NHA NHA AI Analytics NHA AI apps

Analytics

How quickly — and how well — are staff responding to parents through the NHA App?

Pilot phase · 4 schools · 8,670 messages analyzed
What we measure

For every parent-initiated inquiry: did it get a response, and how fast? Reported as response rate and first response time — per school and per staff member.

How we judge "needs a response"

A classifier reads each message: questions and explicit requests need a reply; "got it, thanks" and emoji-only messages don't. Ambiguous cases go to a second pass (AI, Phase 2).

How we measure time

Business hours only — 7am to 6pm, Monday through Friday, excluding holidays. A Friday-evening message answered Monday morning is ~3 business hours, not 64.

Where AI comes in

Today: rules-based only. Phase 2: Claude Haiku scores responses on Quick, Clear, and Kind — three independent dimensions, never averaged into one score.

What this measures and what it does not

This pilot covers four NHA schools: Pathway (OH), Wake Forest (NC), Linden (MI), and Canton (MI). Across all four, 81.9% of parent-initiated inquiries received a response, with the median response landing within two business hours at our most active school.

The numbers produced here are responsiveness signals — not a judgment of any staff member's performance. The philosophy is "data creates conversation, not consequences." Response quality scoring (did the reply actually answer the question? was the tone warm?) is a Phase 2 feature.

Want the full explainer? What it does and how.