Quick answer: In messaging and voice, "aggregator" describes a wholesaler that stands between a platform and the mobile or fixed-line operators that terminate your traffic. A CPaaS vendor buys termination from that wholesaler or connects to operators through its own switch; the difference shows up in your delivery logs as route depth — however many hands pass each message in the middle. "Carrier-of-record posture" is the readable version of that question: which party answers for the route when a destination degrades, and how much of the path you can see. You can measure both yourself, with lookup dips and your own delivery records.
Direct interconnect vs. aggregator: counting the hops
A message or call has a path. Your API call reaches the platform, the platform's route decision reaches a network termination point, and the operator delivers. Each wholesale party between the platform and the operator is a hop, and its wholesale margin, queueing discipline, and failure window sit on your traffic.
A hop-neutral way to read a vendor's routing claim is to ask it how many wholesale parties a typical destination country crosses in its routing, and whether the count varies by country. "4,800 carrier relationships" is a relationship count, not a route-depth answer — the question is how many middle parties a specific country actually goes through, not the size of the Rolodex behind it. Three shapes cover nearly the whole market:
- Direct-to-operator termination. The platform holds its own switch and interconnects directly with the operators that terminate your traffic. No wholesale middle party sits between the platform and the route.
- An owned wholesale switch, resold. The vendor holds its own switch with direct operator interconnects — then resells termination over that switch to other platforms. Your traffic crosses the resale counter, but the underlying network is the vendor's own. Two parties touch a route: the wholesaler's own switch and the operator.
- Aggregated termination. The platform buys termination from one or more wholesale aggregators rather than terminating routes itself. Each reseller hop adds a party whose margin, route discipline, and failure handling enter your delivery, and the aggregator adds its own hops in turn.
None of the three is inherently good or bad — a well-run aggregation layer can outperform a poorly-run owned switch. The shape does, however, change what you can observe when something goes wrong, and that observation is the part buyers actually test.
What carrier-of-record posture means — a tenant-readable question, not a platform mandate
"Carrier of record" is an FCC regulatory filing category in the US voice market: the carrier holding the operating authority behind a number. Used loosely, vendors extend the phrase to mean "we answer for the route." In CPaaS vendor comparisons it stays a claim, not a regulated label — and treating it as one is the mistake this post exists to prevent.
The useful version of the question is not "is this vendor a carrier of record" but "what is this vendor's posture toward carrying its routes":
- Who terminates. Does the vendor terminate routes itself, buy from one wholesaler, or buy from many depending on the destination?
- Who answers for a failure. When a route degrades, whose NOC owns the fix — the vendor's own switch records, or a support case that re-enters a wholesale chain the vendor does not control?
- What's visible to you. Which per-message and per-call routing attributes appear in your delivery records — and which ones the vendor cannot see either, because the aggregator holds them?
Sellers sometimes present the question as a mandate — "must be carrier-of-record" — as if it were a compliance requirement. It is not. It is a routing posture a buyer can read, weigh, and decide on in the same way it weighs per-country pricing. A tenant owns the decision; the platform's job is to make the posture readable, not to mandate an answer.
Latency, visibility, and dispute resolution per shape
The three shapes behave differently along three axes, and the differences are concrete enough to log:
- Latency. Each resale counter adds queueing: the middle party batches, retries, and throttles before the operator ever sees the traffic. On a 2FA or OTP workload, the difference reads in the gap between send and handset delivery. On a voice workload, it reads as an extra set of media and signaling gateways between your call and the called party — each one a dBm of jitter you did not cause.
- Visibility. A single-hop path gives the platform's delivery records a direct view of route, status, and failure: what your logs say, the operator actually returned. Each wholesale counter interposes its own logs — and a dispute over whose record is the real one becomes the dispute-resolution tax. Multi-hop disputes typically resolve as "the upstream returned DELIVRD," with no visibility into whether the terminal operator delivered.
- Dispute resolution. When a route degrades, ownership decides who holds the evidence. An owned switch holds the route decision, the per-attempt failover, and the outcome in one place — the route re-picks from its own records without a support case. An aggregated path turns the same event into a chain of support cases, each party pointing at the next. Buyers feel the difference as time-to-answer on a degraded destination.
The jump in latency and dispute-resolution quality between one- and two-hop paths is the reason the carrier-of-record question matters for OTP and alerting workloads specifically: a failure that takes hours under aggregation resolves in seconds with a single route owner — while a delayed OTP fails closed against the customer, not open.
How buyers test posture themselves
You do not need to trust a posture claim, including this one. Three checks read a platform's actual behavior against what the platform says about itself:
- HLR lookups. Run a number lookup dip against your own destinations. The carrier block — mobile country and network codes, ported flag, line type, roaming state — tells you which operator terminates the number you are sending to. Compare that against what the platform's delivery records say about the route for the same destination: a platform routing directly should show a path consistent with the operator the HLR dip names. A platform whose records cannot line up with the lookup is carrying multiple hops it cannot itself name.
- Mobile-termination-rate (MTR) checks. Pull the published per-country termination rate for a destination, then compare what your per-message or per-minute pricing actually bills at for that destination. A reseller layer hides its margin between the published termination rate and your invoice. A direct-onwards path should show a consistent margin across countries — and a margin shape that varies oddly country-to-country is an invitation to ask which party is eating the difference.
- Delivery-log inspection. Your message and call logs carry per-attempt status, provider response, and disposition. Slice them by destination country and failure class: a country that degrades, then recovers, tells you who re-picked the route. On Orbit, the message sends and call legs visible in your logs are the per-attempt records the routing layer wrote — the same records the delivery receipts explainer reads. If a destination degrades and the route re-picks without your intervention, the records you read back are the platform's own switch deciding. If a degraded destination needs a ticket first, someone else's switch owns the decision — and you have read the posture.
None of these checks requires privileged access. All three run on data any buyer generates within an hour of having the account.
Where Devotel Orbit's posture sits
Devotel Orbit routes outbound voice and SMS over Devotel's own wholesale softswitch, which interconnects directly with 500+ carriers, rather than reselling through a third-party aggregator layer. The failover behavior the compare page describes — a degraded route re-picks a better route per attempt, from the switch's own records — is the same single-switch posture expressed against live traffic. The carrier-of-record compare page states the posture and walks the failover behavior; the carrier-vs-reseller framework lays out the neutral four-question read next to what six vendors (including Orbit) publicly state about their own networks. And because the posture is a claim you can test, the three checks above are the ones we invite: run the HLR dip, read the MTR math in your price list, and slice your own delivery records.
Further reading
- Carrier vs. reseller in CPaaS — the neutral framework for reading a "we own the network" claim, plus what six vendors publicly state about their own networks.
- Carrier-of-record compare page — Orbit's stated posture: the failover behavior, the pricing math on one account, and the capability inventory beside it.
- Phone-number lookup and SIM-swap pre-send — the HLR dip that maps the destination half of the check.
- Carrier-of-record selection criteria — the buyer checklist this posture answers to.
- Delivery receipts, carrier delay, and what a DLR actually guarantees — how to read the per-attempt records the checks rely on.
Frequently asked questions
What is an aggregator in CPaaS?
A wholesaler that stands between a platform and the operators that terminate its traffic. The platform buys termination from the wholesaler rather than connecting directly; each aggregator hop adds a party whose margin and failure windows ride on your delivery.
Does "carrier of record" mean a vendor holds FCC carrier status?
Not in common CPaaS usage. In US telecom, "carrier of record" names the FCC-filed entity holding a number; in vendor marketing it usually means "we answer for the route." Treat the phrase as a posture claim, not a regulatory label, and ask which party terminates which routes.
Do I need a carrier-of-record posture to run compliant traffic?
No. Posture is a routing and visibility question, not a compliance requirement. US sender registration (10DLC, toll-free verification), consent discipline, quiet hours, and opt-out handling are the compliance obligations, and they run identically over any routing shape. The tenant owns those controls; the routing posture decides how visible the route behind them is.
Can an aggregator offer better delivery than an owned switch?
Yes — ownership does not substitute for route quality. A well-run aggregator with deep failover across many upstreams can outperform a poorly-run owned switch, which is why the buyer-side checks (HLR dips, MTR comparisons, delivery-log slicing) matter more than the claim itself.
Does Orbit hold carrier-of-record status?
Orbit's stated posture is narrower and verifiable: outbound voice and SMS terminate over Devotel's own wholesale softswitch, with direct interconnects to 500+ carriers, rather than a resold aggregator hop. The compare page publishes the posture; the framework gives you the four questions to test it against.