CDR-based revenue reconciliation and AI-assisted leakage detection across voice, data, SMS, VAS and roaming. SHIELD checks that every chargeable event is captured, rated once and correctly, and billed — and routes what is not to the people who can recover it.
The pressureFlat ARPU and heavy 5G and fibre capex leave no room for silent leakage.
Why it matters to an operator executiveEvery unbilled CDR, mis-rated bundle or unmatched roaming file is margin lost on a flat-ARPU base. SHIELD gives the CFO a continuous, evidence-backed view of leakage and a team that recovers it.
Highlighted: the stages this use case changes. See how our services map to it →
The fraud and revenue-assurance use cases come together in the SENTINEL command centre →
We start from the KPI an executive owns, not from the technology — then work back to the decisions, data and agents that move it.
Revenue-chain map and leakage value case
Reconciliation rules, leakage taxonomy, KPI baseline
CDR-to-bill reconciliation on 1–2 streams; agent squad live
RA team working the case queue and recovery workflow
More streams, interconnect and roaming; managed RA operations
Same five phases on every engagement; the pilot (phase 3) runs on your data with your team. How the methodology works →
Usage is recorded by the network, collected by mediation, rated by one or more charging systems, billed, and settled with interconnect and roaming partners. Each hand-off is a place where an event can be dropped, duplicated or priced wrongly — and most of those errors are never seen by the customer or the P&L.
Traditional revenue assurance samples records, runs monthly controls and chases exceptions in spreadsheets. By the time a leak is found, weeks of revenue have gone. SHIELD reconciles every record, continuously, on the governed Data Fabric — and puts a prioritised, evidence-backed case in front of the right owner.
SHIELD runs on the DaasLabs Data Fabric Framework, so ingestion, lineage and data quality come with it.
Switch, mediation, charging, billing and partner files land on the fabric, batch or streaming.
Multi-format CDRs (ASN.1, CSV, XML) are parsed, validated and de-duplicated.
Record-level matching of network events to rated and billed events, and to partner statements.
Rules plus ML baselines flag gaps, rating anomalies and unusual variances.
Each leak is typed, sized and routed to its owner, highest impact first.
Re-billing, partner claims and control fixes are tracked to closure.
Configurable starting points our teams tailor to your network, rating logic and controls.
Each panel below is drawn from the platform APIs on this site, over the synthetic operator dataset.
| Event type | Events | Total | Average |
|---|---|---|---|
Agents parse, match, detect, classify and chase recoveries. Revenue assurance analysts approve adjustments and partner claims; every step is logged.
Ingests and parses CDRs from switches, mediation and billing in near real time.
Compares network records with rated and billed events to find unbilled usage and rating errors.
Scores discrepancies and unusual patterns against learned baselines.
Types each leak by stream, severity and impact, and prioritises the queue.
Drafts re-billing, credits and partner claims for approval, then tracks them to closure.
Human approval requiredEscalates high-impact leaks to revenue assurance and finance owners.
| Capability | SHIELD | Manual audits | Basic rules engine | Spreadsheets |
|---|---|---|---|---|
| Detection speed | Continuous, near real time | Monthly or quarterly | Daily batches | Not feasible |
| Anomaly detection | Rules plus ML baselines | Manual review | Static rules only | None |
| CDR coverage | Every record | Samples | Every record | Samples |
| Recovery tracking | Workflow with SLAs | Manual tracking | Basic alerts | Manual |
| Leakage classification | AI-assisted, configurable taxonomy | Manual coding | Pre-defined types | Ad hoc |
| Lineage & audit trail | Built in, event to invoice | Working papers | Partial | None |
| Time to value | 30-45 day pilot | Months | Weeks to months | Weeks |
SHIELD is a configuration of the same governed layers every DaasLabs accelerator uses. See the framework →
A short reading of the panels above, written the way we would brief an operator executive after a first look at their data.
Usage is led by voice, data sessions and SMS, with VoLTE and conference calls a growing tail. Billing shows a material block of failed and overdue payments next to pending bills, and credits and adjustments are frequent enough to point at rating or bundle errors upstream. Roaming revenue is spread across many partners, so no single settlement file covers most of the value.
A leakage baseline for one revenue stream, a ranked case queue with named owners, and the first recoveries tracked to closure — sized on your data, not on benchmarks.
Continuous reconciliation of every record replaces periodic samples, so a broken feed or wrong tariff is caught while it is still small.
Agents assemble the evidence and draft the recovery; analysts decide. Capacity moves from matching to fixing root causes.
Lineage from network event to invoice, with every control result and decision logged for finance, audit and the regulator.
Want a number for your own network? Our value calculator estimates revenue protected from your own leakage and fraud assumptions.
Most operators start with one revenue stream and one or two leakage types, then scale.
Map the revenue chain, data sources and current controls; agree the leakage types and success metrics.
1-2 weeksConnect CDR and billing feeds, configure reconciliation and rules, and run on real data with your RA team.
30-45 daysExtend to more streams, interconnect and roaming; raise agent autonomy where controls allow.
SprintsDataOps, MLOps and AgentOps under agreed SLAs through our Managed Services.
Managed serviceTell us where you suspect revenue is leaking. We'll propose a 30-45 day pilot on your data.