Achieve Education Group
Retention IntelligenceCrystalBall
🔒 All data stays on this computer — nothing is uploaded anywhere
—

EDUPRIZE Queen Creek

1

Withdrawal history

The registrar's weekly flat file of withdrawn students (the same export you have today, both campuses in one file — .csv or .xlsx). This teaches CrystalBall which families already left.

Drop file here or click to choose
2

Currently enrolled students

An export of every enrolled student — same columns as the withdrawal file (minus exit code / end date), plus Initial Entry Date and current ITR Status (codes 1–7). Household ID, Absences, and Tardies all sharpen the picture.

Drop file here or click to choose
3

Generate risk lists

CrystalBall scores every family on this campus and sorts them into High, Moderate, and Mild risk with a plain-English reason profile for each student.

How this campus's algorithm works

Built from 3,000 true EDUPRIZE attrition events (school years 2021–2027) — the registrar's exit-code legend now lets CrystalBall separate real attrition from graduations, in-district transfers, and other non-attrition exits. Of real attrition: 72% transferred to another in-state district, 17% were parent withdrawals pending a new school, 6% went private or out of state, 4% to homeschool. High ≥ 60 · Moderate 40–59 · Mild 25–39. In a backtest, 96% of the families who withdrew in the last two school years were carrying at least one of these signals beforehand — 72% carried two or more.

Every score comes with its reasons spelled out — principals never get a number without the "why."

What CrystalBall still needs to see clearer

The current file tells us who left, when, and from where. These additions would tell us why — and catch families earlier. Ask the registrar to add columns to the weekly flat file in this priority order:

Delivered ✓ (Sept 2026): Household ID, Absences, Tardies, special-education status, Initial Entry Dates, the ITR code legend (1–4 returning · 5 moving · 6 transferring locally · 7 undecided), and the exit-code legend — all wired into the algorithm above.
  1. The enrolled-student roster, weekly — same columns as the withdrawal file plus Initial Entry Date and current ITR Status. This is the file that turns the algorithm into an actual risk list.
  2. A 3-question exit survey — the exit codes now say where families went; a short survey says why they left.
  3. Academic flags — benchmark scores or grades trending down, plus retention-in-grade history.
  4. Discipline incidents and nurse/office visits — friction indicators.
  5. Fee / lunch balance status — financial stress predicts summer non-return.
  6. Parent engagement — event attendance, app logins, unanswered messages.
Also: the enrolled-student file is what turns this from a "who looks risky" list into true percentages — CrystalBall needs to see the families who stay to avoid over-flagging profiles that are actually common and stable.