When a national Christian university college crossed the twelve-month mark in July, we did what we always do: reset the account against the pre-engagement baseline and re-ran the numbers cleanly, without any of the mid-year adjustments the team had made along the way. The 66 per cent cost-per-click reduction is still there. The 360 per cent enquiry conversion lift is still there. But when we sat down to write the year-one internal review, the surprising finding was not any of the headline numbers.
The surprising finding was this: the largest single contributor to the durable CPC reduction was not any of the paid-media changes. It was killing campaign cannibalisation in the account structure. That change alone accounted for roughly forty per cent of the total media saving. And it was the cheapest change to make.
What we thought would drive the savings
Going in, the assumption — informed by every prior engagement — was that the big wins would come from three sources, in this order:
- Landing page rebuilds against sub-sector intent. Improve conversion rate, reduce effective cost per enrolled student, look great in the reporting.
- Performance Max governance. Rein in the algorithmic spend that was serving leads from countries the institution could not actually enrol from.
- Offline conversion imports. Optimise the bidding machine against enrolment value rather than form fills — the change that makes gains durable at month twelve rather than month three.
All three of those matter. All three delivered. But none of them was the largest contributor.
What actually drove the biggest saving
The account had six campaigns running when the engagement began. Three of them were bidding on overlapping keyword sets simultaneously, because the previous agency had built them as parallel "audience-based" campaigns without coordinating negative keyword lists. The domestic-undergraduate campaign and the domestic-postgraduate campaign were both bidding on "study theology online Australia." The international campaign was bidding on some of the same terms too.
Every time a relevant query hit the auction, all three internal campaigns competed. Google's system dutifully picked one of them to serve — usually the one bidding highest that day — and the losing campaigns still paid an internal opportunity cost in the form of budget consumed by lost auctions and worse Quality Scores over time.
Restructuring this into an SKAG-style account with tight negative keyword hygiene between campaigns freed $6,400 per month in the first ninety days. That single change was worth roughly 27 per cent of the pre-engagement monthly spend. It took a week of account restructuring and could have been done by anyone who understood what they were looking at.
Why we didn't see it coming
Two reasons.
First: cannibalisation of this scale is usually not what a well-resourced agency running a well-known institution's account produces. It exists at smaller accounts under agencies short on time. When we did the initial audit, the assumption was that campaign structure would be adequate, and that the real problem would be creative and landing pages. It wasn't. And by not naming this up front, we made the pre-engagement pitch less compelling than it could have been.
Second: the loss shows up in the reporting only if you look for it explicitly. There is no line in Google Ads' standard interface called "budget consumed by internal auction competition." You have to build a search terms report, cross-reference it with the campaign attribution logic, and calculate the loss manually. Every agency who had looked at this account before had produced quarterly review decks that never included that calculation.
What we'd do differently at another Christian higher education institution
Three things.
Audit the account structure before the creative
The instinct — because creative is more interesting and easier to talk about in a first meeting — is to open a discovery process by looking at the ads. That is the wrong order. The account structure audit is a two-day exercise. It is also where the cheapest, largest and most durable wins live. Everything else follows.
Ask specifically about audience-based campaign expansion
Google's own recommendations engine has been pushing "audience-based" and "consolidated" campaign structures for the last three years, on the reasonable premise that machine learning benefits from larger data pools. This is defensible advice for a retail brand. It is often actively harmful for a tertiary institution with sub-sector cohorts that behave completely differently and need separate creative, separate landing pages and separate measurement.
Measure cannibalisation loss in the initial audit report
We now include a "cannibalisation exposure" figure in every pre-engagement audit — even where the exposure turns out to be zero. It sets the expectation that this is a thing that gets checked, and it means the audit report has one number in it that no prior agency has ever produced. That matters for how the conversation goes with the institution's board.
The wider point
The 66 per cent CPC reduction is a good headline number. It is also — and this is the honest part — a headline number that came mostly from correcting a structural error rather than from anything clever. The clever work matters, and the durable enrolment lift will come from the clever work over the next twelve to twenty-four months. But if there is one lesson from year one at a national Christian university college, it is this: the biggest media saving is usually not where the smartest strategy lives. It is where the housekeeping got skipped.
Housekeeping is unglamorous. It is also, in this sector, the difference between a marketing budget that produces enrolled students and one that produces reporting decks.