Applications arrived twice. Now they arrive once.
Enrolment is a paper avalanche wearing a lanyard: web forms, inbox PDFs, employer spreadsheets and photographs of paper — every one of them already filled in by the applicant, and every one of them typed again by somebody in admissions.
Does this sound like your admissions team?
The information was all there. That was the problem.
Every detail the team needed had already been typed by the applicant. It arrived across four channels in four shapes, and the only thing standing between it and the student record system was a person with a keyboard.
THE SAME FORTY APPLICATIONS — A LOST MONDAY VS A QUEUE THAT WAS ALREADY CLEAR
- Capacity was capped by typing speedMarketing that worked produced a backlog rather than students. The only lever for a busy intake period was another pair of hands.
- The expensive gaps surfaced lateA missing ID or funding evidence noticed in week two delays a start date. Caught at intake, it costs one automated email.
- Employer bookings were the worst offenderThe highest-value enquiries arrived in the least convenient shape: many learners, one file, and no two employers formatting it the same way.
- Nobody could see the queueApplications existed in four inboxes and a spreadsheet. “How many are waiting” required a person to go and count.
Every channel read. Every record matched. Every gap chased.
One intake pipeline reads whatever arrives — web form, PDF, employer spreadsheet, photographed paper — extracts what the student record system needs, and checks it against what is already there before writing anything.
- Reads a form, and reads a photograph of a formA structured submission and a phone snapshot of paper are different problems. The pipeline handles both, and knows when it is looking at one it cannot read.
- Matched before anything is writtenEvery applicant is checked against existing records first, so duplicates are prevented rather than merged later.
- Low confidence means one question, not a re-readAn unclear cohort or a mismatched date surfaces with the ambiguity highlighted. The human decision takes seconds instead of a fresh read of the whole application.
- Missing documents chased on arrivalRequests generate at intake and track to completion. The chase starts the day the application lands, not the day someone reaches it.
Built for the shape admissions actually arrives in.
- Four channels, one recordDownstream teams stopped having to know which door an application came through. The record is the same either way.
- Employer batches are first-classA six-learner booking is parsed as a batch rather than as six separate applications, which is what made it the worst job in the building.
- Duplicate prevention, not duplicate cleanupMatching happens before the write. Merging records afterwards is a task that should not need to exist.
- Peak season is a volume, not an eventThe fortnight before a cohort start no longer changes processing time, because processing time no longer depends on how many people are free.
The queue was clear before anyone opened it.
Admissions stopped starting the week behind. The applications that need a decision reach a person; the ones that were filled in correctly never do.
- Peak weeks stopped being peaksVolume in the fortnight before a cohort start no longer changes processing time, because processing time no longer depends on who is available.
- Admissions moved to the conversationsThe team spends its time on applicants who need a decision or a discussion, not on the ones who filled the form in correctly.
- Employer relationships got easier to serveA six-learner booking is now the same effort as a one-learner booking, which changes what the sales team is willing to promise.
What we didn't solve
Roughly 6% of applications still route to a person. Almost all of them are photographs of paper forms taken at an angle, where the cohort is ambiguous. That share has been stable since week four and we do not expect it to reach zero.
The applications were never the problem. The typing was.
Enrolment paperwork exists to protect the learner and the funding. It just arrives in volumes and formats that only a machine should be reading first.
What's hiding in your intake queue?
Send us nothing — just tell us your application volume. We'll show you what machine-read enrolment would return, before you spend a dollar.
Want results like these?
Tell us where your team is losing hours. We'll show you exactly what automation can do about it — and what it's worth, before you spend a dollar.
30 minutes · No pitch deck · An honest first read