Every Workforce Pell compliance question ultimately reduces to a table with one row per student. Can your program show that at least 70 percent of participants completed within 150 percent of normal time, and that at least 70 percent of completers were employed in the second quarter after program exit? If you cannot produce the row-level data behind those two rates, you cannot defend them — to your state, to your governor's certification process, or eventually under federal methodology.

This article defines the minimum schema for that table: the 70/70 cohort file. Start it on day one of your first Workforce Pell cohort. Reconstructing it later from scattered systems is the single most expensive mistake schools make.

The minimum schema

One row per student, per program enrollment. Ten fields are non-negotiable:

The exclusion column deserves its own rule. Under 34 CFR part 690, subpart H, exactly four circumstances let you remove a student from the 70/70 math: the student died, suffered a totally disabling condition, was called to military service for more than 30 days, or was incarcerated. An excluded student comes out of both the numerator and the denominator of both rates — completion and placement alike. Every non-blank value in this column needs an evidence file behind it (death certificate, medical documentation, military orders, incarceration record); an undocumented exclusion is an audit finding, not a judgment call. The 70/70 Report Generator roster template now supports this column and applies the exclusions to both rates automatically.

A derived eleventh column is worth storing rather than recomputing: deadline_150 = start_date + (1.5 × normal_completion_weeks) weeks. For a 12-week program, that is week 18. The 150 percent tracking playbook covers the operational edge cases — leaves of absence, re-entries, transfers.

Beyond the per-student rows, the file (or a companion program record) must carry: the program's length in both weeks and clock hours — the two eligibility bounds run independently; the published tuition and fees for each award year — the future input to the value-added earnings test; and graduate follow-up fields — employer, start date, and occupation with SOC code — which become load-bearing when placement shifts to occupation matching after award year 2028–29.

Sample CSV row

student_id,program,start_date,normal_completion_weeks,actual_completion_date,withdrawal_date,exit_date,employment_status_q2,verification_source,exclusion
S-2027-0142,Commercial Truck Driving (CIP 49.0205),2026-09-08,12,2026-11-25,,2026-11-25,employed,ui_wage_records,

This student started September 8, completed on November 25 — comfortably inside the 18-week (150 percent) deadline — and was verified employed in the second quarter after exit via state UI wage records. That row contributes favorably to both rates.

This schema matches the roster template used by the 70/70 Report Generator. If you maintain your cohort file in exactly these columns, you can upload it and generate both rates — and a submission-ready report — without any remapping.

Rules that keep the file defensible

• Denominators are decided by definitions, not convenience. The completion rate is measured over program participants; the placement rate is measured over completers. Record every enrolled student — including withdrawals — or your completion denominator is wrong. Never delete a row.
• Dates come from source systems. start_date from the enrollment agreement, actual_completion_date from the registrar's completion record, withdrawal_date from the same determination you use for R2T4. If the cohort file and the student file disagree, you have an audit problem, not just a data problem.
• unknown is not not_employed — but plan conservatively. How unverified completers are treated in the rate calculation depends on your state's methodology; until confirmed, assume an unverifiable completer does not help your numerator, which is why chasing verification matters.
• Snapshot it. Freeze a dated copy of the file at each reporting milestone. The version you submitted must be reproducible years later — see records retention for why the retention horizon here is longer than the Title IV three-year baseline.
• One file per program per cohort period. Failure consequences attach at the program level: a program that fails 70/70 loses Workforce Pell eligibility, and it (or a substantially similar program) cannot be re-established for two years.

For award years 2026–27 through 2028–29, completion and placement rates are determined and verified under your state's methodology — confirm specifics with your governor's office or state workforce board before relying on any calculation. Your state may require additional fields (for example, wage amounts, occupation codes for the Phase 2 placement definition starting 2029–30, or demographic data). Treat this schema as the floor, and add state-required columns to the same file rather than starting a parallel one.

Where the data comes from

The file is a join across at least three offices. Assign one owner who reconciles it monthly; the monitoring cadence article describes the rhythm.

What to do now

1. Create the file today for every active Workforce Pell cohort, with the ten fields above plus deadline_150 — even if employment columns sit at pending.
2. Backfill from source systems, not memory: enrollment agreements, registrar records, R2T4 withdrawal determinations.
3. Ask your state workforce board for its data specification and add any extra required columns now.
4. Assign a single owner and a monthly reconciliation step against the SIS.
5. Run the file through the report generator (linked above) to see your current rates and which rows are dragging them.