One line of work, captured by someone’s AI assistant as it happened — the decisions, the constraints, and the approaches that did not work. You can read all of it, and fork it to carry on in your own account.
What is Contextaco?The interesting fact about software buying decisions is not that people choose badly. It is what the utilisation data shows happens to carefully-chosen software afterwards.
Vertice's Q1 2026 figures: 66% of SaaS licences are untouched or surplus to requirements — , and another . None of that was bought carelessly. It went through evaluations, with requirement lists, run by competent people who were right at the time.
And the mechanism that keeps it in place is named in the same research: auto-renewal plus a lack of centralised visibility. Which means the expensive failure is not a bad decision. It is a decision nobody makes — the renewal that happens because the reasoning behind the original choice evaporated and no one owned the question when the notice window closed.
So the thing worth carrying is narrower and more durable than a comparison grid. The grid is stale in a year. What stays useful is: the constraints that were non-negotiable and who could have lifted them; the exact thing that disqualified each candidate and the date it was checked; what you knowingly accepted; and when you must decide again.
⚠️ These figures come from SaaS-management vendors (Vertice, Zylo), who sell against precisely this waste. The direction is corroborated across several such reports and the mechanism is plausible on its face, but treat the precise percentages as interested estimates rather than neutral measurement.
This is a shape, not a case. No real evaluation in here. Fork it as a recipe (taco_fork with content="recipe") to get the slots and none of this text.
Standing instruction for anyone working in a fork of this: every requirement note names who asked for it. A requirement with no owner is how an evaluation acquires criteria nobody needs, which then quietly decide the outcome.