A public university in the Southeast runs a full omnichannel advertising program: paid search, Meta, geofencing, programmatic display and site retargeting. It is also the kind of institution that tests new platforms early, having run campaigns on Twitch, X and Reddit. When Propellant Media brought the opportunity to advertise inside ChatGPT, where prospective graduate students now do much of their program research, the university treated it the way it treats every new channel: as a test worth running.
The open question was not whether prospective students were in ChatGPT. The university already believed that. The question was what an ad there would cost, whether it would earn a click from someone mid-research, and whether that click could be attributed back to an inquiry. No benchmark existed for any of it, and a low channel minimum made building one cheaper than waiting for one.
What Our Team Was Tasked With
Launch ChatGPT advertising for two graduate programs, one in teaching and one in financial technology, and produce a program-level read on cost and engagement fast enough to inform the next round of graduate marketing decisions. Each campaign ran at $25 per day. The first went live on September 4, 2026, the second on September 11, and this write-up covers the preliminary data through September 17.
Media Plan We Developed
- ChatGPT Advertising: One campaign per program, not one per university. Each program received its own audience brief and its own daily budget so performance could be read program by program rather than averaged into a single number.
- Descriptive targeting: ChatGPT targeting works from a description, not a keyword bid. We supplied who the prospective student is, where they are in the decision, the questions they ask about each program and the keywords that matter.
- Program-specific geography: Targeting was fenced to the markets each program actually recruits from, rather than a blanket statewide footprint.
- Attribution build in parallel: GA4 conversion tracking was connected alongside the launch so the click data would eventually read through to inquiries and applications.
Ad copy and creative were written for someone comparing programs, not someone scrolling a feed. The person seeing the ad is mid-research, so the message reads as a next step in that research rather than an interruption of it.
Campaign Results
| Program | Spend | Impressions | Clicks | CTR | Avg CPC | Avg CPM |
|---|---|---|---|---|---|---|
| Graduate teaching program (live Sep 4) | $252.31 | ~13,818 | ~85 | 0.61% | $2.97 | $18.26 |
| Graduate fintech program (live Sep 11) | $106.42 | ~5,065 | ~21 | 0.41% | $5.07 | $21.01 |
| Both programs | $358.73 | ~18,883 | ~106 | 0.56% | $3.38 | $18.99 |
Platform-reported figures, preliminary through September 17, 2026. Impressions and clicks are derived from reported spend, average CPC and average CPM. Conversion tracking into GA4 is being connected, so no conversion figure is reported yet.
Click-through rate, by program
CTR · same $25 daily budget on both
| Teaching program ~13,818 impressions |
0.61% | |
| Fintech program ~5,065 impressions |
0.41% |
The program a week further in also carried the stronger click rate. Treat the gap as provisional until the fintech campaign has fourteen days of its own.
What the Data Settled
Fourteen days in, the teaching campaign returned a 0.61% click-through rate at a $2.97 average cost per click and an $18.26 CPM. That is a program-level benchmark the university did not have two weeks earlier, and it was produced on $252.31 of spend.
Cost per click, by program
Average CPC · through Sep 17, 2026
| Teaching program $252.31 spent · 14 days |
$2.97 | |
| Fintech program $106.42 spent · 7 days |
$5.07 | |
| Channel ceiling top of the $1–$10 range we see |
$10.00 |
Both programs landed inside the $1–$10 CPC range we see on the channel. The teaching program sat near the floor.
The two programs ran on identical daily budgets and returned different results: a 0.61% CTR against 0.41%, and a $2.10 gap in cost per click. Averaged into one university-level campaign, that difference would have disappeared. Kept separate, it tells the university which program is finding its audience faster and which needs another look at its brief.
At $25 a day each, both campaigns produced a readable CTR and CPC before the combined spend reached $400. The low floor is the reason the test was affordable, and the reason it produced an answer in two weeks instead of a quarter.
What We Take to the Next Campaign
- Build one campaign per program, not one per university. Identical budgets returned different CTRs and a $2.10 gap in CPC. Averaged together, that difference disappears.
- Use the low floor to build the benchmark instead of waiting for one. $25 a day per program produced a program-level read in fourteen days.
- Connect attribution before launch, not after. Clicks are real; the enrollment signal is pending GA4. The next program should go live with conversion tracking already connected.
Measurement in this pilot currently ends at the click. The click data is real and reported by program; the enrollment signal arrives once GA4 conversion events are connected. Both campaigns remain live, and the university now has a cost and engagement baseline for a channel that did not have one when the month began. No enrollment outcome is claimed at this stage.
Frequently Asked Questions
What does ChatGPT advertising cost per click for a university?
In this pilot, $2.97 for a graduate teaching program and $5.07 for a graduate fintech program, with a blended $3.38 across both. That sits inside the $1–$10 CPC range we see on the channel, and the teaching program landed near the low end.
What click-through rate can higher education expect on ChatGPT ads?
This university saw 0.61% on its teaching program after fourteen days and 0.41% on its fintech program after seven, a 0.56% blended CTR. ChatGPT ads are a new format, so we do not cross-compare them to display or search benchmarks; these figures are the university’s own program-level benchmark.
How much budget does a ChatGPT advertising pilot need?
This one ran at $25 per day per program and produced a readable CTR and CPC on both before combined spend reached $400 ($358.73 through Sep 17, 2026). The platform’s low minimum is what makes a program-by-program test affordable.
How is ChatGPT ad targeting set up for graduate programs?
Targeting is descriptive rather than keyword-bid based. We supply who the prospective student is, the questions they ask about the program and the keywords that matter, then scope geography to the markets each program recruits from. Each program gets its own campaign so results read program by program.
Can ChatGPT ad clicks be attributed to inquiries and applications?
Not yet in this pilot. The platform reports impressions, clicks, CTR, CPC and CPM by program; GA4 conversion tracking is being connected so clicks can be read through to landing-page sessions and request-for-information events. The lesson for the next program is to connect attribution before the first dollar is spent.
