It is the most common surprise on a first invoice: a campaign you costed at 1,000 messages consumed 2,000 credits. It is not a billing error, it is how SMS works.
Two alphabets, two limits
An SMS does not carry free text, it carries text encoded in one of two alphabets.
GSM-7 is the standard alphabet. It contains letters without diacritics, digits and the usual punctuation. A single message holds 160 characters.
UCS-2 is the alphabet used as soon as any character outside GSM-7 appears. A single message then holds only 70 characters.
Romanian diacritics (ă, â, î, ș, ț) are not part of GSM-7. One of them, anywhere in the text, is enough to switch the whole message to UCS-2 and drop the limit from 160 to 70.
What that means in practice
A typical reminder looks like this:
Hello! A reminder about your appointment tomorrow, 14 March, at 10:30. See you then!
In Romanian that text runs to 84 characters. Without diacritics it fits comfortably in one segment. With diacritics it passes the 70-character limit and becomes two segments, twice the cost for exactly the same words.
Longer messages get worse: concatenated segments give up a few characters to the reassembly header, so the limits become 153 for GSM-7 and 67 for UCS-2.
How to decide
There is no universal answer, but there is a practical rule:
- Operational messages (reminders, confirmations, codes): drop the diacritics. The text stays perfectly readable and the cost halves.
- Brand messages to clients who care about form: keep the diacritics, but write shorter, under 70 characters.
The platform shows you the segment count and the cost before sending, and diacritic stripping can be enabled at account level so you do not have to think about it every time.
What does not help
Replacing diacritics with apostrophes or other marks solves nothing if that mark is itself outside GSM-7. Emoji are outside GSM-7 too and trigger the same drop to 70 characters.