The unit economics of a discovery portal

What a discovery portal actually is

Before the numbers, a definition worth being precise about. A discovery portal is not a marketplace. A marketplace clears the transaction on-platform; a portal introduces and steps aside. A portal is also not a lead-generation shop. A lead-gen shop sells leads without owning the qualification model; a portal owns the qualification model, and everything downstream depends on that ownership.

The distinction matters because it decides which numbers the portal is allowed to be measured by. A marketplace answers to gross merchandise value. A lead-gen shop answers to cost per lead. A portal answers to something in between, and reporting the wrong number is the fastest way to confuse investors, partners, and, eventually, yourself.

The three numbers

Number one: cost per qualified lead (CPQL). Not cost per lead. The word that matters is "qualified," and its definition is negotiated with each counterparty. A university might define qualified as a prospect within a defined age range, with a stated intent to enrol within twelve months, at or above a minimum eligibility bar. A hospital might define it as a self-declared symptom that matches its specialisation. The definition is not the portal's to make; the definition is the portal's to encode. CPQL is the number that separates portal economics from lead-buy economics. A lead-buy business can improve cost per lead by loosening the definition; a portal cannot, because a loose definition breaks the counterparty relationship, which is the only relationship worth having. CPQL constrains the portal to honest acquisition, which is the whole point.

Number two: session-to-inquiry rate. The purest product-market fit signal a portal has. Not bounce rate, not session depth, not time on page: those are proxies for engagement. The number that matters is the fraction of sessions that end in a submitted inquiry aimed at a specific counterparty. Nothing else on the analytics dashboard tells you as directly whether the portal is doing its job. Session-to-inquiry is where product and content pay for themselves. Better search, better filters, better counterparty pages, better calls to action; every product decision on a portal should be measurable against this number, and most eventually are.

Number three: close rate at the counterparty. The number the portal cannot control and must know. Once a qualified inquiry is handed to the counterparty, the portal's leverage ends. What happens next is admissions calling back within the hour or not, the sales team following up in three days or ten, the enrolment process being clear or a maze. The portal's economics are hostage to a workflow it does not run. This is the number most portal teams underestimate, and the number the good portals track with a discipline the counterparties themselves often do not. A portal that knows its counterparties' close rates, by cohort, by channel, by season, has a data asset the counterparties will eventually pay for. A portal that does not is a lead broker whose margin is dictated by whoever cares more about tracking.

Why the multiplication matters

Multiply the three numbers, weight by revenue per closed customer, and the result is contribution per session. That is the number an operator should be able to state without opening a spreadsheet.

The compounding is unforgiving. A portal with a decent CPQL, a decent session-to-inquiry rate, and weak close-rate visibility looks, on any single number, like a portal doing well. On contribution per session, it is losing money on the highest-value cohort and cannot say which cohort that is. A portal that improves any one number by 30% typically improves contribution per session by more, because the other two often improve as a knock-on effect. The model is non-linear both ways.

What the Ideovate operating experience taught

Ideovate Research Pvt. Ltd., which I founded in October 2018 and continue to direct, operates simplidistance.com, a distance and online MBA discovery portal serving prospective learners across Indian universities. What follows is directional rather than exact; specific figures are held back for commercial reasons.

Across five years of operation, three observations have held. First, CPQL varies far more by acquisition channel than by content-side improvement, and the channels that look cheapest by cost-per-click almost never win on qualified cost. Second, session-to-inquiry improves in step-changes, not gradients: a single well-shipped product decision moves the number in a way six months of tuning does not. Third, the counterparties whose close rates the portal knows well pay a materially higher effective rate than the counterparties whose close rates the portal has to guess. None of this is proprietary insight. All of it is easy to say and hard to run.

What separates a portal from a lead-buy business

The three numbers, taken together, describe the difference between a portal and a lead-buy business, and it is worth naming that difference plainly because it comes up in every early conversation with an investor.

A lead-buy business optimises for cost per lead against a spread. Its margin is the difference between what it pays to acquire and what a counterparty pays to receive. It compounds slowly, it commoditises fast, and its defensibility is a media-buying skill and a supplier list.

A portal owns its qualification model, its product surface, and its counterparty relationships as a system. The three numbers sit inside the system, not outside it. That ownership is what compounds. Over years, it produces a data asset (which cohorts convert, at what price, into which counterparties) that a competitor cannot replicate by outspending on Google Ads. This is the compounding advantage; it is also the reason portals take longer to look good than lead-buy businesses, and longer still to look tired.