Higher education
Unimetrica is the causal AI layer for enrolment yield. It tells you which applicants your outreach and bursaries genuinely persuade, which would have enrolled anyway, and which to stop contacting, so you lift yield without lifting spend.
You cannot buy your way to enrolment. Fees are capped, the funnel thins every cycle, and conversion targets keep climbing. So contact effort and bursary money get sprayed across the whole applicant pool, most of it landing on students who were always coming or never were.
Tools that rank who looks likely to enrol let you congratulate yourself for converting the sure things who would have come anyway, chase lost causes who never will, and pester sleeping dogs you actively put off. Likelihood tells you who, never whether your effort changed anything.
Causal AI measures uplift, how much your outreach actually changes a student's decision, and sorts every applicant into four groups. You learn who to contact, who to leave alone, and what moves the ones genuinely in play.
Enrols because you reached them.
The persuadables convert only because you reached them. They alone generate incremental return, and they are where every pound of outreach should land.
Was always coming.
The sure things convert regardless. Contacting them inflates apparent campaign performance while adding nothing; the spend is waste dressed as success.
Never enrolling.
The lost causes never convert, contacted or not. Money spent here is simply lost, but at least it does no harm.
Generic comms put them off.
The sleeping dogs are actively turned away by contact. The forgotten gym membership a promotional email reminds you to cancel. Reaching them destroys value.
Spend on the persuadable. Leave the rest. Hover any box for the detail.
Concentrate outreach and bursaries on the persuadable, and a fixed budget converts more students.
Suppress the sleeping dogs your generic comms put off. You protect yield by contacting less.
For each segment, the model surfaces the offer, message and timing with the highest modelled uplift, so every touch is tailored without adding staff.
Trained on your funnel, your courses, your competitors and your outcomes, not a sector benchmark.
Unimetrica sits on top of your student CRM. It reads your funnel and outcomes, then writes back scores, segments and a do-not-contact list. No rip and replace, no new workflow for your team.
93% of students say a personalised message would make them explore a school further.
EAB, 2024
Target half your applicants and you still capture around 75% of the enrolment gain.
Stanford field experiment, Athey et al., 2024
We prove uplift with a randomised holdout, treated against control, so the gain is measured on your own cohort rather than asserted. Modelled end to end across the full funnel, with warm-start archetypes that work even where a course is data-thin. No UK-active vendor offers causal yield modelling, and OfferFit's 325-million-dollar Braze acquisition shows the category's value.