What a grey route is
Mobile operators charge for terminating application traffic on their networks: the codes, alerts and campaigns businesses send, known as A2P traffic. A direct, or official, route pays that charge through an agreed connection with the operator or its appointed aggregator. A grey route avoids it: banks of consumer SIM cards, roaming loopholes, unmetered interconnects between operators. It is cheap until the operator notices, and then traffic is filtered or blocked, sometimes together with the sender names that were used on it.
You will also hear of white routes for official ones and black routes for outright fraud. The labels are loose, and "direct" is used just as loosely: few providers hold their own connection to every operator they reach, and international traffic commonly passes through one or more aggregators. What matters is not how many hands a message passes through but whether the last hop, the one into the operator, is a connection the operator accepts for application traffic.
How grey routes work
The mechanisms differ, but they share one trick: application traffic presented as something cheaper.
- SIM farms. Racks of devices holding consumer SIM cards send business messages as if they were personal texts, at retail bundle prices. Recipients see a random local mobile number, often a different one each time. Consumer SIM terms usually forbid this, and operators hunt the cards down.
- Person-to-person interconnects. Operators exchange their subscribers' personal texts under bilateral agreements, often at little or no cost. Application traffic slipped into those channels travels cheaply until the receiving operator's filters recognise it.
- Roaming loopholes. Traffic is dressed up as coming from a subscriber roaming abroad or from a foreign operator, to fall under a cheaper tariff.
- Spoofed signalling. At the far end, messages are injected into signalling networks under a false identity. That is no longer arbitrage; it is fraud.
Operators answer with SMS firewalls that inspect incoming traffic and block what does not match its declared origin. So a grey route is never stable: it works, gets detected, closes, and the reseller moves the traffic elsewhere, often without telling its customers. Whether a given route is also illegal depends on the country; at the very least it breaks the contracts it rides on, and the operator can shut it without notice.
What a grey route costs you
- Codes that do not arrive, or arrive after they expired. A one-time code is the message least able to wait.
- Reports you cannot trust. When receipts are generated by an intermediary instead of the handset, the report stops measuring anything, and you cannot tell which messages to resend.
- Your brand on a stranger's number. Where the country delivers names, a grey route replaces yours with a random mobile number, which is exactly what recipients have learned to distrust.
- Content in unknown hands. A message relayed through consumer devices passes through equipment run by people you have never heard of.
- Collateral blocking. When the operator closes the route, names and content seen on it can be blocked too, including yours.
Symptoms
- Delivery reports that are always perfect. Every message delivered, within the same second, including to numbers you know are switched off.
- Sender names quietly replaced by random mobile numbers, in countries that deliver names, with nothing in the report to say so.
- Messages that arrive hours late or in bursts, or not at all on one operator.
- Text that changes on the way: accented letters replaced, long messages arriving as separate texts.
- Quality that decays: a good first week, then a slide as routes are detected and swapped.
- Prices far below the operator's own A2P rate for a destination known to be expensive.
The handset test
A report can be fabricated; a phone on a desk cannot. A test key, ours included, sends nothing, so this test needs live messages. For each destination that matters to you:
- Gather three to five real handsets on different operators in the country: colleagues, customers, test SIMs.
- Send ten messages to each, spread over a day, with a unique code in every text. Vary the content: a short plain message, one with accented letters or another script, a two-part message, one with a link on your own domain.
- Record for each message: did it arrive, how many seconds after the API call, which sender was displayed, was the text intact.
- Switch one handset off for an hour and send to it. Note what the delivery report says while it is off, and when that changes.
- Compare handset arrival times with the delivered timestamps in the report.
Send the same messages through two or three providers at the same moment and the comparison does itself. Repeat the test every few months and whenever results shift: a route that passed in spring can be swapped by summer.
Reading the results
An honest route shows delivered only after the phone has the message, keeps the switched-off handset at sent until it comes back, and displays the sender the country's rule predicts. Check the country page first: where a destination gives carrier acknowledgement only, sent is the final positive state and delivered is not expected.
| What you observe | What it suggests |
|---|---|
| Delivered for a phone that was switched off | Receipts generated by an intermediary, not by the handset |
| Delivered timestamps earlier than arrival on the phone | The same |
| A different mobile number as sender each time, where names are delivered | Consumer SIM cards at the last hop |
| One operator never receives while the others do | The route does not reach that network officially |
| Accented letters lost, or long texts arriving in pieces | The message was re-sent along the way, not passed through |
| Arrival times ranging from seconds to hours | Traffic queuing on a constrained route |
A numeric sender in a country that delivers names is not proof on its own: a legitimate platform falls back to a number when a carrier refuses a name. The difference is disclosure. On SMSMeteor a fallback is flagged in the report and the from field shows the sender actually used; the delivery report gives per-message timestamps to set against your handsets, and route incidents are posted on the status page. Run the test on us as on anyone else.
Questions to ask a provider
- Which destinations return handset receipts, and which carrier acknowledgement only?
- What sender will recipients see in each country, and what happens when a carrier refuses a name?
- How are routes chosen, how often are they reviewed, and how will I know when one changes?
- Is there a cheaper tier, and what exactly is different about it?
- Where are incidents published, and how far back does the history go?
Vague answers are an answer. Ours are public: the receipt type and sender rule on every country page, one route per destination chosen for measured delivery and reviewed daily, no cheaper tier, and an incident history on the status page. How routes are chosen and swapped describes the process.
Why price alone misleads
The useful figure is the cost per message that actually arrived. Take two hypothetical routes: one at 3 cents a message that lands nine in ten, one at 2.5 cents that lands seven in ten. Per message delivered, the first costs about 3.3 cents and the second about 3.6, before counting the customers who never got their code. A route with fabricated receipts hides even that arithmetic: every message looks delivered, so the losses surface as abandoned sign-ups and support tickets rather than in the report.
We keep one route per destination, chosen on measured delivery, and publish one price for it on the rate card; there is no cheaper tier that loses messages. Failed messages reached a carrier and are billed, rejected ones never are, and the report says which is which. For the wider checks worth running before a first deposit, see how to vet a provider.
Something unclear or out of date? Tell us on the support page; corrections are folded back into the guide and the update date changes.
