Rover Insights
For CFOs and finance leaders

Cost Per Lead Is Missing Its Denominator

Across 10,203 recorded conversations, 2.1% of HR buyers were actively looking to replace their system. Priced per unit, that makes the effective rate roughly 48 times the one on the invoice.

Take your cost per lead. Multiply it by the number of leads. That is the line in your model.

Now divide by the share of those people who could actually buy something. Across 10,203 recorded conversations with HR buyers, 2.1% were actively looking to replace their system.

About 48x

Divide your cost per lead by a 2.1% in-market rate and that is the multiplier. At an illustrative $85 a lead, roughly $4,050 per buyer who is genuinely in market.

Use your own cost per lead. The multiplier does not move.

That is the denominator nobody puts in the model. And the vendor selling you those leads is paid on the first number rather than the second.

Who we asked

Before that figure is worth anything, you should know whose answers it is built from.

10,203 recorded conversations with people who run or use an HR system, over twelve months. Nobody applied, nobody was paid, and nobody was entered in a drawing.

34.9% are at companies under 100 employees, 51.8% between 100 and 999, and 13.2% at 1,000 or more. So this is a small-company and mid-market read, not an enterprise one.

46.3% are individual contributors. That is a deliberate trade and it cuts both ways. An HR administrator will tell you the truth about the product and will not tell you the truth about the budget.

Which makes this sample strong evidence on whether a replacement is being discussed, and weak evidence on when it closes.

Why two percent is structural, not a sampling problem

The obvious objection is that we called the wrong people. So here is the same conclusion from filings nobody at Rover Insights wrote.

ADP reports 92.1% client retention for fiscal 2026. Paycom reports 91% annual revenue retention for 2025, and Paychex 82% to 83% payroll client retention for fiscal 2026. When an installed base turns over that slowly, only a thin slice of it can reach the market in any given year.

Sapient Insights Group, surveying 3,318 organizations, finds 10% planning an HRMS replacement, which converts to roughly 2.5% live at any one moment. Different method, same answer, and no amount of media spend changes it.

The 8% figure, which cuts against us

Sapient also reports 8% with an active RFP, close to four times our 2.1%. Better that you hear it here than find it yourself.

The two questions are not the same. An RFP is a procurement artifact that stays open for months, so a survey asking whether one is active catches a wide window. Our caller asks something narrower: is this person working on a replacement right now.

Use theirs for how many accounts will touch a bid this year, and ours for what your reps can call today. We publish the smaller one because it has a person's answer behind it.

The cost that isn't on the invoice

A lead does not cost what you paid for it. It costs what you paid, plus the rep hour spent finding out it was not a buyer.

Reps spend 30% of their week actually selling (Salesforce, State of Sales, sixth edition, fielded March to April 2024, n=5,500). That was 28% in the 2022 edition, so two years of tooling moved it two points.

Meanwhile 67% of B2B buyers now prefer a rep-free buying experience (Gartner, fielded August to September 2025, n=646), up from 61% the year before.

So your reps have about a third of their week, pointed at a population that increasingly would rather not hear from them, hunting a live market of one buyer in fifty.

A thousand records does not create rep hours. It spends them.

Free guide for CFOs

$85 a lead. What did it actually cost you?

Rep hours on leads that never picked up. Pipeline your board reviewed with nothing behind it. Most CFOs have never put a number on either one.

Where the money goes instead

If two percent is the whole market, the honest read is that this category is oversold and the spend is chasing a group too small to carry it. That is not the whole picture.

37.7% of HR buyers who rate their system a 4 or 5 out of 5 can still name a high-priority thing it cannot do. That is 2,476 people of 6,568.

Eleven times the size

Set those 2,476 against the 217 who are actively looking, out of the same 10,203 conversations.

They are content and working around something. Usually a spreadsheet, an export, a step somebody does by hand every month. Your suppression rules almost certainly exclude them, because on every signal your stack can see they look satisfied.

So the reallocation is not from one channel to another. It is from a two percent population your competitors are bidding up, to a population eleven times larger that nobody is contesting.

And one figure that should change what you think you are competing on

Among the things HR buyers say their system cannot do, cost ranks ninth, at 5.0%. Payroll capability ranks first, at 26.7%.

Every vendor in this category believes it is losing on price. The buyers are not talking about price.

Which means the discount your sales org asks you to approve is buying down the ninth objection. That is a gross margin decision made against the wrong evidence.

What changes when the vendor is not paid by the unit

Per-lead pricing is not the natural way to buy demand. It is a leftover from how media was sold, and it survived because it was easy to budget.

The rest of this category already moved. Nobody buys 6sense by the account, Bombora by the topic surge, or ZoomInfo by the record. You buy a subscription and you run it.

The intent and data side went subscription years ago. Content syndication and traditional lead gen kept the per-unit meter running, and that meter is what pays a vendor to send you more of a market that is two percent live.

Why it is one price rather than a menu

The components only work together. The conversations happen because the community exists, the community exists because content runs into it continuously, and the scoring means nothing without a real conversation underneath it.

So the subscription is not a bundle with a discount applied. It is the smallest unit that functions, which is why the price is published rather than quoted.

What we cannot prove

We do not have verified closed-won data across clients, so we are not going to model a payback period for you. Anyone who does is showing you a spreadsheet, not a result.

What we can show you is published pricing and our own score distribution, including the part that does not flatter us. Across 34,016 conversations scored since 2023, 46.2% land below 50, which is the band that says this person is not your buyer right now.

Fewer than one in twenty clears 70. That is what an honest read of a two percent market looks like from the inside.

If your finance function needs client-verified payback before signing, we are probably a quarter early for you.

What it costs

The four Rover Insights subscription tiers, their monthly price, and what each one is built for.
TierPriceBuilt for
Spark$3,000 a monthDigital engagement prospects, content and webinars, limited first-party data
Seed$5,000 a monthAdds conversational intelligence and TruSQL™ scoring
Growth$10,000 a monthAdds conversational and abandonment prospects, third-party activation, dedicated contact
EnterpriseCustomAdds appointment-scheduled prospects, ABM dashboards, native CRM application

You are going to divide it by the lead count anyway

Every finance team does, so here is the honest version. Divided out, the number usually looks worse on arrival than the per-lead price you are paying now.

What the division hides is the disqualification work that moved to our side of the invoice, the rep hours not spent on non-buyers, and the four vendors you are no longer paying separately for content, webinars, data and scoring.

Cost per lead goes up. Cost per closed deal is the number that should carry the decision, and it is the one we would ask you to measure.

Related Questions

It is, and cost per lead is the wrong unit for the comparison. These are recorded conversations rather than form submissions, and the number that should carry the decision is cost per closed deal.
We can't, not with verified closed-won data across clients, and we'd rather say so than model it. What we publish is our pricing and our own score distribution, including the part that doesn't flatter us: 46.2% of 34,016 scored conversations land below 50.
Because filings nobody at Rover Insights wrote say the same thing. ADP reports 92.1% client retention for fiscal 2026, Paycom 91% annual revenue retention for 2025, Paychex 82% to 83% for fiscal 2026. Sapient Insights Group, surveying 3,318 organizations, finds 10% planning a replacement, which converts to roughly 2.5% live at any moment.
It might. Payroll runs 6.1% in the same dataset, almost three times the HRMS rate. The point is not that every category sits at two percent; it is that almost nobody has measured their own number, and the model assumes 100%.
10,203 recorded conversations over twelve months with people who run or use an HR system. Nobody applied, nobody was paid, and nobody was entered in a drawing. 34.9% are at companies under 100 employees, 51.8% between 100 and 999, 13.2% at 1,000 or more, and 46.3% are individual contributors.
Cost per closed deal, not cost per lead, over a period longer than one quarter given the cycle lengths in this category. We would also ask you to track disqualification hours, because that is the cost that never appears on either invoice.

Your Reps
Are Ready for Better Leads.
So Is Your Pipeline.