Revenue Verification at Scale: Auditing Declared Traffic Against Records

From declaration to evidence

In the companion to this piece I argued that most regulators are regulating on trust: you receive what operators declare, and you have no independent means to check it. This post is about what the alternative actually looks like, because “verify it independently” is easy to say and worth being precise about.

The goal is a control point. Somewhere you can take a declared figure (traffic, termination volume, revenue) and set it against records you hold yourself, at the scale the real data comes in, and produce a difference you can defend. Not a suspicion. A number, with the working shown.

The scale is the hard part, and it’s a solved part

The first objection is always volume. A mid-size operator generates billions of call records a month. Ingesting that, keeping it, and querying it sounds like an infrastructure problem most authorities aren’t resourced for.

It was, until fairly recently. It isn’t anymore, and the reason is the shift in how large data is stored and queried. You no longer need a vast, expensive database that couples storage and compute and charges you for both whether you’re querying or not. Records can sit in cheap open storage, described by an open table format, and be queried in place. Years of CDRs stay queryable rather than archived to tape. The cost curve that used to make this the preserve of the operators themselves has flattened enough that an authority can hold the same depth of data the operators do.

That’s the quiet change that makes independent verification possible. Not a new algorithm, a new economics of keeping and querying evidence at scale.

What verification actually checks

Once the records are in a place you control, the checks are not exotic. They’re reconciliations, and they’re the ones a regulator would design themselves.

Declared international termination against the actual inbound international CDRs. If the operator reports one volume and the records show another, that’s a quantified gap, by route and by period.

Interconnect traffic between two operators, each reporting its own version. The authority is the only party positioned to hold both sides against each other, neither operator will reconcile the other’s books, but you can.

Traffic that should attract a fee against traffic that was declared for that fee. This is where bypass surfaces: international minutes behaving like international minutes in the records, but declared as local. The pattern is visible in the data even though it’s invisible in the declaration.

None of these is clever. All of them are impossible without the records, and trivial with them. The capability was never the analysis. It was the access.

The two things that make it stand up

For a regulator, a verification capability is only worth having if its findings survive contact with an operator’s lawyers. Two properties decide that, and they’re worth building for deliberately.

The first is that the data sits inside your own infrastructure, under your control. Operator traffic data is among the most sensitive data a country holds, and an architecture that only runs in a foreign public cloud fails on law and on principle before it fails on anything technical. On-premise isn’t a preference here for most authorities, it's the thing that makes the capability usable at all.

The second is explainability. When you put a discrepancy in front of an operator, you cannot say “the system flagged it.” You have to show exactly how the figure was produced, which records, which period, which calculation, step by step, in a form a tribunal or a minister can follow. An accurate finding you can’t explain is not evidence. The value of working from open, auditable records is precisely that the path from raw data to conclusion is inspectable. You can show your working, because the working is all there.

How Synaptique builds this

This is the capability we build at Synaptique, so it’s worth being concrete about how, rather than leaving it abstract.

Synaptique provides independent revenue and traffic verification for telecom regulators on a data lakehouse platform. Operator records (CDRs, signalling, mediation and billing outputs) are ingested at full volume into open storage the authority controls, described in an open table format (Apache Iceberg), and queried in place. The reconciliations that compare declared figures against actual records run as transparent, documented steps, so every figure a regulator presents can be traced back to the source data.

The result is a single control point an authority owns: declared traffic, termination volumes and revenue, set against the records that either confirm or contradict them, continuously, not once a year, and in a form built to hold up under dispute.

What it changes

The shift is less about catching operators and more about the basis you regulate from. Today a declaration is the end of the conversation. You receive it and act on it. With an independent control point, a declaration becomes the start of a conversation you can actually have, because you have the other side of it.

And the fiscal dimension follows for free. The same records that verify traffic volumes verify the base that licence fees and sector taxes are assessed on. An undeclared minute is also an uncollected levy. The evidence that supports your QoS and bypass work is the same evidence that supports fiscal verification, you build the capability once and it serves several mandates.

Where can you start ?

Not with a procurement. With a scoped test.

Take one period and one defined scope ( a single international route, say) and reconcile the declaration against the records for that period only. It’s bounded, it’s cheap, and it ends on a real number: here is what was declared, here is what the records show, here is the gap.

That number is the whole argument. Once an authority has seen a genuine discrepancy on its own data, for one route, in one month, the case for a permanent capability makes itself, not from a slide, but from its own evidence.

This is exactly where we come in. Synaptique runs that scoped test with you: we ingest the records for the route and period you choose, into infrastructure you control, reconcile them against the declaration, and hand you the result the declared figure, the actual figure, and the defensible gap between them, with the working shown. No permanent commitment, no procurement cycle; just one route, one period, and a number you can stand behind.

If you’d like to see what that involves on your own data, get in touch and we’ll scope it with you.