The assumptions (move them; the math is exposed)
Does it compound? The lever-type switch.
A diagnostic that says "your cost-per-start spread is X" is interesting; a number that says "X is $Y of EBITDA and $Z of enterprise value at your multiple" is underwritable. Advertising at $247.4M is 13.8% of revenue and the largest discretionary slice of the "student services and administrative" line: a few hundred basis points of efficiency there flows close to dollar-for-dollar to EBITDA, because freed ad spend at constant starts is pure margin. At a 12-16x multiple that EBITDA delta is the enterprise-value swing a board reads and a sponsor prices. This tool does the conversion the rest of the package only gestured at.
The honest caveat, surfaced rather than buried: a one-time diagnostic that names the spread once does not move EBITDA: acting on it every quarter does. The durable version is an instrumented spend graph with named owners, re-run against the real internal numbers each quarter, so the efficiency holds and the growth gets visibly less spend-dependent. That second-order effect (qualified-demand efficiency making enrollment growth more repeatable) is what nudges the multiple itself, not just the opex line. The toggle above is the difference between the appetizer and the entrée.
Why pay for this when you have an in-house analytics team?
You do have one: Covista runs a marketing-analytics function with the real per-brand spend in Power BI and Tableau. So this isn't sold as a capability you lack. It's sold on two things an inside team can't manufacture: independence and regulatory defensibility. An internal model that recommends cutting the budget of the brand you grew fastest is a hard memo to write from inside the org chart; an outside read that's allowed to say "spend less" is structurally easier to trust, and to take to a board or a diligence room. And in a 90/10 + Gainful-Employment + FTC environment, the only marketing-efficiency artifact that survives scrutiny is one where every input is a citation. That's the moat: not the math, which your team can redo, but the posture (independent, every-input-defensible, and convertible into the EBITDA/EV language the people above you actually decide in).
Priced accordingly: a one-time diagnostic at $50-75K, a year-one ceiling near $150K with the internal-data extension, and no recurring subscription: because the durable cadence belongs to your team, not to a vendor's seat licenses.
Sources & method
The bridge math
- Dollars freed = efficiency% × hold% × $247.4M advertising (FY25). Efficiency at constant starts means the freed spend is not re-spent to buy the same starts: it drops to the bottom line.
- → EBITDA = dollars freed, near 1:1 (advertising is an operating expense bundled in "student services and administrative expense," $672.0M FY25; trimming it lifts operating income directly).
- → Enterprise value = EBITDA delta × the EV/EBITDA multiple. The 12-16x band is illustrative of where a for-profit-ed asset of this profile trades / would be underwritten; set your own.
Public anchors
- Advertising FY25 $247.4M (13.8% of $1,788.3M revenue); FY24 $227.9M, FY23 $219.4M: FY2025 10-K, accounting-policies note. Operating income FY25 $341.5M; GAAP operating margin 19.1%.
- Q1 FY26 (Sept-2025 quarter): adjusted EBITDA $112M at a 24.2% margin; Walden +13.6% enrollment at 32.6% segment margin, Chamberlain +2.2% with margin down 240 bps; CEO on the "local marketing effectiveness" and "funnel conversion" miss: Q1 FY26 earnings call.
- Rebrand Adtalem to Covista, ticker ATGE to CVSA: 8-K, Feb 5 2026.
This is a sizing tool, not a forecast or a valuation opinion. The efficiency gain is the lever the diagnostic scopes against your real internal data; the dollar conversion shows why that lever is worth scoping.