HLR lookup product economics for wholesale operators

Operators hold the definitive view of reachability and routing. Packaging that view as an API around HLR queries converts a sunk signalling asset into recurring cash. The commercial question is not “can we expose data” but “which outcomes do we price for.” Done correctly, an anonymised, low-latency HLR lookup product can serve A2P aggregators, KYC vendors, and fraud stacks—without leaking PII—while fitting standard wholesale settlement cycles and preserving SMS and voice economics.

Demand, use cases, and volume profiles

The buying logic is simple. Enterprises and aggregators want to know whether a number can be reached, whether it has moved, and whether a transaction should proceed. They do not usually need deep subscriber attributes. They need freshness, coverage, and latency they can model in their own routing engines.

Primary demand comes from A2P SMS aggregators, call verification providers, KYC and AML vendors, fraud and chargeback platforms, and enterprise contact centres seeking pre-dial hygiene. The purchased outcomes are narrow: reachability, MNP status, roaming state, line type, and basic service state. That is enough to route, rate, suppress, or retry traffic without turning the product into a subscriber-data service.

Volumes are predictable at portfolio level but uneven at the minute level. OTP windows create sharp morning, lunch, and evening spikes. Contact-centre traffic follows working hours and campaign calendars. Retail, tax, insurance renewal, and public-service periods produce seasonal uplift. In a mature wholesale portfolio, 60–90% of billable traffic typically comes from a small group of aggregators, with long-tail enterprise accounts contributing higher unit price but more support overhead.

For buyers, the value case is avoidance. A fraud platform suppresses a high-risk transaction before it incurs chargeback cost. An A2P aggregator reduces failed delivery attempts and improves campaign routing. A contact centre avoids dialling dead or unreachable numbers. For the host operator, the revenue line is clean because it monetises pre-transaction hygiene rather than content delivery. Properly priced, it should not cannibalise SMS or voice volumes; it should make paid traffic more efficient.

Latency requirements also support a wholesale product rather than bespoke systems integration. Sub-500 ms p95 is usually sufficient for OTP-adjacent decisioning, provided timeout behaviour is clear. Dialler and batch hygiene use cases can tolerate roughly 800 ms, especially where caching and asynchronous refreshes are permitted. The decisive issue is not the fastest possible query. It is a stable latency envelope that buyers can encode into retry and fallback rules.

A Tier-1 MNO, Western Europe, ~25M subscribers, tested the product shape through two wholesale aggregators before allowing direct enterprise demand. The initial field set was limited to reachable, ported flag, roaming yes or no, and coarse operator bucket. That reduced legal review time and gave finance a clean monthly unit-rate model. Direct enterprise contracts followed only after partner audit routines, use-case approvals, and query caps had stabilised.

Product design and governance: expose utility, not identity

The commercial product should be designed around utility, not identity. A privacy-minimal response schema can still carry enough value for routing and fraud decisions. Typical external responses should include boolean reachable, ported flag, coarse operator bucket, roaming yes or no, and line type. Basic service state may be offered where local rules allow it. Raw IMSI should never leave the operator domain, and ported-to MNC/MCC should not be exposed verbatim to external parties.

Freshness policy is where product and settlement meet. TTLs should align with market portability cadence, commonly 24–72 hours, with explicit cache headers returned to buyers. Operators should publish a delta window for stale-answer disputes. If a number ported within that window and the response caused a demonstrable failed decision, the buyer knows the refund path. Without that rule, every portability edge case becomes an invoice argument.

Jurisdiction shaping is non-negotiable. Field suppression by market, buyer category, and approved use case should be configured before launch, not handled manually after a complaint. ePrivacy, GDPR, telecom secrecy, and national numbering rules do not map cleanly across borders. The product therefore needs a policy layer that can suppress roaming, line type, or portability detail where required. Access logs must be retained for audit, and field-set changes should pass an internal ethics and regulatory gate.

Buyer governance should be contractual and technical. Integrators need strict KYC, per-use-case approval, signed acceptable-use schedules, query rate caps by GT or API key, and immediate revocation rights for breach. Abuse controls should include SS7 firewall rules to prevent grey-route discovery, request fingerprinting to detect enumeration, and synthetic test numbers to identify scraping intent. Quarterly compliance attestations are not paperwork. They are the evidence base that protects the product when regulators ask how number intelligence is controlled.

Unit economics, pricing bands, and settlement mechanics

The margin profile is attractive only if price discipline survives aggregator negotiation. The cleanest model is a per-lookup fee with volume tiers. Revenue share on net billings can work where the buyer’s resale base is fragmented or opaque, but it weakens auditability unless floor rates and reporting obligations are explicit. Hybrid structures—minimum monthly commits, step-down tiers, and usage true-ups—often give finance the best balance of volume certainty and upside.

Costs sit in two places: signalling and disputes. Live queries over MAP through STP or hub arrangements carry a real transit cost, while cache hits have near-zero marginal cost. Cache policy therefore becomes a commercial control, not only an engineering choice. High-velocity MSISDNs in OTP flows can produce strong cache hit rates if TTLs are tuned carefully. Over-aggressive caching, however, pushes stale MNP disputes into settlement.

Wholesale sell price (aggregators)

$0.0015–$0.004 per lookup, tiered by monthly volume

Direct enterprise sell price

$0.004–$0.012 per lookup with compliance obligations

Signalling/transit cost per live query

$0.0002–$0.0007 (SS7 MAP via STP/hub); cache hit ≈ near-zero marginal

Typical cache hit ratio (tuned TTL)

35–70% depending on market MNP churn and buyer mix

Latency SLA target

p95 200–500 ms; timeout budget 1.0–1.5 s with cached fallback

Refund/dispute rate (stale MNP/timeouts)

0.5–2.0% of billable volume; cap via delta windows

Settlement cadence

Monthly invoicing, Net 30–45; 10–20% rolling holdback against disputes

Expected QPS envelope (early scale)

500–5,000 QPS with OTP peaks; size for 3× burst

Settlement rules should be written before the first billable query. Monthly invoicing with Net 30–45 terms is conventional for established aggregators. New integrators should carry prepaid balances or tight credit limits until volume quality is proven. A 10–20% rolling holdback protects both sides from timeout disputes, stale portability claims, and reconciliation lag. Service credits should attach to SLA misses, not to raw volume, otherwise one outage can erase the margin of an otherwise profitable month.

The CFO-level metric is contribution after transit, cache, refunds, and support. A portfolio that sells mostly to aggregators may show lower unit price but simpler collections. A direct enterprise book may triple the sell price per lookup while adding onboarding review, security questionnaires, and bespoke reporting. Both can be profitable. Mixing them without separate margin views usually hides the cost of complexity.

Delivery models and SLA engineering

The technical architecture must support the commercial promise. Queries typically originate through MAP SRI or ATI via owned GTs or a contracted signalling hub. Where policy permits, Diameter or HLR-HSS interop can support equivalent state checks, but external exposure should remain behind the same API and policy layer. Buyers should not need to know which signalling path resolved a number.

Topology matters because p95 and p99 tails are what buyers feel. Regional PoPs with local STP adjacency cut round-trip time and reduce dependence on a single signalling path. Active-active service across regions limits invoice-impacting outages. When the HLR or hub is congested, the platform should fast-fail to a cached answer where policy allows, flag the response as cached, and preserve an audit trail for settlement.

Caching requires more discipline than a global TTL. Per-MSIDN TTLs should reflect market MNP behaviour, observed query velocity, and buyer use case. Negative-cache invalids should be short, because invalid or unreachable state can change quickly in fraud and prepaid contexts. Cached fallback should be visible to the buyer through response metadata, but not so verbose that it reveals internal routing or signalling topology.

The API layer should be conventional and controlled: REST/JSON, signed requests, idempotency keys, per-key rate plans, partner IP whitelisting, and asynchronous callbacks for slow-path completions. Spend controls should connect to OCS and BSS/OSS so credit stops, caps, and invoice records match the commercial contract. Observability should report per-partner p95 and p99, timeout and error codes, MNP delta counters, cache hit ratio, and real-time billable volume.

Security is part of the SLA. Signalling firewall rules, GT reputation monitoring, credential rotation, and routine access audits reduce the chance that a number-intelligence product becomes an enumeration tool. Partner support teams also need clear runbooks. A commercial dispute over a timeout cannot wait for an engineering post-mortem that lacks request IDs, timestamps, cache status, and signalling outcome.

Contracts, revenue share, and market routes

Number intelligence should be contracted as a wholesale product, not as a string of bespoke data-access exceptions. The strongest constructs are fixed per-lookup buy rates for aggregators, revenue share on net billings with floor rates, or hybrid minimum commits with step-down tiers. Each structure needs a clear billable event definition, timeout treatment, stale-answer policy, refund cap, and audit process.

Settlement protection should be explicit. Rolling credit limits and prepayments are appropriate for new integrators. Established wholesale buyers can move to monthly invoicing once payment behaviour is proven. Holdbacks of 10–20% should be tied to defined dispute categories, not kept as a vague reserve. SLA-linked credits should compensate for measurable service failure, but they should not become volume-linked rebates that undermine the published tariff.

Distribution can remain concentrated. Two to four aggregator logos often cover more than 70% of addressable enterprise demand because they already serve banks, marketplaces, logistics firms, and contact-centre platforms. Direct enterprise routes are useful where the buyer has compliance maturity and enough volume to justify review. Otherwise they can consume legal, security, and support capacity faster than they add margin.

For an MVNE servicing 12+ tenants in EMEA, the more strategic route was bundling number intelligence into tenant fraud and onboarding tools. The API was included as a priced platform option rather than sold only as a standalone wholesale feed. Tenant churn improved because fraud tooling became part of the operating stack, not an external bolt-on that could be replaced at renewal. The contract still separated field rights, geography, audit rights, and misuse clawbacks by tenant.

The operating lesson is narrowness. Start with a small field set, two or three aggregator partners, conservative TTLs, and unambiguous settlement terms. Expand only when audit logs, cache performance, refund rates, and buyer behaviour are visible. HLR-driven number intelligence is a credible wholesale line item with controllable risk, clear SLAs, and repeatable unit economics. The operators that price the outcome, govern the buyer, and reconcile disputes cleanly will keep most of the margin.