Two operators, two CDR sets, one number that does not reconcile.
Interconnect partners exchange call detail records every month, and routinely disagree on what they show. When the gap is large enough and old enough, it stops being an accounting matter and becomes a disconnection notice before the regulator.
Interconnect is the plumbing of the industry and the source of some of its largest disputes. When a call crosses from one operator's network to another's, the terminating operator is owed a fee, and the accounting runs on call detail records exchanged between the partners. The mechanics are unglamorous — TAP files and ASN.1-encoded records, decoded, rated, and matched on call type, destination, duration, and cost band — and the discrepancies are constant. Missing records, timing differences, and allegations of traffic gapping mean the two sides routinely book different numbers for the same traffic.
Most months the gap is reconciled quietly. When it is not, it escalates fast. In January 2024 the Nigerian Communications Commission granted MTN Nigeria approval to partially disconnect Globacom over long-standing interconnect debt — a step that would have degraded service for millions of subscribers. The NCC suspended the approval for twenty-one days to allow reconciliation, and withdrew it in February once the parties settled. The episode is the template: a billing disagreement that the partners cannot resolve becomes a regulatory proceeding with subscribers as collateral.
The regulatory frame is explicit about evidence. The NCC's interconnection rules require the exchange of call detail records for reconciliation before a disconnection can be approved, and set a minimum indebtedness period — generally sixty days — before an application can be made. In South Africa, ICASA has the statutory authority to adjudicate interconnection disputes referred to it under the Interconnection Regulations, and the major operators must publish reference interconnection offers. In every case, the operator that prevails is the one whose numbers are traceable to records the regulator can inspect.
This is where revenue assurance teams spend their reconciliation effort, and where the work is most brittle. Matching billions of records across two billing systems, identifying the missing ones, and characterising the gap is slow, and the analyst who has done it knows the reconciliation often comes down to a judgement about which side's records to trust on a disputed traffic stream. When that judgement goes to the regulator as the operator's position, it has to be defensible line by line, not asserted as a total.
The temptation with AI here is to let a model produce the reconciliation figure and the dispute narrative. The danger is the same one that runs through every governed workflow in this sector: a number in a regulatory submission that cannot be traced back to the records underneath it is worse than no number, because it undermines the credibility of everything filed alongside it. The reconciliation has to be reasoned, not generated.
The problem compounds inside large groups. An operator with affiliates across several markets reconciles not only with external partners but across its own legal entities, each booking traffic under different local rules and currencies, and a group treasury function has to consolidate positions that were never built to reconcile cleanly. When an intra-group figure cannot be substantiated, the exposure is not only commercial but regulatory, because the movement of value between jurisdictions draws the attention of more than one authority. A reconciliation that holds for the communications regulator has to hold for the tax position underneath it too.
Roaming settlement carries the same problem in a more complex form. International roaming runs on TAP files exchanged through clearing houses under GSMA standards, and the disputes that follow are harder to adjudicate than domestic interconnect because they cross jurisdictions and currencies. Layered over both is the revenue leakage from interconnect bypass fraud — SIM-box operations that disguise international traffic as local to dodge the higher termination rate — which both distorts the call-record picture and is itself a matter the operator and the regulator must reconstruct from records. A reconciliation that cannot separate a genuine traffic dispute from bypass-distorted records is not a reconciliation the regulator can rely on, and not one the partner will accept.
At the regulator, a reconciliation total is worth nothing. A reconciliation that traces, line by line, to the records is worth everything.
Where each sits.
Akki holds both sides' call detail records — the operator's own and the partner's exchanged set — as a governed substrate, decoded and matched, with the lineage of every record preserved. Rather than a reconciliation spreadsheet assembled by hand and discarded after the dispute, the matching is inspectable and reproducible, which is exactly what a regulator's adjudication requires.
Solva reasons over the discrepancy and drafts the operator's reconciliation position, with each claim traceable to the underlying records and the confidence in each disputed stream made explicit. Where a gap cannot be evidenced — where the records genuinely do not support a claim — Solva declines to assert it rather than inflating the operator's position with a number that will not hold at the NCC or ICASA. The discipline that looks like caution is what wins the adjudication.
SyniSense does less here than in the customer-facing articles, and it is worth being honest about why. Interconnect records are commercially sensitive but largely not personal data; the boundary problem is confidentiality between partners and across affiliated legal entities, not citizen privacy. Where reconciliation does touch subscriber-identifying detail, SyniSense keeps it inside the perimeter, but the weight in this workflow sits with Akki and Solva.
Operationally, reconciliation stops being a heroic monthly effort that lives in one analyst's spreadsheets. The matching is reproducible and the gap is characterised with its basis attached, so the team spends its time on the genuinely disputed streams rather than on assembling the comparison from scratch each cycle.
In a dispute, the operator's position arrives at the regulator already evidenced. When the NCC requires the exchange of records for reconciliation, or ICASA convenes an adjudication, the operator can show a position that traces to source rather than a total it then has to justify. That is the difference between negotiating from strength and negotiating from a number.
For revenue assurance, the leakage closes from both directions. Discrepancies that previously went unchallenged because chasing them was not worth the effort become tractable, and the operator stops conceding disputed streams simply because it cannot evidence its own claim under time pressure.
For the relationship with the partner — and these are long relationships — a reconciliation process both sides can inspect lowers the temperature. Disputes that escalate to disconnection are expensive, public, and bad for subscribers; a defensible, traceable position resolves more of them before they reach that point.
For the regulator, the quality of the operator's evidence raises the quality of the adjudication. Bodies such as ICASA, which require the major operators to publish reference interconnection offers and which adjudicate referred disputes, depend on the parties bringing traceable numbers. An operator that arrives with a position evidenced to source shortens the proceeding and narrows the dispute to the genuinely contested traffic — which is in everyone's interest, including the subscribers who would otherwise be the collateral of a disconnection.