Support designed to work, and designed to end

Validity asks whether the measure is right. Efficacy asks whether the support that follows the score works. They are separate questions needing separate evidence. Cognassist answers validity on its science page: a normative sample of 26,301 UK apprentices, published index reliability coefficients, and three strands of validity, two of which are doctoral, directional and not peer-reviewed. Efficacy is answered on this page.

Here is what Cognassist holds. An Ofsted inspection of Aspiration Training in May 2026 found that, as a result of this support, the achievement rate of apprentices with barriers to their learning has substantially increased and is in line with their peers. Aspiration Training reports a 20% increase in the proportion of learners receiving personalised support within two months, which is a measure of reach rather than of effect. And the platform reports operational evidence to leaders: the share of learners moved off proactive support with a recorded reason, plans co-created and agreed with the learner, support recorded monthly against Intent, Implementation and Impact, and reviews reaching an explicit continue, adjust or step-off decision. There is no published efficacy study of the support plan.

No. There is no controlled comparison, no effect size and no peer-reviewed paper on the Cognassist support plan, and anyone claiming otherwise about any support tool should be asked for the study. Customer-reported and inspector-reported outcomes are real and worth weighing, but they are not a study: no control group, several things changing at once, encouraging rather than conclusive. The work that would close the gap is a study of the step-off and outcome data the platform already holds.

Because a support model that only ever adds learners is measuring its own activity. The cycle is built so a learner can come off proactive support on the evidence that it worked, with the decision and its reason recorded. Stepping off is different from declining support, which is a legitimate recorded choice revisited at the next review. Either way the safety net stays: Kai and the monthly self-serve module library remain available to every assessed learner.

That is what the design is for. The plan is drafted from the learner's assessed needs, each adjustment pre-written in Intent, Implementation and Impact, and the tutor curates from a library of modules created by the Cognassist Science Team rather than authoring from scratch. A confident non-specialist can therefore produce a plan of a consistent standard, tutor to tutor, in around ten minutes. Specialists are amplified rather than replaced, and concentrate on the most complex needs.

A system of record stores what a tutor wrote. It does not decide what should be in the plan, coach the quality of the record as it is written, or hold the decision to continue, adjust or end support. Cognassist traces every adjustment back to the measured barrier it answers, and that trace reads backwards from an evidenced impact to the original assessed need.