About School Choice Community

We believe families should not have to become researchers to make a good school decision. Choice Architect translates your priorities into a shortlist you can understand, with clear "why this?" explanations and practical map context - always paired with honest limits on what any tool can know.

Who we are

Dr. Kai Lennox and a small founding team are building School Choice Community as a two-sided platform: families stay centered in the advisor experience, while school leaders get paths to tell authentic stories and see how families are searching. We combine education experience with product craft so the system feels human, not robotic.

Two-sided platform vision

Families use chat, maps, and profiles to explore options. School leaders use Tell Your School's Story to submit narratives we review before publishing. Each submission is added to our searchable index so families can find the school by name and context; we still recommend verifying details directly with the campus.

Coverage, freshness & limitations

We expand state-by-state from an NCES-backed spine, with metro weighting for priority regions (including New York and California) alongside Arizona, Michigan, and Florida. National mode searches across ingested states; naming a metro or ZIP still gives the best commute context. School counts and accountability fields are refreshed on a development schedule, not live with every state release - always confirm enrollment, transportation, and eligibility on official sites.

Thin regions may show a "growing coverage" note. School leaders can help us expand faster through For School Leaders.

How the product works

Retrieval blends semantic similarity with lightweight intent cues (grades, commute language, location).
Public data anchors many rows where we have ingested extracts; some catalog copy is illustrative for demos (especially marketplace-style items in Arizona).
AI responses summarize the shortlist you see - they can be wrong; use them as a starting point, not the final word.

Official sources & attribution

NCES EDGE / ArcGIS PublicSchools for nationwide coordinates, grade spans, and NCES IDs.
NCES CCD (Common Core of Data) directory + universe indicators (where available) for public school context like locale, staffing ratios, and related directory fields.
Michigan CEPI / MDE and Florida FLDOE files for accountability joins where automation can reach them.
Urban Institute Education Data Explorer referenced for CCD-style investigations during pipeline development.
NCES PSS (Private School Universe Survey) for private-school locations, enrollment, and grade span signals (2021-22 public-use file in this sprint).
Stanford Education Data Archive (SEDA 5.0) supplies pooled grades 3-8 cohort-scale benchmarks we merge by NCES school ID (national percentile ranks are computed during ingest; when only pooled math and combined averages exist, reading uses an implied ELA composite). Cite: Reardon et al., SEDA v5.0 (Stanford Digital Repository). Refresh by running scripts/ingest_seda.py or national ingest with a cached pool CSV.
EdFacts / Ed Data Express is used as a fallback performance layer when SEDA is missing. We currently ingest percent proficient on statewide math and reading/language arts assessments (All Students, All Grades) and label it clearly by school year; because these are state tests, they are not perfectly comparable across states. Refresh with scripts/ingest_edfacts.py.

Performance freshness: star ratings and tiers summarize archived pooled benchmarks. They are orientation signals only - pair them with current state accountability releases and campus visits. Coverage varies: not every school appears in the Stanford pool; we merge by NCES ID when possible, then careful name-and-state matches, then district-level estimates when school data is missing.

Privacy

We use privacy-minded, mostly anonymous analytics (see our full Privacy Policy). Optional "Use my region" uses coarse IP-based US state only, never precise GPS. Terms of use: Terms of Service.