Ask a performance marketer what an inbound call is worth and you'll get a rate-card answer: insurance calls go for this, home services for that. Ask an economist and you'll get the correct answer: a call is worth whatever the mechanism that sells it is designed to let buyers reveal. The price of a phone lead isn't a market fact. It's an output of auction design — and most of the pay-per-call industry is running a design that quietly punishes honest bidding.
Display advertising already ran this experiment at scale. For over a decade, programmatic exchanges used second-price auctions — the Vickrey design, where the highest bidder wins but pays the runner-up's bid. The elegant property of second-price is that truth-telling is the dominant strategy: you bid your actual maximum value, because you'll never be charged it unless a competitor forces the price there. If you value an impression at $50 and the next bid is $20, you win, pay roughly $20, and keep the surplus. Then, between 2017 and 2019 — culminating in Google Ad Manager's switch — the display industry moved to first-price auctions, where the winner pays exactly what they bid. The motivations were transparency and yield: first-price is deterministic, easy to reconcile, and immune to the suspicion that a hidden fee sits between the bid and the clearing price.
The cost of that switch landed entirely on buyers. In a first-price auction, bidding your true value is a losing strategy — your whole surplus transfers to the seller the moment you win. So buyers responded with bid shading: models that estimate win probability at every price point and bid below true value, recalculated in under 200 milliseconds per auction. The industry traded a truthful mechanism for a deterministic one, and bought itself a permanent cat-and-mouse game in exchange.
Calls Are Not Impressions
Here's what the display precedent misses when you port it to pay-per-call: a call auction is a thin market. A display auction clears among thousands of bidders over billions of daily impressions, which gives shading models dense clearing-price history to learn from. A call auction clears among the handful of buyers who survive the eligibility filters — the ones in the right geography, inside their business hours, under their spend caps, below their concurrency limits, with a funded balance. That's often two to five bidders, competing for a live, perishable, exclusive good while the phone is actually ringing.
In a thin first-price market, shading gets worse, not better. There isn't enough clearing-price data to model competitors accurately, so rational buyers shade conservatively — bidding well under true value to protect margin against uncertainty. Sellers earn less than the calls are worth, buyers miss calls they'd have happily paid more for, and the market never finds its level. And unlike display, the typical pay-per-call bidder is a roofing contractor or a law firm intake desk, not a trading desk with a bid-optimization team. Asking them to out-model each other is asking them to lose.
This is why second-price auctions matter in pay-per-call — and why it's notable that they're rare in this market. Dial800's Leads Marketplace supports both designs, with minimum bids: first-price where sellers want deterministic pricing, second-price where the goal is healthy competition. Under second-price, a buyer can set their bid to the true maximum a qualified call is worth and leave it alone. No shading, no modeling arms race, no winner's curse. Truer bids, in a thin market, mean better payouts for sellers — the surplus gets competed over instead of hidden.
The Auction Is the Last Step, Not the Whole Mechanism
Auction type gets the economics headlines, but it's one node in a longer decision pipeline, and every stage is mechanism design too. Qualification rules decide what's billable at all — on attempt, on connect, or only past a duration threshold, so a five-second wrong number never becomes an invoice line. Repeat-caller windows deduplicate the same caller returning within a defined period — the commercial cousin of the CDR dedup problem, except here a duplicate record is a duplicate charge. Prepaid balance gates skip out-of-funds buyers automatically, so sellers always get paid. Prime-Time Pricing sets per-weekday, per-time-window overrides, because a Monday 9 a.m. call and a Saturday 11 p.m. call are not the same product.
The part I'd argue matters most is explainability. Every one of those stages is a place a call can route somewhere a seller didn't expect, and in most marketplaces the answer to "why did Buyer B get that call?" is a support ticket. The marketplace should answer it structurally: a Test Route trace that simulates any call and shows every buyer with eligibility, priority, weight, bid, and the reason — "outbid," "caller state outside geo filter," "prepaid balance exhausted — skipped." Disputes die when the decision trace is readable. And because the marketplace runs as a node inside AccuRoute call flows, a call can be qualified by an IVR or AI agent first, auctioned mid-flow, and pulled back into your own routing if no buyer matches — with full call tracking, recording, and AI analysis riding along on every leg. Seller payout and buyer charge land on the same call record, which makes margin a first-class reporting number instead of a month-end reconstruction.
If you run calls through a marketplace — either side of it — ask two questions. Which auction design clears the price, and can anyone show you the decision trace for a single call? If the answers are "first-price only" and "no," the mechanism is extracting value from you, and it's designed so you can't see where.