every time, even when the invisible number is bigger.

Thirty minutes to make it visible

We did a short exercise before the renewal went through. Listed every workaround the team had built around this vendor. Mapped each one to engineering hours per month. Mapped those hours to dollars.

The cost of the workarounds was 2.3x the cost of switching.

Nobody had ever connected those dots, and the reason is worth understanding because it’s structural, not careless. The workarounds had been built one at a time over eighteen months by different people, each solving what looked like a small local problem. No single decision was wrong. The sum of them was a second, shadow contract with the same vendor, paid monthly in engineering time, that never appeared on any invoice.

The debt that doesn’t live in the backlog

This is the dangerous category of technical debt. Not the kind that gets logged in a backlog and argued about in planning. The kind that lives invisibly inside your operating expenses, getting paid every month in hours you can’t see. A backlog item at least has a name. This has camouflage: every workaround looks like the team being resourceful, right up until you add them up.

It’s the same blind spot I wrote about in technology as your largest unwatched expense, one level deeper. There, the problem is that nobody reviews the line items. Here, the problem is that the real line item doesn’t exist anywhere a CFO could review it. The invoice says one number. The relationship costs another.

And the longer it runs, the more the exit options decay. Eventually somebody proposes fixing it all at once, under deadline pressure, which is how you end up with the rushed migration that costs more than the vendor ever did.

The quarterly question

Once a quarter, force this question with your portfolio companies, or your own team: what’s the most expensive vendor relationship you can’t measure? Then go measure it. The exercise is the same one we ran: workarounds, hours, dollars, compared against the switching cost everyone already knows.

Sometimes the answer is “keep the vendor, kill the workarounds.” Sometimes it’s “the switch pays for itself in a quarter.” Either way you’re deciding with real numbers, which beats auto-renewal by default.

Most portfolio CTOs I work with are surprised by what this surfaces. Worth being honest about which contracts you’re renewing on purpose and which ones are renewing themselves.