For family engagement coordinators: the families you most need to reach are the ones your platform reports as unreachable
The families who show up to curriculum night are not the ones your funding is measured on. The families you are accountable for are the ones whose row in the report is blank, and a blank row has at least eight different causes with eight different fixes.
| Cause | How it looks in a report | The evidence that identifies it | The fix |
|---|---|---|---|
| Dead or wrong number | Sent, never delivered | Carrier failure on every attempt across months | Re-collect at a physical touchpoint — registration, a sports form, the front desk |
| Wrong guardian | Delivered, never a response | One contact on file for a student in two households | Add the second household as its own guardian record on the student |
| No consent for SMS | Email only, silently | Consent state on the contact point is unset or STOP | Voice and email until consent is re-collected; never assume STOP was a mistake |
| Language mismatch | Delivered, opened, no action | Home language in the SIS differs from send language | Translate before sending and check the translation |
| Channel mismatch | Email delivered to a dormant address | Zero engagement on email, response on voice | Move the household to voice or SMS as primary |
| Timing | Delivered overnight or mid-shift | Send timestamps cluster outside waking or non-working hours | Quiet hours, and scheduling sends for when the household is reachable |
| Distrust | Delivered, read, ignored, consistently | Household responds to the school but not the district | Send from a named person at the building, not the district identity |
| Genuine disengagement | Everything above ruled out | All channels valid, all delivered, no response over a term | The only one that is a home-visit problem rather than a data problem |
Why the table above is the whole job
Seven of those eight causes are data or delivery problems that a coordinator gets blamed for as if they were relationship problems. A district that reports "31% of families unengaged" and treats all 31% as a single population will spend its Title I parent-engagement budget on the wrong intervention, because the household with a disconnected number and the household that reads every message and disagrees with the school need completely different responses.
The reason platforms do not separate them is that separating them requires storing, per contact point, both whether the channel works and whether the family has agreed to it — and then reporting those two things apart from each other. Kastr stores validation state and consent state as distinct fields on every contact point, alongside the source it came from and its rank. That is not a feature so much as a precondition for the diagnostic being possible at all.
The related distinction is between a send and a reach. See true reach for why a delivery report that counts "sent" inside "delivered" — which is common, and which we have had to correct in our own analytics code — makes the table above impossible to build.
Language: what we do, and the number we will not inflate
Translation runs through DeepL, and you can render a draft in several languages inside the composer before anyone receives it. That preview is the point. Machine translation of school language is unreliable in specific, predictable ways — "referral", "retention", "suspension" and "conference" all carry meanings in education that a general model does not reliably choose — and a coordinator who reads the Spanish before it sends will catch it. A coordinator who receives a complaint the next morning will not.
Three honest limits. DeepL supports roughly 30 target languages, not the "47" that appears on some vendor sites including, historically, our own; if a family's home language is Marshallese, Mam or Chuukese, machine translation is not going to serve them and you need a human. There is no district glossary — no way to force "IEP" or a programme name to render consistently — and we are not going to claim one until it exists. And there is no "see original" footer appended to translated messages, so a bilingual parent cannot check our work the way you can.
The translation cache is keyed on a hash of the source text and language pair, which means we never pay to translate the same sentence twice, and also that the cache holds no index of who a string belongs to: a row is retrievable only by somebody who already has the exact source text.
Volunteers, PTA groups and the admin account you should not hand out
The recurring governance failure in family engagement is that the bilingual parent liaison or the PTA president needs to post to their group, so somebody gives them a district administrator login. That account can see every family in the district.
Kastr separates group administration from platform administration. A volunteer can be made an administrator of a single group: they post as the group identity, manage that group's membership, and see nothing outside it. Every group-identity post still records the true author's person ID in the audit log, so "the PTA posted it" never means "nobody is accountable for it".
Families can report a post, which lands in a moderation queue with four outcomes — dismiss, warn the author, remove the post, archive the group — and each action is audit-logged. That is a governance mechanism rather than a moderation product, and it is sized for a district that wants parent-led groups without wanting to read every message in them.
One person, one identity — including the ones who are staff too
A large share of the households you most need to reach contain someone who works for the district: bus drivers, paraprofessionals, cafeteria staff, custodians. In most platforms that person has two accounts and two sets of preferences, and the guardian one is the one that goes stale.
Kastr models one person row with many effective-dated role rows. A paraprofessional who is also a guardian at the same district is genuinely both, with one phone number, one language preference and one set of consent states. Roles are effective-dated, so a role that ends in June stops applying in June without deleting the history of it.
What this will not do for you
Set against the engagement programmes you are probably running: there are no push notifications — Kastr sends SMS, email and voice only, and there is no mobile app to push to. There are no attachments, photos or video, so a family-night photo recap is not something you can send from here. Auto-escalation does not fire; rules can be configured and nothing reads them. There is no digest batching, so a family's daily and weekly digest preferences are stored and never acted on. And audience targeting by school, grade or language does not resolve — only explicit person lists and everyone.
The one commitment that runs the other way: contract clause §9.4 forbids us from marketing to your families, selling to them, or taking any family-facing subscription revenue, permanently and regardless of who owns us. Engagement platforms that monetise the family side eventually ask your parents for money, and that request arrives wearing your district's name.
Questions people actually ask
How do I tell the difference between a family who ignored a message and one who never received it?
By reading validation state and consent state separately from delivery outcome. A contact point that has never accepted a message on any attempt is a data problem; a contact point with clean deliveries and no response is an engagement problem. Any report that collapses those two into one number — and most do, including reports that count 'sent' as 'delivered' — makes the distinction impossible.
Can a parent volunteer run their own group without district admin access?
Yes. Group administration is a separate capability from platform administration. A volunteer group admin can post as the group and manage that group's members, and can see nothing outside the group. The true author is still recorded on every group-identity post for audit purposes.
How do I communicate with a family whose language is not in the composer list?
You use a human. DeepL covers roughly 30 languages and school districts routinely serve families whose home language is outside that set. We would rather tell you that than publish a language count we cannot support. For languages that are covered, translate in the composer, read the preview, and edit the source English if the translation is wrong — there is no glossary to pin a term with.
Can one person be both a guardian and a staff member without two accounts?
Yes. One person row carries many effective-dated role rows, so a bus driver who is also a parent at the same district is one identity with one phone number and one set of preferences. This is unusual: most platforms in this category create a separate record per role and let the two drift apart.
Will you ever advertise to our families?
No, and it is a contract term rather than a policy. Clause §9.4 prohibits marketing to, selling to, or generating any family-facing subscription revenue from district families. Clause §11.2 gives you a 90-day exit with export and a prorated refund if we are acquired or materially change our data terms, which is what stops §9.4 from being undone by a change of owner.
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.