What a UK text message has to satisfy before it sends

Each row below is a rule we apply to UK traffic, with where it comes from. The sender, registration, content and opt-out positions are our own UK country record, verified against Ofcom numbering rules and UK operator sender policy on 12 October 2026 and updated as operators change their handling. Length and price come from the character set reference and the coverage page, both of which are published in full rather than summarised here.

Two things the table deliberately does not say. It does not promise a delivery rate: the only delivery claim we publish is structural, that messages are submitted to the UK networks and receipts are reported back. And it does not claim a direct connection to every network, because most UK traffic is carried through hubs, including the networks' own.

A UK recipient sees either an alphanumeric sender ID of up to 11 characters, a numeric sender, or a short code. The alphanumeric form is what makes a message look like it came from a business rather than a random number, and it is the common choice for transactional and notification traffic. It is also the form with the most setup behind it, because identity in the UK is protected by default rather than by a registry.

There is no national sender register to join, and no campaign approval to wait for. What exists instead is a set of reserved and restricted sender IDs maintained by the UK networks, plus per-network requirements where an operator wants an originator registered before it will carry it. Before you launch we check your ID against those lists, and register it where a network requires that, so the first message is not the test. If your ID collides with a reserved one, a network can substitute a different sender or reject the message, and neither outcome is visible in your own logs unless you are watching receipts.

Because UK numbers are portable, the prefix no longer tells you which network a handset was issued on, and after porting it may not tell you where it lives now. Routing decisions based on number ranges are therefore approximate by nature, which is one of the reasons we do not publish per-network delivery percentages that would rest on that approximation.

Registration is the same for a local and an international origin in the UK, which is unusual. What we ask for is your company name, website and country, the industry you are in, a message example, your opt-out method and how it is set up, and the relationship between your company and the sender ID. Turnaround is typically one working day. The full requirements for every market we serve, including what changes if you send from outside the country you are sending to, are in the country restrictions reference.

The restricted categories for UK traffic are gambling, phishing, delivery phishing, other scams, regulatory content and explicit content. Gambling is the one with a licensing rule rather than a taste rule: it is accepted for licensed operators, with raffle traffic excepted, so a competition that is legally a raffle can pass where a betting promotion cannot. Explicit content is judged by the receiving network, which means the same message can be treated differently by two operators.

Separately from content categories, some senders are protected. Where a brand name is protected, the network may want a letter of authorisation from the brand before it will accept traffic carrying that name as a sender ID. This catches resellers and agencies more often than direct customers: the person sending is not the person whose name is on the message, and the network has no way to know they are entitled to use it.

The category list is per market and changes with operator policy rather than with law. The full list for every country, with the detail behind each category, is in the country restrictions reference; our own sending-side rules, including how we treat sender IDs and originator registration, are in the anti-spam policy.

UK networks do not make an opt-out keyword mandatory the way US carriers make STOP mandatory. That is a network rule, not the legal position, and treating the two as the same thing is how senders end up compliant with the operator and not with the regulator.

Under PECR, marketing by text needs consent or the soft opt-in, and every marketing message needs a working way to opt out. Service messages — a one-time code, an order update, an appointment reminder — are not marketing, so they do not need marketing consent, though UK GDPR still applies to the data behind them. That split is a decision about what a message is for, made at the point you build it, not an assessment to make afterwards. The UK SMS compliance page sets out the consent position and the soft opt-in in detail, and the transactional versus marketing comparison covers where the boundary falls in practice.

What we do on the platform side is keep the two classes separable: message type travels with the message, so whether a given text was transactional or marketing is a property of the record rather than a question for the legal team. Two-way messaging is available on UK traffic, which means an opt-out or a reply can be received and acted on rather than only sent into a void.

A single SMS part is not 160 characters in every encoding. It is 160 characters of GSM 7-bit text, which is the alphabet that covers basic Latin, most Western European accents and a set of Greek capitals, and 70 characters of UCS-2 Unicode for everything else. The moment a message needs more than one part, each part loses space to the header that lets the handset reassemble them: 153 characters of GSM text per part, 67 of Unicode.

A few GSM characters are encoded as an escape plus the character, so they cost two positions: ^, {, }, \, [, ], ~, | and €. A message that measures 160 characters and contains one of them spills into a second part. The full tables, including what happens at each concatenation boundary, are on the SMS length reference, with the character sets themselves on the GSM and ASCII pages.

Each part is billed. UK destinations are 10.71 credits per part on the coverage page, so a message split into three parts costs three times the single-part rate, and the split is decided by encoding and length rather than by anything in the message text you can see. Our API concatenates up to nine parts; anything sent beyond that limit is discarded by the gateway without an error, which is why our Zapier action refuses a message that needs a fourth part instead of trimming it.