An emoji in an SMS forces Unicode encoding, cutting the message capacity from 160 to 70 characters, and some emoji take two Unicode units, so a single symbol can add a whole segment.
No emoji is part of the GSM-7 alphabet, so one emoji switches the whole message to UCS-2. Most emoji also sit beyond the range UCS-2 was designed for and travel as a surrogate pair, two 16-bit units, so they take two of the 70 places. Some take more: a skin-tone variant or a flag is made of two such characters, four units, and combined emoji such as families join several with invisible joiners. A few older symbols, such as the heart, take one unit, or two when followed by the invisible selector that requests emoji-style display.
The cost effect is large because two things happen at once: capacity falls from 160 to 70, and the emoji itself counts double. A hypothetical 150-character marketing text is one GSM-7 segment. Add one emoji that takes two units and it becomes 152 units of UCS-2: three segments, 67 + 67 + 18, three times the price.
Rendering is the other risk. The picture the recipient sees is drawn by their phone, so the same emoji can look different from one handset to another, and an older phone may show an empty box. Emoji can be worth it in a short promotional message where the visual matters. They are never worth it in codes, alerts or receipts, and they are best avoided in campaigns to countries where the route replaces or filters them.
Three mistakes are worth avoiding:
- Counting an emoji as one character because it looks like one.
- Adding a decorative emoji to a text written to fit one segment, which can triple its cost.
- Using emoji in one-time codes, where they add cost and nothing else.
At SMSMeteor: the composer names the character that forced Unicode, an emoji included, and shows the new encoding and segment count before anything is sent. The API sends text exactly as submitted and returns encoding and segments, so a test key shows the effect of an emoji without sending or billing anything.
