The problem this solves
An assessment company reaches a point where buyers stop asking whether the product works and start asking for the evidence that it does. Enterprise procurement wants validation documentation. Investors want to know whether the psychometric claims in the deck would survive an expert reading. Regulators, and under the EU AI Act increasingly the law itself, want fairness evidence that was generated rather than asserted.
Most companies at that stage do not need a full time head of measurement, and could not recruit one quickly if they did. What they need is someone who can say what evidence is missing, design the studies that produce it, and put a name to the resulting claims.
What the arrangement covers
Reviewing the validity evidence you already hold and stating plainly where it will not bear weight. Designing and overseeing validation, invariance and adverse impact studies. Drafting the technical documentation that procurement asks for. Reviewing model based scoring for fairness properties before a client does it for you. Acting as named scientific advisor where that is useful. A retainer runs at one of two levels. The lighter one is a standing scientific advisor with a small monthly allocation of advisory hours and a periodic written note on the state of your validity evidence. The heavier one is closer to a fractional head of measurement, with hands-on review of instruments, differential item functioning and invariance, and sign-off on specific validity claims, supported by a standing validity dossier. Both run on a three month minimum with thirty days’ notice either side, and rates are sent on request.
Where it does not fit
This is advisory rather than implementation. Building the instrument, running the fieldwork and shipping the product remain yours. An advisory relationship also rules out expert witness work against the same company, in either direction, which is why the conflict register is checked before either one starts.