How to read this
A start is a student who begins classes, not an applicant or a deposit. "Enrolled start" is the spend-to-tuition milestone, and 10 to 40% of admits never reach it (the "summer melt"; NCAN). Institution type is a cohort filter on purpose: for-profit and nonprofit economics differ enough that they are never pooled, so your peer set is matched on sector before any number is drawn.
1 · Match your peer cohort
2 · Your last 12 months
A · Where you sit: cost per enrolled start across the cohort
Only 43% of higher-ed marketers track cost per enrolled student at all. Knowing your number and where it sits is already a competitive position. Industry-wide it averages about $1,505 per enrolled undergrad and $3,804 per grad student.
B · The funnel, you vs. cohort median (per 100 inquiries)
C · Your top three levers
This is where the conversation starts.
You've just seen your funnel against - matched peers. A Level strategist can walk the same slice with you, channel by channel and program by program in the outcome calculator, and show which lever moves first.
Talk to Level about your cohort →Prototype note: this is the qualified-lead handoff point (no form here). In production this exact UI runs on Level's real benchmark dataset, and the slice you built becomes the first slide of the call.
One-pager: the benchmark mirror
The 2026 Higher Education Benchmarks Report is a genuinely scarce asset ($100M+ in managed spend, 500K+ student inquiries) locked in a static gated PDF. Prospects skim the averages, file it, and forget it. What they actually want is to see themselves in the data. And the timing is sharp: with Google AI Overviews on roughly 48% of queries and paid CTR down from about 19.7% to 6.3%, inquiry volume is falling while inquiry quality rises, which means every benchmark from the volume era needs re-reading, and institutions know it.
- A peer-cohort slicer over the benchmark dataset (degree level x region x institution type x program size) with honest small-n warnings.
- The prospect's own funnel computed from four numbers they already know, positioned against the matched cohort: percentile, distribution, stage-by-stage leak.
- Three quantified levers generated from where they lag the cohort: the sales conversation pre-written by the data.
Same UI on Level's real benchmark dataset, anonymized and aggregated so no client is identifiable. The slicer sits behind the existing form gate, but instead of a download counter, every session emits a scored lead: cohort selected, numbers entered, gaps found, levers shown. A CRM webhook drops that context straight into the seller's hands, so the first call opens on the prospect's own funnel instead of discovery questions. Two to three weeks from dataset access to live.
I've sat on both sides of this exact table. At Sparkroom I was Director of Solutions for higher-ed performance marketing: I closed and implemented $1.5M/yr in education SaaS, selling measurement tools to the same enrollment-marketing buyers this targets. At Meta I was a senior data scientist leading a 10-person ads analytics team; benchmarking advertiser performance at scale was the day job. Jeff Pinto · jeff@jeffpinto.com · jeffpinto.com
The synthetic cohort is generated to published 2026 higher-ed norms so a prospect's real numbers land in a believable distribution: average cost per enrolled student $2,849 ($1,505 undergrad, $3,804 graduate), average cost per inquiry $140 (range $29-$450), and the fact that only 43% of institutions track cost per enrolled student at all. Search Influence 2026 / UPCEA. Enroll and start are kept distinct because 10-40% of admits never start (summer-melt research). In production this runs on Level's own benchmark dataset; the synthetic version exists only so the prototype is shareable without exposing client data.