Delivery Analysis
Methodology.
What we measure when a manager asks us to look at their own tape, what we deliberately do not infer, and the terms that govern both.
Four steps
The individual commissions it
Analysis begins with a request from the person who will be analysed, covering footage of their own sessions. There is no standing collection, no fixture monitoring and no back catalogue. If we have not been asked, no analysis exists.
Scope and consent are set in writing
Which sessions, what is measured, who receives the report, how long the footage is kept. Consent is specific to that scope and withdrawable at any point, and withdrawal triggers deletion of the source footage and the analysis rather than archival.
Delivery features are measured
Speaking pace and where it shifts, pause structure and length before key answers, emphasis placement, filler and self-correction rate, hedging versus commitment phrasing, and answer length under pressure. Each is expressed against that individual's own prior sessions, not against anyone else.
The report goes to them
One recipient: the person who commissioned it. Not the club, the board, a recruitment function, a data partner or a third party. We do not retain a usable copy beyond the retention period they set.
What the report never contains
The distinction that matters is between reporting what a delivery did and asserting what a person felt. The first is measurable and contestable by the person it describes. The second is an inference about someone’s mind, and the scientific basis for drawing it from facial or vocal signal is weak enough that the EU legislature said so in terms when it wrote the AI Act.