> ## Documentation Index
> Fetch the complete documentation index at: https://docs.robase.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Segments & encoding

> How SMS is billed — and why emoji costs more.

SMS is billed **per segment**, not per message. Understanding segments saves you money on long or non-Latin sends.

## GSM-7 vs UCS-2

SMS has two encodings:

| Encoding  | Character set                            | Per-segment cap (single) | Per-segment cap (concatenated) |
| --------- | ---------------------------------------- | ------------------------ | ------------------------------ |
| **GSM-7** | Latin letters, digits, basic punctuation | **160 chars**            | **153 chars**                  |
| **UCS-2** | Unicode (emoji, Arabic, Chinese, …)      | **70 chars**             | **67 chars**                   |

We detect the encoding automatically: the moment any character in your body falls outside the GSM-7 alphabet, the entire message switches to UCS-2 — you don't get to send 160 Latin chars plus one emoji as one segment.

## Examples

| Body                                           | Length | Encoding                  | Segments        |
| ---------------------------------------------- | ------ | ------------------------- | --------------- |
| `"Hello Chidi, your ride arrives in 3 min."`   | 42     | GSM-7                     | **1**           |
| 160 chars of Latin text                        | 160    | GSM-7                     | **1**           |
| 161 chars of Latin text                        | 161    | GSM-7                     | **2** (99 + 62) |
| `"Your ride is here! 🚗"`                      | 20     | **UCS-2** (because of 🚗) | **1**           |
| `"Your ride is here! 🚗 " + repeated 60 chars` | 80     | UCS-2                     | **2**           |

Preview the segment count before sending using [`POST /v1/sms-templates/:id/render`](/api-reference/sms-templates/render) for templates, or compute it locally:

```typescript theme={null}
function segments(body: string): { encoding: 'GSM7' | 'UCS2'; count: number } {
  const hasUnicode = /[^\x00-\x7F]/.test(body);
  const len = [...body].length;
  if (!hasUnicode) {
    return { encoding: 'GSM7', count: len <= 160 ? 1 : Math.ceil(len / 153) };
  }
  return { encoding: 'UCS2', count: len <= 70 ? 1 : Math.ceil(len / 67) };
}
```

## Why the count drops on concatenation

Multi-segment SMS uses a small header on each part to tell the phone how to reassemble the message. That header eats 7 chars in GSM-7 and 3 characters in UCS-2 — so a "160-char" segment really only has 153 of your characters to work with once it's concatenated.

## Cost implications

A friendly emoji can 2x or 3x your cost if you weren't expecting UCS-2:

```
"Thanks for signing up!"                  GSM-7 → 1 segment @ ₦4.00 = ₦4.00
"Thanks for signing up! 🎉"               UCS-2 → 1 segment @ ₦4.00 = ₦4.00
"Thanks for signing up! Visit ... 🎉"     UCS-2 → 2 segments @ ₦4.00 = ₦8.00  (same text, no emoji would be GSM-7 → 1 segment)
```

## Best practices

<CardGroup cols={2}>
  <Card title="Keep transactional under 160 GSM-7">
    Tight, no emoji, no fancy quotes. One segment per send.
  </Card>

  <Card title="Ascii-ify URLs">
    Don't use URL shorteners with non-Latin chars — those force UCS-2.
  </Card>

  <Card title="Watch smart quotes">
    Word / Notes replace `"` with `"` and `'` with `'` — both UCS-2.
  </Card>

  <Card title="Preview campaign copy">
    Use the template render endpoint in CI to assert segment count ≤ 1.
  </Card>
</CardGroup>

## Handy bytes worth avoiding

These tiny Unicode characters force the whole body to UCS-2:

* `"` (U+201C) and `"` (U+201D) — curly quotes
* `'` (U+2019) — typographic apostrophe
* `…` (U+2026) — ellipsis
* `–` (U+2013) and `—` (U+2014) — en/em dashes
* Non-breaking space (U+00A0)

Replace with `"`, `'`, `...`, `-`, `--`, and plain space respectively.