Every program's earnings accountability test projected as a range, months before the Department's result, with a decided response for each one at risk
Every program's earnings test projected months ahead, with a response for each at risk.
See one case, screen by screen ↓Earnings Accountability Monitoring lets institutional research and financial aid offices project each program's earnings accountability test and flag the programs at risk, months before the Department publishes its result. Agents rebuild each program's federal aid completer cohort, match it to state wage records and public earnings data, project the next three results as ranges with the chance of failing two of three, and draft responses for each program at risk. People confirm the cohorts, certify the analysis and decide every response.
AI in education: a buyer's guide to agentic AI for colleges, universities and school districts →
Updated
The earnings accountability test holds every Title IV program to one comparison from July 1, 2027: the median earnings of its completers who received federal aid, four tax years after completion, against the median earnings of working adults aged 25–34 with only a high school diploma in the college's state. A program that fails two of three consecutive years becomes a low-earning outcome program and loses Direct Loan eligibility for at least two years, and after a first failure current students must be warned within 30 days. The Department measures earnings from IRS data the college never sees, so by the time a result is published, the cohorts that decide the next two results have already graduated.
The counting is easy to get wrong. A program is a six-digit CIP code at a credential level; cohorts under 30 combine prior years, up to four, then the four-digit CIP, and a completer coded to the wrong CIP in the aid system moves a median. Today an IR analyst rebuilds cohorts in spreadsheets from public field-of-study data and the college's own extracts, and the result is a single number per program, with no range and no decision attached. Earnings Accountability Monitoring puts seven agents on that rebuilding, matching and projecting, so the IR director, the financial aid director and the vice president spend their time confirming, certifying and deciding.
Estimated split of an IR and financial aid team's working days for one full projection, by hand and with the agents; the 'after' column is mostly people confirming and deciding. Our arithmetic, kept modest.
Seven agents rebuild each cohort, match wage records and project three results; Financial Aid confirms, IR certifies, the vice president decides.
Brandvale Community College has 140 programs and $48.6 million a year in Title IV aid. It is Tuesday, February 9, 2027, 142 days before the Department's first determinations. Here is how the AAS in Early Childhood Education goes from a row below the line to a decided response, screen by screen, in the working solution.
The agents rebuilt 140 cohorts from the student information system and the financial aid system and matched 3,794 graduates to state wage records. For the July 1, 2027 result, 96 programs are measurable: 7 are projected below the $32,600 state benchmark, 12 are too close to call and 77 are clear. The other 44 are too small to measure, and the ladder lists them so nobody assumes they passed.
Early Childhood Education: cohort 42, 80% federal aid, projected range $28,300–$31,200.
The program file shows the three results that decide eligibility. 2027: $29,800 against $32,600, a $2,800 gap, and its earnings year has passed. 2028 is a close call. 2029 is $300 below with a range that crosses the line, and its earnings year is still ahead, so it can still move. The chance of failing two of three is 72%.
I can't certify a number. I can certify a range and the records behind it.
34 of the 42 cohort members are found in the state wage records, and 13 of them (38%) work part-time. The 21 who work a full year for one employer earn a median of $34,100, above the benchmark. The agent names the 8 the records cannot see (4 moved out of state, 3 self-employed home child care providers, 1 with no record) and widens the range for them instead of dropping them.
The Cohort agent rebuilt the cohort step by step: 58 completions recorded, 55 students who completed, 44 who received federal aid, 42 in the cohort. The Department's draft completers list counts 44, because two students who completed the Child Care Provider certificate (19.0709) were coded to the AAS in the aid system. Tomás Rivera prepared the correction and Denise confirmed the cohort before the review window closes on February 26.
Correcting the Department's list adds $300 to the 2028 median. Holders of the state director credential earn $4,100 more in the region; offered now to the 44 graduates of 2022–23, with 40% expected to take it, it lifts the 2029 median by about $1,600, for about $38,000 in tuition waivers. Redesigning the AAS around the credential and a paid practicum protects every cohort from 2027–28 on.
With the correction and the credential sprint chosen, the chance of losing Direct Loans falls from 72% to 41%.
Marcus chooses the correction, the credential sprint and the redesign with reason code R1, redesign to raise earnings, and keeps the borrower bridge as a contingency. Anyone else sees the decision disabled, and the screen says why. The evidence file holds the certified projection, the confirmed cohort and correction, the banded wage summary, the labor data and the decision.
Keep the program: it fills a regional shortage. Correct the list, run the credential sprint now, redesign for 2027–28; hold the borrower bridge as a contingency.
The decision opens a plan with seven milestones, from the credential sprint offer on February 23 to the Department's first result on July 1. The dean accepts it. The $38,000 in waivers moves money, so only the Director of Financial Aid can approve that line.
Plain-language warnings for the seven programs below the line are drafted in English and Spanish, for current and prospective students. Each says what the Department measures, that Direct Loans end if it happens again, and that Pell Grants do not change. Nadia checks the numbers, Denise approves the wording, and nothing goes out before the Department's result.
623 of 6,780 federal aid recipients (9.2%) and $4.1 million of $48.6 million (8.4%) sit in programs projected below the line. Even if every close call failed too, the shares are 21.7% and 21.2%, well under the 50% limit that would put all federal aid for those programs at risk.
The audit trail records who, what, when, why and which evidence, for people and agents alike: the overnight runs, the flagged completers, the correction, the certification and the decision. A board pack, an appeal file or an auditor pack exports in one click, with student identifiers redacted.
The same program, seen by the five people who carry it, from the cohort to the decision: what their work looked like, and what it looks like now.
Seven agents rebuild each cohort, match wage records and project three results; Financial Aid confirms, IR certifies, the vice president decides.
Plans the monthly run for every program, tracks the three results and routes each item to the right person.
Rebuilds each program's federal aid completer cohort under the rule's counting, from the student information system and the financial aid system, and compares it with the Department's draft completers list.
Matches cohorts to state wage records, annualizes earnings, separates part-time work, names the coverage gap and blends public field-of-study earnings.
Sets the state benchmark from the American Community Survey (ages 25–34, high school only), applies the in-state test and projects the next years' benchmarks with a range.
Projects each result with an 80% range, the chance of failing two of three, and the institution-wide exposure against the 50% limit.
Drafts response options with their modeled effect on the result each one moves, cites the labor data and assembles the evidence file.
Drafts the current and prospective student warnings in English and Spanish, and tracks the 30-day clock and the acknowledgments.
Decides each response with a reason. The agents propose; a named person decides.
Every program file has an ask box, and the assistant answers across all programs. Answers cite the records they come from, and some act on the page.
Why did the 2027 result miss?
Part-time work, not pay rates. 13 of the 34 graduates found in the wage records worked fewer than four quarters or below full time; the 21 full-year workers earned a median of $34,100. The projection treats the 8 not found with public field-of-study earnings as a prior.
What can still change the outcome?
Only the 2029 result is still movable: its graduates (2022–23) are measured on 2027 earnings. A director credential sprint lifts that median by about $1,600; with the correction to the Department's list, the chance of losing Direct Loans falls from 72% to 41%.
Which close calls could fail two of three?
Twelve programs have a range that crosses the line. The highest chances of failing two of three: Baking and Pastry 34%, Phlebotomy 31%, Barbering 29% and Fitness Trainer 27%.
How close are we to the 50% limit?
Far from it. 623 of 6,780 federal aid recipients (9.2%) and $4.1 million of $48.6 million (8.4%) are in programs projected below; even if every close call failed, 21.7% and 21.2%.
12 screens from the working solution, on its sample data. Pick one to see it large.
The rules every projection runs on are settings, versioned and logged; the ones set by regulation are locked.
| Setting | Default | Choose from |
|---|---|---|
| Range level | 80% | 80% or 90% |
| Close call | Range crosses the line | Or within $1,000 of the benchmark |
| Minimum cohort | 30 completers | Fixed by regulation |
| Combining small cohorts | Prior years up to four, then the four-digit CIP | Set by the rule |
| Benchmark | State · 94% in-state | State, unless fewer than half of students are in-state |
| Cohort match confidence | 0.95 | 0.90, 0.95 or 0.98; below it, a person checks the match |
| SLAs | Cohort confirmation 3 business days · decision 10 · dean review 20 | Edited by an administrator with a second approver |
| Student-level wage matches | Two named stewards | FERPA and the state data-sharing agreement |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
Sources: NASFAA, The One Big Beautiful Bill: Accountability Framework Planning & Implementation Checklist (July 1, 2026) · McGuireWoods, Education Department Proposes New Earnings Accountability Framework (May 2026) · Sikich, Breaking down the OBBBA earnings accountability (cohort of 30 and combining rules, two-year ineligibility, 30-day warnings and acknowledgments, tipped-income delay) · IHEP, citing a PEER Center estimate (about a quarter of undergraduate certificate students in programs that would fail) · "demo" = seen in the working solution, on its sample data · "target" = design goal, measured in the live solution · "typical" = published figures and the rule's own numbers · "estimated" = our arithmetic, kept modest. Brandvale Community College, its programs, earnings, benchmark and credential figures, and the people named on this page are fictional: characters and sample data in the working solution.
Earnings Accountability Monitoring lets institutional research and financial aid offices project each program's earnings accountability test and flag the programs at risk, months before the Department publishes its result. Agents rebuild each program's federal aid completer cohort, match it to state wage records and public earnings data, project the next three results as ranges with the chance of failing two of three, and draft responses for each program at risk. People confirm the cohorts, certify the analysis and decide every response.
From July 1, 2027, the Department of Education compares the median earnings of each Title IV program's completers who received federal aid, four tax years after completion, with the median earnings of working adults aged 25–34 whose highest credential is a high school diploma, in the college's state. A program that fails two of three consecutive years becomes a low-earning outcome program and loses Direct Loan eligibility for at least two years.
The final rules take effect July 1, 2027, when the Department makes its first low-earning determinations. Because a program must fail two of three years, the earliest loss of Direct Loan eligibility is July 1, 2028. Programs that lead to occupations with significant tipped income are first counted on July 1, 2028.
After a first failure, a college that keeps the program must warn current students within 30 days and prospective students before they enroll, with a written acknowledgment from each. A second failure in three years removes Direct Loans for at least two years. If more than half of a college's Title IV recipients or Title IV dollars sit in low-earning programs in two of three years, those programs lose all Title IV aid and the institution may be placed on provisional certification.
The Cohort agent counts completers who received a Pell Grant or a Direct Loan, at the six-digit CIP code and credential level. Cohorts under 30 add prior years, up to four, then combine the four-digit CIP; still under 30, the program is not measured. It then compares the rebuilt cohort with the Department's draft completers list, so errors can be corrected inside the review window.
Not the Department's figure, which comes from IRS data the college never sees. The solution projects it from the college's own records and state wage records, as a range rather than a single number, names the graduates the wage records cannot see, and marks which results are already fixed and which can still move.
Yes. The agents rebuild, match, project and draft; they never write to the student or aid systems, certify, decide or send anything to a student or the Department. The Director of Financial Aid confirms each cohort, the IR director certifies the analysis, and only the vice president decides a response, with a reason code on the record.
The Agentic Solution Engine builds and deploys it from your requirements and documents, such as your program list, completions and aid extracts and your state's wage-record agreement, and it goes live once every quality gate has passed. We will walk you through it on your own programs first.
No uncited sentence reaches the accreditor, and likely findings are answered first.
1-min demoIPEDS ReportingEvery survey cell rebuilt from source, every variance explained before lock.
1-min demoFaculty CredentialingEvery section cleared before the term starts, with every credential tied to its proof.
1-min demoWe'll run Earnings Accountability Monitoring on your own completions, aid records and programs.