← Portfolio · ← Collaboration 2.0
2.6 · Measurement, feedback and refinement
Sub-sections: 2.6.1 The four declared KPIs · 2.6.2 How the first reading was taken · 2.6.3 The gap worth naming
2.6.1 · The four declared KPIs
Four whole-organization KPIs were declared as what the intervention would be judged by. They were named before anything was disturbed, for the reason given in 2.1.3:
| KPI | What it was meant to capture |
|---|---|
| Accountability | Whether you can tell who owns a decision |
| Communication ratio | How much people talk across team lines, and how far it reaches |
| Time-efficiency | How long it takes to get from a need to a decision, and how long a clarification takes |
| Intellectual utilization | How much of the relevant expertise actually reaches the decisions it bears on |
The fourth one closes a loop opened at the start of section 1. 4.1 established that the knowledge was there. Intellectual utilization asks the other half of the question: how much of it reaches the decisions it bears on.
The diagnosis and the measurement design are two ends of the same claim. If only one of the four had ever been measured, this is the one that mattered.
What the communication ratio is a ratio of. The developers were suffering — some of them without noticing it — from not knowing each other, and from how hard it was to reach anyone. The measure is an aggregate of things the platform can count: the volume of text people write, the number of different people someone gets into a question-and-answer exchange with in the comments, and how often each person comments at all. At the start of a rollout only a few people commit text and reviews. The number is expected to climb as Collaboration 2.0 is actually used, and that climb is the KPI.
The survey (4.3) was the main instrument for the first and the fourth.
It also asked a direct time-efficiency question: on average, how long does it take to receive the clarifications you need from another team? That is the one question in the set that gives you a number you can compare before and after.
2.6.2 · How the first reading was taken
The KPIs needed a reading taken before anything was disturbed. Getting one was not straightforward. The people being asked had as their main complaint that speaking honestly could cost them their career. Any instrument that could be traced back to a respondent would have measured the political weather instead of the thing it was pointed at.
The answer was a survey handed out as individually printed sheets, each carrying a token that works once. How that instrument was built, which alternatives were rejected, and what the choice cost are in 4.3. It was answered in late 2025, after a twenty-minute explanatory video and before any part of the framework was in use.
One thing about the order of events, because it decides what this evidence is worth. The diagnosis in section 1 and the design of Collaboration 2.0 both came before the survey. They came out of years of watching the place, not out of these answers. The survey was built afterwards as the baseline measurement. A large part of what it returned matches what the diagnosis had already claimed, which is support for the diagnosis rather than its source.
The reading itself, question by question, with the charts and the caveats, is 4.3. It was meant to be run again rather than once; 2.7.1 is what repeating it was for.
2.6.3 · The gap worth naming
The first reading exists. The second one does not, and the programme stopped before it could (3.1).
So the comparison the whole measurement design turns on was never made. What is above is a baseline with nothing to hold it against.
Three of the four are also still attitudes rather than observations, because the honest source was a survey rather than the work itself.
Put the shared platform in place and the logs would give you two of them directly. Who contributed to which document. How many of those contributions crossed a team boundary. How long a raised question took to get an answer.
That would turn things people report about themselves into things you observe. What that would take, in detail, is 2.7.1.
⟵ Up: Collaboration 2.0 · Previous: 2.5 · The adoption argument · Next: 2.7 · What would have come next →