What Your Verification Records Should Tell You (And Why Most Providers Don't Show Them)

Ask most teams how much they spend on phone verification and you'll get a number. Ask them what that number is made of — how many sends, to which countries, on which routes, how many actually completed — and the room goes quiet.

That's not carelessness. It's usually that nobody ever gave them the data. Verification is one of the few line items in a modern stack where teams accept a monthly total the way you'd accept a utility bill: as a fact of nature rather than something you could audit.

It's auditable. Here's what you should be able to see.

The record underneath the invoice

Every verification attempt generates a record — a call detail record (CDR) for voice, its equivalent for messaging. At minimum, a useful one contains:

• A unique identifier for the attempt
• A timestamp
• The destination — country and network, not just a number
• The channel used
• The status — what actually happened, not just what was attempted
• Duration / response time, for voice
• The cost of that specific attempt

Your invoice is the sum of those records. If you only ever see the sum, you're being asked to trust an arithmetic you can't check.

What aggregation hides

There's nothing sinister about a summary view. The problem is when the summary is the only view, because a monthly total is exactly the right shape to hide the three things you most need to catch.

Where the money went. A total tells you that you spent more this month. It doesn't tell you that the increase is concentrated in four unfamiliar country codes you don't do business in — which is what AIT fraud looks like from the inside, right up until the invoice lands.

What you paid per attempt. Blended rates are averages, and averages are where margin lives. If you can't see the cost of an individual send to a specific destination, you can't tell a good route from a marked-up one.

Whether it worked. "Delivered" is a status reported by somebody in the delivery chain. Sent, delivered, and verified are three different events, and only the third one is a real user. A provider that reports the first and bills you for it has told you nothing about the third.

The three ratios that surface a problem early

You don't need a data team for this. If you can export raw records, three numbers will tell you most of what's wrong, and they're all ratios rather than totals — which is precisely why an invoice can't show them.

1. Sends ÷ completed verifications. The single most useful number in verification. It's the share of your spend that produced nothing. It should be stable. When it climbs while your signups stay flat, something is triggering sends that no human ever finishes — bot abuse, a broken retry loop, or pumping. This ratio moves days before the invoice does.

2. Spend by destination, ranked. Sort your cost by country and look at the top ten. You know which markets your product is in. Any destination in that list that you can't explain is either a market you didn't know you had, or traffic you're paying for and shouldn't be. Fraud concentrates here because that's where the payout is.

3. Completion rate by channel and corridor. A channel that verifies fine in one market and quietly fails in another is invisible in an aggregate. Broken out, it's obvious — and it's usually a routing problem, not a user problem.

None of these require new tooling. They require records with enough resolution to compute them.

Why "no black boxes" is a product decision, not a slogan

We give 100% call detail record visibility, per attempt, with cost broken out and CSV export — and we publish per-attempt rates per country on the pricing page so you can check a record against the rate it should have been billed at.

That combination is the point. Rates without records means you can't verify you were charged correctly. Records without published rates means you can see what you paid but not whether it was right. You need both to actually audit anything, which is why vendors who give you one and not the other haven't given you much.

There's a self-interested version of this argument and an honest one. The honest one: transparency is only meaningful when it can embarrass the vendor. A provider that publishes its rates and hands you the records has given you the tools to catch it overcharging you. That's the entire test. Everything else is a dashboard.

What to ask your current provider

Whoever you're with today, these four questions are worth asking this week:

1. Can I export raw per-attempt records with cost? Not a summary. The rows.
2. Does the record show the destination network, or just the country?
3. Can I distinguish sent from delivered from verified?
4. Is your per-country rate published, or do I have to ask?

Four yeses means you can audit your own spend. Any no is a place where you're trusting a number instead of checking it.

You can't fix a cost you're not allowed to see — and you can't catch fraud in a total.