It depends on how the verification is run and how the vendor is paid, and the price on the quote is the least useful number in the comparison. This page covers the pricing models you will see, the ranges I have seen from building and running these systems, the formula that shows what verification costs you, and why prices in this category are rarely published, ours included.
The TPV Buyer's Checklist
What to measure, and what to make any TPV vendor prove, including us.
Two vendors can quote the same unit price and cost you very different amounts, depending on what they are paid for and where the cost of a failed verification lands. Start with the model.
You pay when a verification reaches a decision, and in the sharpest version only when it is approved. It looks safe because failed calls seem free. It is the model to question hardest: a verifier paid only on confirmations has a financial reason to confirm, and the cost of the calls that never complete is still in the price, spread across the ones that do.
You pay for every verification attempt, whatever the outcome. The vendor earns the same for a rejection as for an approval, so it has no reason to push a customer toward yes, and you see the cost of abandoned and rejected attempts directly instead of buried in a higher unit price. This is how we price.
You pay for call time, which rewards longer calls. It is easy to confuse with the per-minute rates voice-AI platforms publish for raw infrastructure. Those rates are not a TPV price: they leave out script control, evidence and retention, independence, integration, live-person transfers and quality review, which is most of what you are buying.
You pay for verifier capacity, used or not. Predictable on paper. In practice peak hours queue anyway and quiet hours are paid for regardless, so the cost per retained enrollment moves with your sales calendar.
These ranges come from years building and running these systems, not a published study, because there isn't one. Treat them as a starting point and check them against your own data.
| How it is run | Typical cost | Drop-off |
|---|---|---|
| Live agent, offshore or nearshore | $2 to $5 per verification | 15 to 25% |
| Live agent, onshore | More than offshore. I do not have a range I would stand behind, so I am not giving one. | |
| Automated IVR | $0.20 to $0.80 per call | 30 to 50% |
| Conversational voice agent | $0.30 to $0.70 per call | 5 to 10% |
Drop-off here means verifications that end before a decision. The ranges describe each category, not any single vendor, and not us. We do not publish completion or drop-off rates for our own program; if a vendor shows you theirs, ask exactly how each one is defined.
Read the two columns together. A model with a low price per call can still be the expensive one if its drop-off sits at the high end of its range, because each verification that ends early can be a closed sale you lose. That cost never appears on the vendor's invoice.
Count everything in the top line: attempts, retries, transfers to live verifiers, manual review, integrations, record storage, and every closed sale lost before verification reached an outcome. Per-call price is not this number, and the two regularly point in opposite directions.
Some inputs are hard to see, false passes most of all, because they surface later as disputes and cancellations. Include post-verification cancellations and disputes when you can. What a false pass is and how to find one.
Prices depend on things that vary by program: the length of the script, the number of states and channels, languages, live-transfer coverage, how long records are kept, the integration work, and volume.
Units differ, too. A number published per completed verification invites a comparison with a number per attempt or per minute, and the comparison misleads both ways. Contracts can also carry minimums, setup fees and storage charges that a headline price leaves out.
And the per-minute rates voice-AI platforms publish get mistaken for TPV prices, which pulls every published number toward the wrong comparison.
We do not publish ours either. We price per attempt, so a rejection costs us nothing, and we will give you a number for your program when we talk.
These ranges come from years building and running these systems, not a published study, because there isn't one. Offshore or nearshore live agents run about $2 to $5 per verification, automated IVR about $0.20 to $0.80 per call, and conversational voice agents about $0.30 to $0.70 per call. Onshore live agents cost more than offshore. Check the ranges against your own invoices, and compare total program cost rather than unit price.
It looks safe to the buyer, because failed calls seem free. But a verifier paid only on confirmations has a financial reason to confirm, and the cost of the calls that never complete is still in the price, spread across the ones that do. Ask whether the vendor will price per attempt instead.
Per-minute rates published by voice-AI platforms are prices for the voice and phone layer. They leave out script control, evidence and retention, independence, integration, live-person transfers and quality review, which is most of what a verification program needs. Compare total program cost divided by retained, compliant enrollments.
Every attempt and retry, transfers to live verifiers, manual review, integrations, record storage, and every closed sale lost before verification reached an outcome. Divide the total by retained, compliant enrollments. Per-call price is not this number.
Cheapest per call and cheapest per retained, compliant enrollment are different questions. A model with a low price per call can be the expensive one if customers abandon it, because each abandoned verification can be a closed sale you lose. Work out the total-cost number from your own data for each option.
No. Prices depend on the script, the states and channels, languages, live-transfer coverage and volume, so a single published number would mislead more than it helps. We price per attempt, whatever the outcome, so a rejection costs us nothing, and we will give you a number for your program when we talk.
Bring it to a vendor call: the number that matters, the nine numbers to make any vendor separate, the five things to make them prove, and the two questions that sort vendors fastest.
Tell me what you verify and roughly how many a month, and I will reply by email with how I would work out your number. If you would rather talk it through, pick a time on my calendar.
Pick a time on my calendar
Ryan Schuetz
Founder, Automatdo