Sends, deliveries, reach: how to tell whether your messages actually land
Every communications platform reports a big number and calls it delivery. Almost none of them mean what a superintendent thinks they mean. This is how to build a reach figure you can defend when somebody asks a follow-up question.
| Metric | What it proves | What it does not prove | How it gets inflated |
|---|---|---|---|
| Sent | Your system created a message record | That anything left the building | Counting one broadcast as thousands of "messages sent" |
| Accepted | A carrier or mail provider took the message | That it reached a handset or inbox | Reporting acceptance as delivery — the most common inflation in the category |
| Delivered | The carrier confirmed handset receipt, or the mail server accepted for the mailbox | That a human saw it | Folding "sent" or "accepted" rows into the delivered count |
| Opened | An email client loaded a tracking pixel | Anything at all about SMS or voice; increasingly little about email | Privacy proxies pre-fetching images, inflating opens to near 100% |
| Reached | A student has at least one guardian with a validated, consented contact point that received the message | That the guardian understood it | Counting contact points instead of students, so one over-connected family masks ten unreachable ones |
| Acted | A form was returned, a reply arrived, attendance changed | Reach for the silent majority | Attributing an action to a message that was one of five touches |
The worked funnel, and where it collapses
A 4,000-student district sends an early-dismissal notice. The dashboard says 3,900 delivered, which sounds like near-total coverage. Work it downwards instead:
- 4,000 students, of whom 3,780 have at least one guardian contact point on file. 220 students have none. That is your first and largest hole, and it appears in no delivery report anywhere.
- 3,780 students, but 410 of their primary contact points have never been validated — disconnected numbers, landlines flagged for SMS, addresses that bounced in September.
- 3,370 students with a plausibly working channel. Of those, 290 have no SMS consent recorded, so an SMS-only send does not lawfully reach them at all.
- 3,080 students genuinely addressable by this send. Of those, roughly 240 are in households whose preferred language is not English, and the message went out in English only.
- 2,840 students reached — 71% of enrolment, against a dashboard reporting 3,900.
The gap between 3,900 and 2,840 is not a rounding error, and it is not the platform lying exactly. It is that the platform counted contact points that received something, while the question you actually had was how many children's families know school ends at noon.
Eleven reasons a message shows delivered and reaches nobody
- Unregistered A2P traffic filtered by carriers. Silent. No error. The most damaging one, and the reason 10DLC registration matters.
- Number is a landline. SMS accepted, nothing arrives.
- Number belongs to somebody else now. Delivered, to a stranger.
- Wrong guardian ranked first. Delivered to the parent who does not do school logistics.
- Email routed to spam or a promotions tab. Delivered by every technical definition.
- Email address is a work address the parent no longer holds. Bounces are often not fed back into the roster.
- Language mismatch. Arrives, is not readable.
- Voice call to an unanswered line with a full voicemail box. Recorded as attempted, sometimes as delivered.
- Family opted out months ago and nobody noticed. Consent state and validation state are different fields; conflating them hides this.
- Message arrived at 9:40pm. Delivered, and never read.
- Student withdrew. The family receives messages about a child who left, which is worse than not reaching them.
Nine of those eleven are contact-data problems rather than platform problems, which is why buying a new platform rarely fixes reach on its own. The families-who-never-respond guide is the operational routine for the data half.
The five numbers to put in front of a board
Report these, with definitions attached, and refuse to report anything else:
- Guardian coverage. Percentage of enrolled students with at least one contact point of any kind. This is the ceiling on everything else.
- Validated coverage. Percentage with at least one contact point that has been validated and not subsequently failed.
- Consented reach by channel. Percentage reachable by SMS, by email, by voice, counted per student rather than per contact point.
- Language coverage. Percentage of students whose household preferred language is one you can actually send in, and the count of students in languages you cannot.
- Failure rate by school. Not district-wide. School-level, because the variation is where the actionable information lives — one building is always materially worse, and it is usually the one with the highest mobility.
Track all five per term rather than per send. Per-send numbers are noise; the term-over-term movement in guardian coverage is the thing that actually changes outcomes.
Our own honest disclosure about this
We do not publish a reachability percentage, and here is why. A figure like "99% of your families are reachable" is trivially produced by counting contact points rather than students, or by counting acceptance as delivery. We have seen it in this category and we are not going to compete with it.
More specifically about our own product, in August 2026: our insights view currently counts messages in the sent state inside its delivered figure. That contradicts the principle we publish elsewhere — that a vendor should never report "sent" as "delivered" — and it is a defect on our list rather than a design decision. We would rather you read that here than discover it in a board meeting.
The general rule, applied to us as much as to anyone: ask a vendor to show you the state machine. How many delivery states exist, what transitions between them, and which ones are being added together in the number on the dashboard. A vendor who can answer that in specifics is measuring something. A vendor who answers with a percentage is not.
Questions people actually ask
Does 'delivered' mean a parent read it?
No. For SMS it means a carrier confirmed handset receipt, which happens whether or not the phone is looked at. For email it means a mail server accepted the message, which happens whether it lands in the inbox or in spam. No platform can report human attention, and one that implies it is measuring a pixel rather than a person.
What is contactability and is it a real measurement?
It is a real measurement if it is defined per student rather than per contact point: the percentage of enrolled students with at least one validated, consented contact point on a channel you can actually send on. Defined that way it is the most useful single number a district can track. Defined as a count of phone numbers on file, it is marketing.
Why do our open rates look great and our attendance calls still come in?
Because open rates measure the families who were already reachable. The families driving your attendance calls are usually the ones missing from the denominator entirely — no valid contact point, no consent, or a language mismatch. Improving a 68% open rate to 74% does not touch them; adding 200 students to guardian coverage does.
How do we measure reach for families who only receive voice calls?
Answered-call rate and voicemail-deposit rate, tracked separately, and never folded into a combined delivery number. A voice attempt to an unanswered line with a full mailbox reached nobody, and if your report cannot distinguish that from a completed call, the voice portion of your reach figure is unusable.
What should we report to the board?
Guardian coverage, validated coverage, consented reach by channel, language coverage, and failure rate by school — each per term, each with its definition written beside it, and each counted per student. Do not report total messages sent. It is the largest number available and it proves the least.
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.