One respondent, no evidence standard, generous anchors. The calibrated version is thirty-five questions, four interviews conducted separately, and evidence demanded for every high score.
The gap between the two is typically fifteen to thirty points — and that gap is the most useful thing either number tells you.
Three structured interviews conducted separately — engineering lead, delivery manager, CFO — plus a fourth with whoever owns pricing. Never as a group: a group produces one consensus answer where you needed three independent ones, and the divergence between them is frequently the most valuable finding.
Any score of 3 or 4 requires evidence produced during the conversation. A dashboard, a log, a configuration. Where nothing can be shown, the score is 2 or below. That single rule accounts for most of the difference between a self-assessment and a calibrated one, and it is not because anyone lies — organisations routinely believe capabilities exist that stopped working, were never finished, or are true of one team and assumed to be general.
Only the last ninety days count. Roadmap commitments score zero. Pilots not in production score zero.
Seven dimensions, four ceilings, with the evidence behind every point and the reason for any ceiling applied.
Your self-assessment against the calibrated result, question by question, with the reason for each material difference.
Five functions scored separately. Whether capability is institutional or concentrated in one team — which are different problems with different fixes.
Revenue per employee, human minutes per outcome, no-touch rate. Measured, with method and confidence stated.
Revenue by service line against exposure and pricing basis, plus what reprices in the next four quarters.
Named workflows in priority order, with effort, dependency, owner, and expected point movement for each.
An assessment produced by someone who also sells the remedy is worth very little unless its independence is explicit. So:
| The fee is never credited or discounted against any subsequent engagement. |
| The roadmap names off-the-shelf and do-nothing options where those are the right answer. |
| Any commercial relationship with a vendor named in the recommendations is disclosed in the report. |
| Anchors, weighting and ceilings are published in advance and never adjusted for an individual client. |
| You may commission a second opinion on the same published instrument, and the evidence is made available to support it. |
| The fee is fixed regardless of the score. No part of it is contingent on the result. |
Best fit: IT services and product engineering firms, 200–2,000 people. Buyer is the CEO, COO or delivery head — not the innovation team.
Trigger: shrinking deal sizes, clients asking what your AI story is, and margin pressure you cannot explain away.
Not for: organisations wanting validation of a decision already made.
Tell me roughly where you are and I will tell you honestly whether an assessment is the right next step. Sometimes it is not.