Sessions completed per week is proposed as a tester productivity target — how do you respond?

### Why the proposal is tempting Sessions are the first thing exploratory testing has ever offered that looks countable. A lead who has spent two years unable to answer "how much testing did we do?" now has a unit, and the pull toward turning that unit into a target is immediate. The reasoning even looks sound: sessions are roughly uniform by construction, so more sessions should mean more testing. ### Why it collapses The uniformity is a convention, not a property of the world, and the person producing the number is the person being measured by it. Every lever that raises a session count is cheaper than testing well: - **Split the mission.** One coherent charter becomes three narrow ones, and the week's count goes from four to twelve without a minute of extra testing. Charter granularity is a judgement call, so this is undetectable from the outside. - **Choose the easy area.** Familiar, well-seeded, quick-to-set-up areas yield clean sessions. The nasty area with no test data — the one carrying the risk — costs a session and a half and yields a messy sheet, so it drifts to the bottom of the list. - **Skip the expensive parts.** Setup, careful notes and the debrief all consume the block. Under a count target, the block shrinks toward pure execution and the reporting quality dies, which removes exactly the artefacts that would reveal the problem. - **Stop pairing.** Two testers on one area produce one session; separately they produce two. The count rewards splitting up precisely where paired work is most valuable. - **Corrupt the time split.** The estimated setup / design / investigation breakdown only works because nobody is judged on it. Attach consequences and every sheet converges on a socially safe distribution, and the one number that reliably exposed a broken environment stops working. This is the ordinary pattern of a measure becoming a target and ceasing to be a good measure. It bites unusually hard here because the measurement instrument is a self-report from the measured party, and because testing has no accepted individual productivity metric at all — the evidence base for any of the candidates is thin and genuinely contested, which is worth saying out loud rather than replacing one confident number with another. ### What to offer instead Separate the two questions hiding in the proposal. *Capacity.* Sessions answer this well, in aggregate and unattributed. "This team sustains about eleven to fourteen sessions a week" is a real forecasting input for how much chartered testing a release can get, and it is safe because no individual is graded on it. *Effectiveness.* Report it as areas and obstacles rather than volume: which areas have had attention, which have had none, what is currently making sessions expensive. A team whose session count fell from fourteen to eight because the payment sandbox was down for a day and a half has not become less productive, and the obstacle list says so. *Individual performance.* Judge it where a lead has direct evidence: the debrief. Does this tester's reasoning hold up when questioned, are their notes usable by someone else six weeks later, do they take the ugly areas or avoid them, do their charters improve after a finding? That is qualitative, it does not aggregate into a dashboard, and it is the honest answer. ### Holding the line in practice Agreeing to publish the count as a team-level capacity figure, with the areas grid attached and a stated rule that it is never broken down per person, usually satisfies the real need behind the request. Where the request survives that, it is worth naming the trade explicitly to the person asking: you can have a number that goes up, or you can have session sheets you trust, and within about two quarters you will not have both. A lead who insists after hearing that has made an informed choice, and the useful move then is to keep an unmeasured signal — sampled debriefs, or simply reading sheets — so you can tell how far the reported number has drifted from the work.