In higher education the decision journey is long and multi-touch. A prospect searches for the programme, compares on social media, fills in a form, applies weeks later — or doesn't. If what you call a "conversion" is the form submission, every channel claims the same prospect and budget flows to whoever brings the cheapest form. The cheapest form is usually the least qualified prospect.
Which KPI buys which behaviour
Whatever you choose as the optimisation target in this sector, the platform will bring you more of it — quite literally what you asked for. Each link in the chain rewards something different:
- Form submission: The easiest signal to produce. It also comes from the wrong country, the wrong budget, and from people who were "just looking". Optimised for alone, cost per form drops while cost per enrolment rises.
- Qualified lead (MQL): The prospect who answers the phone, verifies their email, applies to the right programme from the right country. Measuring it requires CRM integration, but it is what first turns the budget in the right direction.
- Application: The signal closest to the real business outcome. Volume is low, so platform learning is slow — use it together with qualified leads, not alone.
- Enrolment: The final outcome — months after the ad cycle. Use it for channel and period strategy reviews, not campaign optimisation.
Extend the funnel
The first thing we did for two universities at Eduhub was to extend measurement from the form to the CRM: form steps via GTM/GA4, and leads landing and qualifying in the CRM tracked with channel tags. The picture changed overnight: Meta brought many forms but unqualified; search brought fewer forms and more real applications. Budget shifted accordingly.
Building the chain: five steps
- 1 · UTM discipline. A consistent utm_source/medium/campaign template for every campaign and ad set. If "facebook", "Facebook" and "fb" land in reports as three different channels, the chain breaks at the first link.
- 2 · Hidden form fields. UTM parameters and click IDs (gclid, fbclid) are carried into the CRM as hidden fields on submission. Without this, nobody can tell which ad a CRM prospect came from.
- 3 · Stage definitions in the CRM. "New", "contacted", "qualified", "applied", "enrolled" — defined with clear criteria, together with the admissions team. Vague stage definitions produce vague reports.
- 4 · Offline conversion upload. Qualified-lead and application stages are sent back to Google Ads and Meta with the click ID. The platforms now learn to find qualifying prospects, not form-fillers. This is where the whole setup starts paying for itself.
- 5 · One report. Channel-level form → qualified → application rates in a single table, same numbers for everyone. Marketing and admissions quoting different figures is this sector's classic time sink.
Country-level patience
An international prospect pool behaves very differently by country: search language, device, decision speed. In some markets the decision takes weeks. Cutting the learning period short and declaring "it doesn't work" is one of the most expensive mistakes. My practical rule: I don't judge a market's performance before data has accumulated for at least that market's own average decision time. Closing a six-week-decision country in week two is ploughing the field because it hasn't flowered yet.
Programme page = landing page
A programme page written well for SEO is also the best landing point for paid traffic. Instead of managing two pages we optimised one for both: programme content, admission requirements, fees and scholarships, FAQs and one clear form. Building a separate "campaign landing page" while neglecting the organic page means doing the same job twice and leaving both mediocre.
Three common mistakes
- The shorten-the-form reflex: Fewer fields make forms cheaper and quality lower. Qualifying questions like country and programme must stay; friction is not always the enemy — sometimes it is the filter.
- One creative for all markets: Assuming the same visual and message works everywhere. Decision motivation differs by market: rankings and prestige in one, fees and visa ease in another.
- Flat budget through the season: Spreading budget evenly across the year when application windows are known. Where the demand curve goes, the budget curve should follow.
Reporting rhythm: weekly signals, monthly decisions
In long-decision-cycle sectors the reporting rhythm must match the funnel. The weekly meeting discusses forms and qualified leads — fast signals; campaign tuning runs on them. Applications and enrolments come to the table monthly and per intake — strategy metrics; channel and market decisions run on them. Mixing the rhythms produces one of two mistakes: panicking over application counts in a weekly meeting (the data hasn't matured), or discussing cost per form in the monthly one and missing the real question. The most functional format I've seen is a one-line health card per channel: cost per form, qualification rate, application rate, that market's decision time. Four columns tell the whole story.
Measurement chain quick checklist
- Is the UTM template written down and used identically by everyone?
- Does the form carry UTM + gclid/fbclid into the CRM via hidden fields?
- Were CRM stage definitions written together with the admissions team?
- Do qualified leads and applications flow back to the platforms as offline conversions?
- Are phone and WhatsApp contacts part of the chain?
- Are channel-level form → qualified → application rates in one report?
- Is each market's average decision time known — and budget judged accordingly?
The lesson
Whatever your industry, measure conversion where the business actually happens. Forms, appointments, carts — intermediate steps. Budget decisions belong to the final outcome. Once measurement reaches the CRM, the "which channel is better" debate ends; the data speaks and the meeting gets shorter.
Short questions, short answers
We have no CRM — can this still be done? At small scale even a spreadsheet builds the chain: hidden form fields land in the sheet, the admissions team updates a stage column, a weekly export is uploaded to the platforms. Discipline decides, not tooling.
How much data does offline conversion upload need? There is no hard threshold; platforms start learning even from single-digit weekly counts per stage. Little data beats no data.
Can an agency do this setup alone? The technical part, yes; stage definitions and CRM discipline don't happen without the institution. The model that works best: the agency builds the measurement, the school owns the process.