Estimation and Team Stability: Sizing Work for a Team That Keeps Changing Shape

The material here comes from an organisational change engagement inside an international research and advisory group, in one of its largest business units. Nine teams spanned product and technology, generating hundreds of millions of dollars in sales a year, with leadership aiming to double that figure. We built the evidence base by scoring twenty-five dimensions of the organisation and adding a qualitative read for whatever the numbers alone could not show.

The analysis started with the map.

The map opens centred on team-level planning and estimation: whether the teams can size their work. Its ties run broad and moderate rather than deep, and the strongest single pull isn’t a planning dimension at all. It’s release processes. What’s missing from the neighbourhood matters as much as what’s in it: no estimation-technique dimension, no tooling dimension. Whatever was going wrong with sizing here, the map says it wasn’t happening in the planning room.

On the scores alone, this page wouldn’t exist. A median of 8 out of 10; of sixty-one respondents, only seven scored it at 4 or below. Most assessments would file estimation under “working” and move on. Then you read what the same people wrote.

Barely touched by the team survey’s coded stream: two observations, both negative. Widen to every coded mention, team survey and leadership interviews together, and the count is eleven: ten negative, one positive, against a scored median of 8. That gap is the finding, and most of the sore units come from the leaders, the people whose seat looks across teams. Read in sequence, they name three absences, none of them a technique:

“Teams are not always independent units, because ppl are always being borrowed. This makes it tough to estimate.”

“There is no defined framework to say what the team capacity is, sizing the work, to ensure that the teams are able to do the work which has been assigned to them.”

“There isn’t a review or sizing workflow.”

“Time and resource constraints mean tickets are typically sized by just a few people, versus the whole team.”

No stable team to estimate for, no agreed picture of capacity, no workflow with a slot for sizing in it. You can’t size work for a team that keeps changing shape, and the units knew it, even as the scores said 8.

Eight of the nine teams hold medians of 7 or above, and one sits at 3, against a best of 9.5. That shape fits the borrowing story: estimation feels fine from inside a team whose membership holds, and impossible from inside the team the people are borrowed out of. The average hides exactly the team the words describe.

The counter-example was already inside the same dataset. One group had stopped trying to estimate better and stabilised the other side of the equation instead:

“We pushed the program managers to give us a yearly estimate of capacity for every sprint throughout the year… Now we have identified when we want to have code freezes, and adjusted capacity allocation during those times.”

“We established predictive capacity planning. Sprint by Sprint capacity planning.”

They didn’t sharpen the estimates; they stabilised the denominator, and planning held through their high season. Here is the check that travels. Take three recent sprints and list the people who worked in them but weren’t planned in. If that list isn’t empty, your estimation problem is a team-stability problem, and no pointing technique will fix it.

A real engagement; an unnamed client, and it stays that way. Team names and identifying details are removed, and quotes are lightly edited for anonymity. Every chart here draws its numbers directly off the underlying assessment data as the page renders, out of a set of about 2,000 qualitative and quantitative data points, never simply asserted. One organisation, correlations not causes.