FaithDash
← All articles
How-to ·FaithDash

Half the Message Was in Luganda. The Cost Doubled.

A perfectly natural bilingual church announcement quietly tripped an encoding rule most admins have never heard of — and cost three times what it should have.

A young woman enjoying her mobile phone outdoors
Photo by Zeal Creative Studios on Pexels

Sister Nabirye writes the way most bilingual Ugandans talk — English for the formal parts, Luganda for the warmth. Her Sunday reminder, the way she'd always written it, read: "Reminder: service starts 9am. Mujje mangu, tetulwa!" — come quickly, let's not be late. It's the kind of line that makes a text feel like it's actually from your church and not a company. She'd been writing messages this way for months without a second thought.

What she hadn't noticed — because nothing had ever told her to look — was that the character counter at the bottom of the composer had quietly crept from "1 part" to "3 parts" the moment she added a particular kind of punctuation mark she habitually used before Luganda phrases, a stylized quotation mark her phone's keyboard auto-corrected into. The message wasn't any longer, really. It just cost three times as much, for the exact same handful of words, and she only found out because she happened to glance at the total cost figure for four hundred recipients and did a double take.

Why one character can triple a bill

This isn't really about Luganda specifically — it's about character encoding, and it catches people writing in English just as often. Standard SMS uses an encoding called GSM-7, which fits a generous 160 characters into a single message part cheaply. The moment a message contains even one character outside that GSM-7 set — certain accented letters, curly "smart" quotes your phone auto-corrects to, emoji, some symbols — the entire message switches to Unicode encoding, which allows far fewer characters per part, around 70. A message that would have comfortably fit in one part at 160 characters can split into two or three the instant a single incompatible character sneaks in, even if the total character count barely changed.

Luganda itself, written plainly, mostly uses standard Latin letters that fit GSM-7 just fine — "Mujje mangu, tetulwa" on its own wouldn't have been the problem. The trap was almost always something smaller and sneakier: a phone's autocorrect swapping a straight apostrophe for a curly one, or someone pasting text from WhatsApp that carried an emoji or a stylized character along with it. It's an easy, invisible mistake, and it has nothing to do with which language you're writing in and everything to do with which exact characters land in the box.

What actually caught it

The reason Sister Nabirye noticed at all was the live composer counter — character count, GSM-7 versus Unicode encoding, number of parts, and cost for the selected recipient group, all updating as she typed, before she sent anything. She went back, found the one curly apostrophe her keyboard had swapped in automatically, replaced it with a plain one, and watched the count drop straight back from 3 parts to 1 — and the total cost for four hundred people drop with it.

A generic bulk SMS portal, in most cases, will show you a character count too — but not always this clearly, and not always broken down by encoding type so you can actually see why a message tripped into extra parts rather than just seeing a number climb with no explanation. Without that visibility, the honest failure mode is that you don't notice until the bill is higher than expected, and by then the message is already sent.

What still requires a bit of judgment

None of this means avoiding Luganda, or any local language, in church SMS — that would be a strange and slightly sad way to solve a technical quirk, stripping the warmth out of a message to save a few shillings. The actual fix is smaller: type plainly, watch the live counter, and if it jumps to Unicode for no reason you can explain, look for the culprit character rather than assuming it's the language itself. Plain apostrophes over curly ones. Plain hyphens over em-dashes. Save the emoji for WhatsApp, where they don't cost anything extra.

What this actually costs, in real numbers

At UGX 35 per SMS part, a message that should have been one part costs 35 shillings per recipient. The same message accidentally split into three parts by one stray character costs 105 shillings per recipient — for four hundred people, that's the difference between 14,000 shillings and 42,000 shillings, for wording that reads identically to the person receiving it. Nobody on the receiving end can tell the difference. Only the church's balance can.

Sister Nabirye still writes her reminders the same warm, bilingual way. She just watches the little "GSM-7 · 1 part" note in the corner now, the way you'd glance at a fuel gauge — not because it changes what she wants to say, but because it tells her, before she sends, whether she's about to pay for saying it three times over by accident.

Send from your member list, not a spreadsheet.

FaithDash keeps visitors, members, attendance and messaging in one place — and can follow up automatically.

Open FaithDash

Related articles