Which channels produced enrolled, tuition-paying students?

Application-to-start rates vary enormously between channels. A channel producing applications at half the cost can produce starts at twice the cost, and the reporting most schools use cannot see that at all.

Ranking on tuition from actual starts is the correction, and it usually redirects a meaningful share of the budget within one cycle.

Vocational education is a heavily scrutinized category, and outcome claims are examined. An attribution model that guesses is not something you want in the file.

Only exact matches count automatically. Unmatched tuition stays in Direct / Unknown rather than being distributed across channels, so every figure traces back to two records.

Much of the inquiry volume in vocational education is bought rather than generated. An aggregator runs the advertising, captures somebody interested in a career field, and passes that person to several institutions at once. The prospective student then hears from four admissions teams in a day, which affects how they behave and how likely they are to start anywhere at all. On a cost-per-inquiry report that source frequently looks like the best one available. On a start-and-tuition report it frequently does not.

Combining each aggregator's own lead export into the lead file, each row tagged with the aggregator, is what makes the comparison possible. Every source is then ranked on tuition from students who actually started, over the same period and on the same axis, including your own campaigns and the inquiries arriving through your own website. The spread between aggregators may well be far wider than the spread in what they charge, which is not something any aggregator will tell you.

There is a second reason to hold those files. An inquiry sold to four schools may appear in four aggregators' lists and start at one of them, and a school that cannot see which list a starter came from has no way of pushing back on an invoice. Matching on the student's own contact detail settles it: the inquiry either appears in that aggregator's export or it does not. Where a starter appears in two aggregators' lists, the start is credited to whichever delivered the earlier lead, not to both.

A six-week certificate and a two-year diploma are different businesses sharing a building. They cost different amounts to deliver, they carry different tuition, they attract different people, and the marketing that fills one does very little for the other. A school with a broad catalog that ranks its channels once, across everything, gets a list describing whichever program happened to be recruiting hardest that quarter, and then allocates the following quarter's budget from it. The resulting drift is slow and hard to see from inside.

Adding a program column, mapped as Product / service, shows tuition by program with how much of each traces to a channel. It does not rank sources within each program, but it shows whether the programs carrying the highest tuition are the ones marketing actually reaches, and it can expose a program whose starts arrive almost entirely with no trace in the lead file. None of this is visible in a blended table, and none of it needs new data — only a column the student record system already holds.

Continuation belongs in the same file if the school wants the whole picture. A student who starts and leaves in week three has generated an enrollment and very little tuition; a student who finishes and returns for a further qualification has generated considerably more. Exporting tuition as it is actually billed, term by term or stage by stage, rather than as one headline figure at enrollment, is what lets the report tell them apart without anybody building a separate retention analysis to do it.

Vocational schools do not recruit continuously. Cohorts start on fixed dates, sometimes monthly and sometimes three times a year, and an inquiry arriving the week after a cohort fills is not lost — it is waiting. Reporting on calendar months therefore produces a figure that oscillates with the intake calendar rather than with anything the marketing did, and a month with no start date in it can look like a collapse in performance. It is a gap in the timetable, not in demand.

Comparing intake to equivalent intake is the correction, and it is available immediately because nothing needed to be running when those cohorts started. A school can load two years of starts and two years of inquiries on the first afternoon and compare the same intake across both, which is the only comparison in which the calendar is held still and the marketing is the variable being tested. Waiting to accumulate that comparison costs a recruitment year and buys nothing. A school that waits is simply choosing to make next year's decisions on this year's guesses.

The waiting period between the inquiry and the next available start date is worth measuring in its own right. If one channel produces people who enroll into the next cohort and another produces people who defer twice before starting, those channels have different working capital implications even when their tuition totals eventually match. Because there is no window, both are credited correctly whenever the start happens, and the interval itself becomes a number the admissions team can act on rather than an impression.