Controlling SMS spend without making your messages worse
Almost every district that overruns its messaging budget does so for the same two reasons, and neither is sending too many messages. It is segment inflation from characters nobody thought about, and voice failover from contact data nobody cleaned.
| Encoding | Single message | Per segment when concatenated | Triggered by |
|---|---|---|---|
| GSM-7 (standard) | 160 characters | 153 characters | Plain ASCII letters, digits, basic punctuation |
| UCS-2 (Unicode) | 70 characters | 67 characters | Any single character outside GSM-7 anywhere in the message |
| Character | Where it comes from | Effect on a 150-character message |
|---|---|---|
| Curly apostrophe ’ | Autocorrect in Word, Google Docs, iOS | 1 segment becomes 3 |
| Curly quotes “ ” | Same | 1 segment becomes 3 |
| Em dash — or en dash – | Autocorrect, or pasting from a document | 1 segment becomes 3 |
| Ellipsis … as one character | Autocorrect | 1 segment becomes 3 |
| Any emoji | Deliberate, usually to seem friendly | 1 segment becomes 3, and emoji cost 2 characters each |
| Accented names in the body | Unavoidable, and correct | 1 segment becomes 3 |
| Non-breaking space | Pasted from a web page, invisible | 1 segment becomes 3, cause undiscoverable by eye |
The arithmetic that catches everyone
A 150-character message in plain ASCII is one segment. Paste the same message from a word processor, where the apostrophe in "school's" has been autocorrected to a curly one, and the entire message switches to UCS-2 encoding. Now the limit is 67 characters per segment, so 150 characters becomes three segments. You have tripled the cost of every district-wide send without changing a visible word.
Multiply through: 3,000 recipients, three segments instead of one, is 9,000 segments instead of 3,000. Do that on the twenty-seven district-wide sends in a school year and the difference is 162,000 segments. That is the single largest controllable line in most districts' telephony spend, and it is invisible in every report except the invoice.
Two practices fix it permanently. Compose in a plain-text field rather than pasting from a document. And treat the accented characters you genuinely need — a family's name, a Spanish message — as a deliberate cost rather than an accident, because those are worth paying for and a curly apostrophe is not.
A cost model for a 5,000-student district
Roughly 7,500 contactable guardians. Three sending intensities, at the deliberate one-cent-per-SMS upper bound the Kastr composer uses for estimates:
- Light — 18 district sends a year, one segment each. 135,000 segments, about $1,350. Building-level traffic adds roughly half again: call it $2,000.
- Typical — 27 district sends plus 4 per school per month plus attendance traffic. Around 420,000 segments, about $4,200, assuming disciplined single-segment messages.
- Undisciplined — the same volume with average two-and-a-half segments per message and 12% voice failover. Over a million segments plus roughly 50,000 voice attempts. Comfortably three to four times the typical figure.
The gap between the second and third rows is not caused by sending more. It is caused by longer messages, unicode characters, and failing contact data converting texts into voice calls. All three are fixable without sending less.
Two honest caveats about the numbers you will see in-product. The composer's estimate uses one cent per SMS as a deliberate upper bound, so it over-states rather than surprises. And single-segment sends can round to $0 in the insights view, which makes small sends look free when they are not quite. Use the composer estimate for decisions and your carrier invoice for the budget.
Rewrites: same meaning, one segment
Each of these goes from three segments to one. The originals are the shapes districts actually send.
- Before (214 chars): "Dear parents and guardians, please be advised that due to inclement weather conditions in our area, all after-school activities scheduled for this afternoon have been cancelled. We apologise for any inconvenience."
After (92): "All after-school activities are cancelled today due to weather. Normal dismissal at 3:10pm." - Before (188): "This is a reminder that parent-teacher conferences will be taking place next week and we encourage all families to sign up for a time slot using the link that was previously sent."
After (98): "Parent-teacher conferences are next week. Book your slot: lincoln.k12.example/conf" - Before (172): "We wanted to let you know that your child was marked absent today. If this is an error please contact the main office at your earliest convenience so we can update our records."
After (86): "Your child was marked absent today. If that is wrong, call the office on 555-0100."
The pattern is consistent: cut the greeting, cut the apology, cut "please be advised", lead with the fact, end with the instruction. Shorter messages also generate fewer front-office calls, which is the second saving and usually the larger one in staff time.
When voice is cheaper, and when it is not
Per recipient, voice is the most expensive channel and always will be. It is nevertheless the right choice in three cases:
- The family has no working text channel. A voice call that reaches them costs less than a text that does not.
- The message is long. Beyond about four segments, a 20-second voice message can be comparable in cost and considerably better received.
- It is a genuine emergency. Redundancy across channels is the point and cost is irrelevant.
What is not a good reason: using voice as a default because it feels more serious. Kastr fails SMS over to voice on terminal failure, which is the right shape — voice as a fallback for the recipients who need it rather than as a parallel channel for everyone. The corollary is that failover cost is a direct function of contact-data quality: every bad number you leave in the system buys a voice call every time you send.
Forecasting for next year is straightforward once you have a semester of data: take actual segments, apply your expected enrolment change, add 15% for the sends you did not anticipate, and hold the messaging-discipline improvements as your contingency rather than as a saving you have already banked.
Questions people actually ask
Why did a 165-character message cost double?
Because it crossed the single-message boundary. A plain GSM-7 message holds 160 characters; beyond that it is split, and concatenated segments hold 153 each, so 165 characters is two segments. If the message also contains a curly apostrophe or an emoji it switches to Unicode encoding at 67 characters per segment, and 165 characters becomes three.
Does adding an emoji really change the price?
Yes, substantially. A single emoji anywhere in the message forces the whole message into UCS-2 encoding, dropping capacity from 153 characters per segment to 67, and each emoji itself consumes two characters. One smiley can turn a one-segment message into three for every recipient.
Is voice cheaper than text for a whole-district message?
Almost never per recipient. Voice is worth its cost for families with no working text channel, for messages longer than about four segments, and for genuine emergencies where channel redundancy is the point. Using voice as a default because it feels more official is the most common way a telephony budget disappears by March.
How do we forecast next year's SMS budget from this year's usage?
Take actual segments rather than message counts, adjust for expected enrolment change, add about 15% for unanticipated sends, and treat any planned improvement in message discipline as contingency rather than as a saving already achieved. Present it as a range with the assumptions attached; a point estimate in year one will be wrong in a direction you cannot predict.
How do we stop staff sending expensive messages by accident?
Make the cost visible at the moment of sending. Kastr's composer shows a live per-channel dollar estimate as channels are selected, which changes behaviour more reliably than a policy document. Pair it with one rule staff can remember: compose in plain text, no emoji, no pasting from Word, and lead with the fact.
One price. Every feature. Locked for three years.
$3.50 per student per year under 5,000 students. No tiers, no add-on modules, no per-message fees. Published on the site because you should not have to book a call to learn a price.