The college campaign that looked huge

Not on this evidence alone. The headline numbers are weak: 900 clicks from 4.2 million impressions is a 0.02% click-through rate, and 11 enrollments is too small a sample to support a confident channel verdict. But the click-to-enrollment pattern is striking. TikTok produced about 198 clicks and 8 enrollments, roughly 4.0%. Instagram produced about 702 clicks and 3 enrollments, roughly 0.4%. That is a tenfold difference in this campaign, though with only 11 outcomes the confidence interval is wide.

Last-click attribution likely favors Instagram because it is the channel people return to before converting. TikTok may be creating awareness that later converts through search, email, or a direct visit. Cutting it would remove a possible upper-funnel driver based on a metric that cannot see that role.

Recommendation: do not cut TikTok. Rebalance to roughly 60% TikTok and 40% Instagram for the spring test, or keep the split and add measurement before moving everything. Set a decision rule in advance: if TikTok still shows at least double the click-to-application rate after 30 days and at least 20 enrollments total, increase its share. If not, shift budget. The advancement office’s proposal is understandable but premature; it mistakes last-click credit for causation.

Both practices carry avoidable risk. The influencer’s photo of a current student needs a release for the college’s marketing use, not just the influencer’s personal post. A student’s enrollment status may also be private under education privacy rules. The college should ask the influencer to remove the image, stop using it in ads, and obtain a proper consent form if the student is willing. The student should be able to decline without pressure.

Scraping 1,200 public comments to build a retargeting audience is riskier than it looks. Public visibility is not consent for advertising segmentation. The comments may include minors, sensitive personal details, or usernames that identify students. Scraping may also violate the platform’s terms, and using the data for retargeting lacks a clear lawful basis. The privacy office should treat the audience as compromised: stop using it, delete the derived list, document what was collected and where it went, and notify the appropriate internal owner.

Replace both practices with permissioned methods: platform-native audience tools, lookalike audiences from consented first-party data, an opt-in form, and influencer contracts that require disclosure and written consent for identifiable people. The remediation should be recorded, not just promised.

Define the decision first: the college needs to know whether TikTok adds enrollments beyond what Instagram would have produced. The outcome metric is enrolled students, not clicks. Because 11 enrollments is too small, the plan needs both more volume and better leading indicators.

Measurement design:

• Use unique UTM links and separate landing pages for TikTok and Instagram, plus a self-reported “where did you hear about us?” field on the application.
• Track the funnel: impressions, 3-second views, clicks, application starts, completed applications, admits, and enrollments.
• Add a geo or time-based holdout if feasible, such as pausing TikTok in two service areas for four weeks, to estimate incremental effect.
• Pre-register the decision rule and the minimum sample before launch. Do not change the rule after seeing results.
• Report cost per application and cost per enrollment by channel with a range, not a single number. With small counts, show the uncertainty.
• Run a short survey of enrolled students and compare it with platform data; neither source alone is reliable.

Eight weeks is short, so start tracking now and avoid a mid-campaign budget switch based on daily noise. The deliverable should be a one-page decision memo with the pre-registered rule, the observed funnel, and a recommendation for the next term.