The next partner dollar: cheap inquiry, or enrolled start?
A fee-for-service OPM does not own the enrollment P&L; the university partner does. So the metric that pays at renewal is the partner's cost-per-enrolled-start, demonstrated, not cost-per-inquiry. This splits one partner program's media buy across channels under per-channel ceilings (the boast) and a total cap (the bound), then flips the objective from inquiries to enrolled starts. Switch the partner program in the selector and the same method reruns: that portability is the point. The per-channel cost numbers are synthetic and labeled; only the public funnel and base are real.
One method, many partners. Switching reloads the public anchors and reruns the same optimizer: the resellable-across-partners proof.
Try a scenario: one click sets the plan
Fine-tune (optional)
Inquiry cost is the fast metric a channel team reports. Cost-per-enrolled-start is the number that wins the partner's renewal. The finding: an inquiry-optimal plan is usually pessimal on enrolled starts.
Hard upper bound on this partner program's quarterly media spend. The synthetic default scales to the public enrolled-start base of the selected program. No increase in total cost.
Flat ceiling each channel may grow over last period. A flat percent is a placeholder; the honest version is a per-channel number negotiated against what each vendor can actually deliver. That negotiation is the real engagement.
Multi-period weighted scorecard; the rest splits 5 / 10 / 15 across the older three periods. You want more than one bar to point at when a partner asks why the buy moved.
Channel
Cost / inquiry
Cost / start
Plan $
vs base
Sources & method
Funnel shape and enrolled-start base: tuned to the public structure of a program family on IPEDS and College Scorecard (program enrollment, completions, cost, outcomes by UnitID). The applicant-to-enrolled-start ratio is public-shaped; the absolute counts here are illustrative.
Real labor demand: each program's good-fit story is checked against the BLS Occupational Outlook Handbook for the credential it sells (nursing demand is the canonical strong case). This is the anti-predatory-volume check: optimize toward enrolled starts in programs with genuine labor-market demand.
Per-channel cost-per-inquiry and inquiry-to-enrolled-start rates: SYNTHETIC and labeled, calibrated to the structure of higher-ed paid media (a brand or top-of-funnel channel buys the cheapest inquiries and converts to enrolled starts at the worst rate, which is where the trap springs). iDesign discloses no media spend and partner programs disclose no realized conversion rate, so no number here is anyone's real figure. The real cost-per-enrolled-start is sized only on a partner's own export, after the data-readiness read.
Method: boast-and-bound, published at jeffpinto.com/notes/gcu-media-planning. Solver: greedy fill of the highest-yield channel to its ceiling, then the next, optimal for a single linear objective under one budget bound plus box ceilings; it agrees to the dollar with the scipy LP in the published repo. What it deliberately can't do: multi-touch attribution across the months-long inquiry-to-enrollment gap. Read the gap between the two objectives, not a single channel's bar as a real spend order.