Research · our own product sprint

Same rubric, same answer —
whoever ran it.

Product research is where opinion quietly replaces evidence. Two stages, hard gates, a written scoring rubric, and a decision per candidate that does not move with anyone’s mood on the day.

Our own product sprint, scored against a rubric written before it started. It ran hard for one sprint and has been idle since — which is said here rather than hidden.

How it works Book a 15-minute call Your numbers, on the call
Hard gates before any scoring A wide candidate list meets hard gates, drawn as a refusal: fail one and the candidate is out regardless of how promising it looks. Only survivors reach stage two, where eleven weighted criteria score them to a maximum of 110. STAGE ONE — CAST WIDEEvery candidate foundHARD GATES — FAIL ONE, OUT REGARDLESSSurvivors onlyStage two — score11 weighted criteria, max 110
Gates reject outright. Only survivors are worth scoring.

Every figure here is a build fact — counted, not claimed. No outcome numbers appear on this site until a client’s system produces one.

  • 11weighted criteria
  • 110maximum raw score
  • 2stages
  • 4decision bands
  • 80 / 70 / 60the thresholds, fixed in advance
  • 0criteria added after the scoring began
  • samescore for the same input, any month
  • 1sprint — and idle since

A rubric everybody has in their head, written down instead.

Everyone doing product research has a rubric in their head, and it changes. The candidate you found yourself scores higher than the one somebody else suggested. Without the criteria and their weights written down first, the research becomes a way of confirming a decision that was already made.

So stage one casts wide and applies hard gates — a candidate that fails one is out regardless of how promising it looks. Stage two scores the survivors against eleven weighted criteria, computes true net after payment fees and sales tax, and derives units-to-target. Thresholds are fixed in advance, not fitted afterwards.

How it works

Two stages, and a gate between them.

The gate rejects outright rather than deducting points, which is the difference between a filter and a preference.

  1. 01

    Cast wide stage one

    A broad candidate list gathered before anything is judged. Narrowing early is how a favourite survives to the end of a process that was supposed to test it.

  2. 02

    Hard gates fail one, out

    A candidate that fails a gate is removed regardless of how well it scores elsewhere. A gate that deducts points is not a gate — it is a preference wearing a gate’s clothes.

  3. 03

    Score the survivors 11 criteria, max 110

    Weighted criteria applied to everything that got through, in the same order, with the same weights, every time.

  4. 04

    True net after fees and tax

    Payment fees and sales tax are computed rather than assumed, and units-to-target falls out of the arithmetic instead of being chosen to look reachable.

  5. 05

    Decide fixed thresholds

    Four bands set before the sprint began. The band a candidate lands in is arithmetic, not a conversation.

The part that matters

The scorer is deterministic: given the same inputs it returns the same number today and next month, run by anyone. That is the product — not the score itself, but the fact that the score is not negotiable.

Two stages with hard gates between A wide candidate list is narrowed by hard gates, which reject outright rather than deduct points. Survivors are scored against eleven weighted criteria to a maximum of 110, then sorted into fixed thresholds that were set before the research began. Stage one — cast wideHard gates — fail one, out regardlessSurvivors onlyStage two — score11 weighted criteria, max 110Thresholds fixed in advance80+ strong finalist70+ shortlist60+ borderlineunder 60 reject
Gates reject outright. Only survivors reach the scorer.

What you get

What a written rubric is actually worth.

Repeatable

The same answer, run by anyone

Two people scoring the same candidate reach the same number. Disagreement moves to the weights, where it can be argued about openly, instead of hiding inside a judgement call.

Cheap to redo

Re-run it next quarter

Because the criteria are written and the scorer is code, running the whole thing again costs almost nothing — which is what makes it a process rather than a one-off exercise.

Hard to fool

Thresholds set before the data

Fixing the bands in advance removes the most common failure in research: choosing the line after seeing where your favourite landed.

The same discipline is what a client is buying elsewhere on this site: a rule written down beats a judgement made in the moment.

Want this one walked through against your own numbers? Email info@adapton.io and we will show you the parts that transfer and the parts that do not.

The limits

Status, said plainly.

It would be easy to present this as a system running today. It is not one.

  • It ran hard for one sprint and has been idle since. It is a completed piece of work, not a live system.
  • It scores candidates. It does not source them, buy them, or run the business that follows.
  • The weights are a judgement — a written, arguable, reusable one, which is the whole point, but a judgement all the same.
  • A deterministic score is not a correct score. It is a score that cannot quietly change to suit the person reading it.
Deterministic, and idle The same inputs produce the same score in any month, and the thresholds were fixed before the sprint rather than fitted afterwards. Below the refusal line, the honest status: idle since the sprint, a completed piece of work rather than a running system. DETERMINISTIC BY DESIGNSame inputssame score, today and next monthThresholds fixed first80 · 70 · 60, written before the dataNOT FITTED AFTERWARDSIdle since the sprintfinished work, not a running system
The last box is the one most case studies leave out.

Next step

Show us the thing you do every morning.

Fifteen minutes. We ask three questions and do the arithmetic with your real numbers. If it is not worth automating we will say so on the call.

Send us the details

All case studies · info@adapton.io