Pay-Per-Call Has a Lemons Problem. Here Is How Buyers Fix It

You pay one price for a call. The source that sold it may know far more about that call than you do. It may know whether the caller rang you last week, whether the same caller is being offered elsewhere this morning, and how the caller got to the phone. You learn some of this later. Some of it you never learn.

Economists have a name for a market built on that gap. George Akerlof described it in 1970 with used cars, and it fits pay-per-call closely. This piece walks through his argument, what it predicts for call buyers, and which parts of the gap a buyer can close before paying. Numbers in the examples are illustrative.

The source can know things about each call that the buyer cannot see

Akerlof's setup is short. Ross Anderson restates it in a few lines:

“

"suppose that there are 100 used cars for sale in a town: 50 well-maintained cars worth $2000 each, and 50 'lemons' worth $1000. The sellers know which is which, but the buyers don't." — Ross Anderson, Security Engineering (§ 8.3.3 Asymmetric information)

Swap cars for calls and the seller is the call source. Some facts sit with the source, some with you, and some with nobody yet:

The first rows are facts somebody holds that you don't see at the bid. The last two are facts nobody knows yet. They need different fixes, and this piece takes them in turn.

A price set at the average can push good sources out

If you can't tell a good call from a bad one at the bid, you can't pay more for the good one. You pay for the average call. Anderson draws the consequence:

“

"When users can't tell good from bad, they might as well buy the cheapest." — Ross Anderson, Security Engineering (§ 8.3.3 Asymmetric information)

Whether good sources leave depends on their costs. Here is the case where they do.

That is the spiral Akerlof described. The bad calls are not the whole loss. The larger loss is the good calls that never get offered, because the price can't reward them.

Flat per-call deals settled over spreadsheets are exposed to this. The price is fixed before anyone knows what any call was, and quality gets argued after the money has moved. The comparison of RTB and direct buying covers how each model works. This piece is about which facts each party can inspect, and when.

Call buyers care which caller they get, not only how many

Most markets don't work this way. A grocer doesn't care who buys the apples. Liran Einav, Amy Finkelstein and Ray Fisman call the markets where it matters "selection markets". They introduce them with an airline pass that drew only the heaviest flyers:

“

"But it is just one example of a much larger problem, one that afflicts any market in which sellers care not just about how much they sell but also about whom they are selling to." — Liran Einav, Amy Finkelstein and Ray Fisman, Risky Business (ch. 1, "Death of a Market")

In their insurance cases, the customer holds the private facts. In pay-per-call the roles turn around. The buyer is the party that cares who is on the line, and the source is the one that may know more about it.

A Medicare agency wants a caller in a county its plans cover, who hasn't just spoken to a competitor, and who will reach a licensed agent with time to talk. Volume alone cannot establish any of that. Ten of the right calls can be worth more than a hundred of the wrong ones.

Checks before the bid turn some hidden facts into visible ones

The cheapest way to shrink the gap is to look before you pay. Several facts in the table can be checked while a partner waits for an answer. On TapQuality, a partner ping passes these checks before a campaign bids:

Only then does the blind first-price auction run. TapQuality's own time per ping is under 200 ms; outside lookups you add, like the floor check, come on top. The real-time bidding page explains the checks before bidding.

Each check reveals one fact, and none reveals whether the caller will buy. They don't make a source honest. They make some of what is known visible at the moment it matters, which is before you pay. A source sending clean calls loses nothing by them.

Billable time sets a floor every paid call must clear

Some facts can't be checked before the call, because nobody knows them yet. Whether the caller stays on the line is one. A used-car buyer facing the same problem asks for a warranty: pay for the car only if it runs. Billable time is the call version.

On TapQuality, a call bills only when it passes your billable time. You set it per campaign; if you don't, it's 90 seconds. Short calls, duplicates and declined pings aren't billable spend, so they add nothing to the 1% platform fee. Phone numbers and minutes are passed through at carrier list rates, and voice-agent screening minutes are quoted separately.

Duration is a billing condition, not proof of quality. A 90-second call can still be the wrong caller. What billable time does is move the cost of the shortest, cheapest junk back to whoever sent it.

A recorded billing reason gives both sides the same record to check

Hidden facts cause a second cost after the call: the dispute. The buyer says the calls were bad, the source says they were fine, and each side argues from its own logs. Steven Shavell, writing about disclosure rules in law, describes what sellers do when nothing makes them speak:

“

"sellers will tend not to disclose unfavorable information—such as that the basement of a home leaks—because this information would tend to lower the price; thus a disclosure requirement results in revelation of unfavorable information by sellers and lower prices." — Steven Shavell, Foundations of Economic Analysis of Law (§ 6, Information Disclosure)

A call platform can do part of that job. On TapQuality, a call that didn't bill carries the reason on its record, such as a duplicate or a call that ended before billable time. Partners get a portal with their own reporting. Postbacks can be scoped so a source hears only about its own traffic.

Disclosure cuts both ways, and that is the point. A source with clean calls gains most, because it can point to the record instead of asking to be trusted. That is how a market keeps its good sellers.

The fact that matters most stays hidden for months

For a Medicare agency, the fact worth paying for is whether the sale lasts. A call can pass every check, bill and close, and the policy can still never take effect, or lapse within weeks. No pre-bid check sees that, because nobody knows it yet, the source included.

That gap closes only with time and joined records. TapQuality's Buyer Insights joins call data to your dialer, CRM, appointments and enrollment records, so you can judge each source by policies still in force 90 days after they take effect. The Buyer Insights page shows the funnel from landing page to 90-day retention, with illustrative data. Why short windows mislead, and how to read the slow numbers, is in Judge Call Sources by Policies That Last.

Five questions reveal what a call platform can prove

1. Which facts about a call can I check before I bid, and which only after?
2. Does a caller who already billed with me from the same source bill again?
3. Can I see when the same caller is turning up through several sources?
4. When a call doesn't bill, does the record say why?
5. How long after the sale do I learn whether the policy stuck?

The strip on How it works follows one partner call through seven steps: ping, checks, auction, optional screening, transfer, billing and outcome. To see the checks run against your own call history, book a walkthrough.