# Data qualityData qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.Voir la définition complète → dimensions for subscriber and network records
A single subscriber can exist eleven times in one telecom's systems: once in CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.Voir la définition complète → (customer relationship management), twice in billing under slightly different spellings, three times in the HLR (Home Location Register, the database that tracks which network a mobile subscriber is registered on), and more in legacy prepaid platforms nobody has fully decommissioned. Multiply that by tens of millions of subscribers and the mess becomes a governance problem with real financial and regulatory consequences, from misrouted bills to failed lawful-intercept requests.
This lesson gives you a working framework to score telecom data against five dimensions, with the specific record types where each failure mode shows up.
Telecom operators run on a denser web of interdependent systems than most industries: OSS (Operations Support Systems, managing network infrastructure), BSS (Business Support Systems, managing customers and billing), and network element databases that update in near real time.
Key datasets you'll encounter:
Because these systems were built at different decades by different vendors, they rarely share a single source of truth. That's the root cause behind most quality failures below.
Are required fields populated? A CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.Voir la définition complète → record missing a billing address won't stop a call from connecting, but it will stall a regulatory obligation like number portability porting requests or emergency-services address lookups (in the US, this maps to FCC E911 location requirements).
Quick check: percentage of subscriber records with all mandatory fields populated.
$$\text{Completeness rate} = \frac{\text{records with all mandatory fields filled}}{\text{total records}} \times 100$$
Example: if 4.7 million of 5 million postpaid records have complete address fields, completeness is 94 percent. Industry practice generally treats anything below roughly 95 percent on regulatory-linked fields as requiring remediation, though thresholds vary by operator and use case.
Is the value correct, not just present? A populated address field that says "123 Main St" when the subscriber moved two years ago is complete but wrong. Accuracy failures are the hardest to detect automatically because the field looks fine structurally.
Common telecom accuracy failure: CDR misrating, where a call is logged against the wrong tariff plan due to a stale rate-plan flag in billing. This directly causes revenue leakage, a term for revenue an operator earns but fails to bill correctly. The TM Forum, the telecom industry's main standards body for OSS/BSS, publishes revenue assurance frameworks specifically because this problem is endemic.
Telecom data has a shelf life measured in seconds for network state and days for billing state. An HLR entry showing a subscriber registered on a network tower they left twenty minutes ago causes dropped call routing and failed SMS delivery.
Timeliness is usually measured as data latency: the gap between when an event occurs and when it's reflected in the system of record.
A stale HSS entry that still shows a subscriber as active on a decommissioned network slice is a timeliness failure with direct customer experiencecustomer experienceThe overall perception a customer forms of your brand across every interaction, from first touch to post-purchase support.Voir la définition complète → impact.
Do the same facts match across systems? This is the classic telecom pain point: a subscriber's address in CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.Voir la définition complète → says "Apt 4B" while billing has "Apartment 4-B" and the field in the fraud-detection system is blank. None of these are technically wrong, but they can't be reliably joined or compared.
Consistency problems compound at merger integration. When one operator acquires another (for example, T-Mobile's acquisition of Sprint in the US, completed 2020), reconciling two full BSS/OSS stacks, each with its own MSISDN formatting conventions, address schemas, and CDR structures, is a multi-year data governancedata governanceData governance is the set of policies, roles, and processes that ensure data is accurate, secure, well-defined, and used responsibly across an organization.Voir la définition complète → project, not a weekend migration.
A simple consistency check in practice:
SELECT crm.msisdn, crm.address, billing.address
FROM crm
JOIN billing ON crm.msisdn = billing.msisdn
WHERE crm.address <> billing.address;Every row returned is a consistency exception requiring reconciliation logic (fuzzy matching, standardized address libraries, or manual review).
Is each real-world entity represented exactly once? Duplicate MSISDNs are the canonical telecom uniqueness failure: the same number appearing under two different customer IDs, often after a SIM swap, porting event, or system migration that failed to deduplicate.
Duplicates distort nearly every downstream metric: churn ratechurn rateChurn rate is the percentage of customers or revenue lost over a period. It measures how fast a business loses its existing customer base.Voir la définition complète →, ARPU (Average Revenue Per User), and active subscriber counts. If an operator reports 30 million subscribers but 900,000 MSISDNs are duplicated across systems, the real base is closer to 29.1 million, a 3 percent overstatement that misleads both internal planning and, if disclosed externally, investors and regulators.
$$\text{Duplicate rate} = \frac{\text{duplicate MSISDN records}}{\text{total MSISDN records}} \times 100 = \frac{900{,}000}{30{,}000{,}000} = 3\%$$
Vérification des acquis
1. Why do telecom operators tend to accumulate multiple, inconsistent records for the same subscriber across systems?
2. Why is the MSISDN's role as a join key across systems particularly risky for data quality?
3. A subscriber's billing address is correctly formatted and spelled consistently in one system, but it no longer reflects where they actually live. Which underlying data quality issue does this best illustrate?
4. Select ALL correct answers about why fragmented subscriber records create real financial and regulatory consequences for telecom operators.
Sélectionnez toutes les réponses correctes.
5. Select ALL correct answers about the systems and datasets described as sources of telecom data quality challenges.
Sélectionnez toutes les réponses correctes.
Operators typically track a Data Quality Score (DQS), a composite weighted across the five dimensions, reported at the dataset level (for example, separately for CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.Voir la définition complète →, billing, and network inventory).
Rough industry-cited benchmarks (treat as estimates, not audited figures, since no universal regulator mandates disclosure):
In Europe, GDPR (General Data Protection Regulation) Article 5(1)(d) explicitly requires personal data to be "accurate and, where necessary, kept up to date," which makes data qualitydata qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.Voir la définition complète → a direct legal compliance matter, not just an operational nicety. In the US, the FCC's CPNI (Customer Proprietary Network Information) rules similarly require operators to maintain accurate safeguards around subscriber data, indirectly reinforcing quality discipline.
🎬 [VIDEO: "What is Data QualityData QualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.Voir la définition complète →? (And Why It Matters)" - youtube.com - search for TM Forum or IBM Data QualityData QualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.Voir la définition complète → explainer channels for a concise walkthrough of the core dimensions applied to enterprise data]
A practical audit workflow:
1. Pick the dataset (CRMCRMCustomer Relationship Management: software and strategy to manage and analyse customer interactions throughout their lifecycle.Voir la définition complète →, billing, HLR/HSS extract).
2. Score each of the five dimensions with a simple percentage metric.
3. Weight by business criticality (billing accuracy matters more than a marketing opt-in flag).
4. Set remediation thresholds and owners (data governancedata governanceData governance is the set of policies, roles, and processes that ensure data is accurate, secure, well-defined, and used responsibly across an organization.Voir la définition complète → teams, often reporting into a Chief Data Officer role, increasingly common at major operators like Vodafone and AT&T).