A five-month Google Search and YouTube campaign for a private healthcare university college: the conversion tracking system built from scratch, the five-lane account structure, and the result, 362 completed applications on a RM4,900 monthly budget, with cost per click cut by 43.6 percent along the way.

Client

A private healthcare university college in the Klang Valley

Sector

Higher education, nursing and allied health sciences

Service

Digital marketing: Google Search, YouTube, conversion tracking

Period

April to August 2026, five months

Media budget

RM4,900 per month, RM24,500 total

Outcome

362 completed application forms, at roughly RM68 of media each

The starting point

In April 2026 the college had a live Google Ads account and no idea what it was doing.

That is not a criticism of the college. It is the normal state of affairs in Malaysian higher education. Ads were running. Money was leaving the account. Applications were arriving through the website. Nobody could draw a line between the three.

Specifically, at the point we took the account on:

• No conversion tracking existed on the website. Not on the Apply Now button, not on the application form, not on the enquiry banner. Spend and enrolment were two unconnected facts.
• Cost per click sat at RM3.51, high for the category and rising.
• The account defended the institution’s own name and did little else. Students searching by programme rather than by institution never saw it.
• The monthly budget was RM4,900 and it was not negotiable. Every competing institution in the same auction was spending multiples of it.

The challenge

Three jobs, all of which had to be done inside the same RM4,900.

1. Protect the name

Every search for the college by name had to return the college, not a competitor bidding on it. In higher education a student searching your name by hand is the most valuable search there is. Losing it to a rival is the cheapest mistake an institution can make and the most common.

2. Create demand that did not exist

Most prospective students do not search for institutions. They search for programmes. Nursing. Physiotherapy. Medical imaging. A specialist college is invisible to all of them unless it competes on course terms, where it has no name recognition to lean on.

3. Take share from larger institutions

The college shares its auction with multi-faculty universities carrying substantially larger budgets. Showing up is not automatic. It has to be earned position by position.

The approach

We did not ask for a bigger budget. A small budget does not fail because it is small. It fails because it is spread thin across campaigns that borrow from each other. So the account was rebuilt as five separate lanes, each with one job, none of them allowed to subsidise another.

Underneath all five, a measurement layer built and maintained by 3thirds Inc: Apply Now button clicks, application form submissions, home page banner clicks and GA4 engaged sessions, each defined as a separate action so that nothing could be counted twice or mistaken for an application when it was not.

Execution, month by month

Phase one, April: instrument before you optimise

The first month bought very little media improvement and a great deal of clarity. We specified, built and deployed the full conversion tracking system on the college website, then connected it to Google Ads and GA4. It went live at the end of April.

That timing matters and we report it plainly: April’s spend sits outside the attributed numbers, since the tracking was not yet live to capture it. Everything from May onward is fully measured, which is exactly the point of building the layer first.

Phase two, May and June: build the volume

With measurement live, the brand lane was pushed to 82 to 88 percent impression share on core name terms, at Google Quality Scores of 8 to 10, the highest relevance band Google awards. In parallel the faculty campaigns opened across nursing, allied health sciences, managerial science and continuous professional development.

May produced 93 applications. June produced 105. The account had gone from unmeasurable to a predictable monthly number in eight weeks.

Phase three, July and August: buy the same result for less

With volume established, the work turned to cost. Ad copy, landing page relevance and keyword structure were tightened until Quality Scores climbed, and Google discounted the clicks accordingly. Cost per click fell from RM3.51 at launch to RM1.98 in August, a 43.6 percent reduction, with volume held.

The competitor lane opened in this phase, and YouTube was extended into Chinese and Tamil to match the catchment. July produced 75 applications, August 78, at materially lower cost per application than May.

Where it ended, August 2026

Where the applications came from

Conversions here are all tracked actions, not applications. Applications are the 362 form submissions counted separately above. We keep the two apart deliberately, because conflating them is how agencies inflate a case study.

What made this work

Six things behind these numbers, stated plainly, including the two that did not go entirely to plan.

1. Tracking came before scale

The tracking system went live at the end of the first month, and only once it did was the account pushed for volume. April carried the campaign while that groundwork was being finished, so its spend sits outside the attributed numbers above. Every account we run now gets full tracking sign-off as a milestone before media scales, which is exactly the discipline that produced everything from May onward.

2. Brand terms convert. Course terms grow.

73.3 percent of conversions came from students already searching the college by name. On a spreadsheet that argues for cutting everything else. It would have been the wrong call. The faculty lane delivered 3,834 clicks and 446 conversions from students who had never heard of the institution. Cut it and the account looks more efficient next month and smaller every month after that.

3. Efficiency came from relevance, not from bidding down

Cost per click fell 43.6 percent while volume held. That did not come from lowering bids, which would have cost position. It came from Quality Scores climbing to 8 to 10 as ad copy, keywords and landing pages were brought into line. Google discounts relevance. That discount is the only free money in the auction.

4. Volume found its ceiling, and pointed to the next lever

Applications by month ran 93, 105, 75, 78. The account found its ceiling in June and held it at a lower cost from there. Search demand for these programmes inside this catchment is finite, and RM4,900 reached the end of it quickly, which is itself a useful reading: it tells us precisely where optimisation stops paying for itself and a bigger conversation, about budget or new demand, starts.

5. YouTube is a demand channel, not a conversion channel

778,705 video impressions produced 33 last-click conversions. Judged on last click, that is a poor result and an easy line item to cut. Judged on what it is actually for, it fed the branded search volume that produced 73 percent of the applications. The mistake is not running YouTube on a small budget. The mistake is measuring it with the wrong instrument.

6. Conquesting is a cheap lane, never a big one

Competitor terms produced only 2,916 impressions in five months because that is all the available inventory there was. 143 clicks, 17 conversions, at a fraction of the account cost. Worth running every time. Never worth planning volume around.

Client name withheld at the institution’s request. All figures drawn from the client Google Ads account, 1 April to 31 August 2026.