The Retraining Payback Worksheet, One Credential Priced to a Break-Even Month

This is the worksheet for pricing a certificate or bootcamp before you enroll or quit your job. You take one program, pull its outcome records and the target role's entry wage from public data, discount by the odds that you finish and get hired, and come out with an expected break-even month plus a downside case you can defend. I carry one credential, CompTIA A+ into entry-level IT support, through every step, and I leave in the wrong turns that make most return-on-investment calculators say yes.

What a retraining break-even month measures

A break-even month is how many months of higher pay it takes to earn back everything a credential cost you, after you weight that higher pay by the odds that you actually finish and get hired. It is an expected value, not a promise, so it always needs a downside case beside it.

The calculators that come up when you search for a certificate return on investment calculator mostly skip the weighting. They take the program's advertised salary, subtract your current pay, and divide the tuition by the difference. That gives the best case: you complete, you get placed straight away, and you earn what the brochure says.

Two terms carry the rest of this guide:

• Counterfactual wage: what you would have earned over the same period without the credential. Your gain is the difference between your new pay and this figure, not between your new pay and zero.
• Entry wage: what a new hire in the target occupation earns. In practice that is the 10th to 25th percentile of the occupation's wage distribution, not the median.

The counterfactual matters because most credentials do not change it much. In Burning Glass Institute and AEI research covering 23,444 credentials from more than 2,000 providers, about 12% offered wage gains of 10% or more that graduates would not otherwise have had. Only 18% of earners saw any wage increase at all. The top 10% of credentials raised annual earnings by nearly $5,000 within a year. The average credential raised them by about $1,200.

There is one important exception, and it decides how you should read your own case. In the follow-up framework, counting mobility and field entry raises the success rate from 1 in 8 to 1 in 3. A credential that gets you into a new occupation can be worth far more than the same credential added to a job you already hold. Gains also differ by group: one-year gains were $1,600 for women and $916 for men. So the question is never just "does a certification increase pay". It is whether this credential, for this person, makes the move happen.

The worked example: CompTIA A+ into entry-level IT support

The teardown follows one person aiming at IT support specialist (SOC 15-1232) through an A+ preparation program. A SOC code is the federal Standard Occupational Classification code that wage data is filed under. I use stated assumptions wherever a public record does not exist, so you can swap in your own numbers.

• Target occupation: IT support specialist, SOC 15-1232, national wages. Use your metro if you can; the reason is in the next section.
• Credential: CompTIA A+, through a short program.
• Tuition: $3,000, the figure New America uses in its worked example.
• Current pay: about $24,000 a year ($2,000 a month). That is roughly what the two most common eligible-training programs lead to: medical assistant and nursing assistant both pay under $6,000 a quarter, according to the Harvard Project on Workforce.
• Program length: 3 months, inside the 8 to 15 week range for Workforce Pell eligible programs.
• Odds: the Workforce Pell floor of 70% completion and 70% placement, used as a stand-in when the program's own record is suppressed.

Before pricing anything, I checked whether A+ looks like a hiring gate for this role. If people get hired without it, the credential cannot be credited with the whole pay jump.

What this proves. The great majority of people holding these titles do not list A+. Either the credential is not the hiring gate, or holders leave it off their profiles. I cannot tell which from the index alone. Treat it as an open question and test it against live postings in your metro.

How it lies. Profile skill lists undercount credentials, because people forget to add them. The sample also contained noise. Of the 25 A+ holders I reviewed, 3 listed Per Scholas (2) or CompTIA Tech Career Academy (1) as their current employer. They were trainees listing their training provider, not support hires. And the Senior band returned 0 because these roles are not tagged Senior at all. The support ladder has to be traced through title changes, not seniority labels.

A cheaper test than enrolling: have Refolk score your current work history against a handful of real IT support postings. If the fit gap is customer-facing troubleshooting experience rather than the certificate, the certificate will not close it.

Where to find completer earnings, completion and placement rates

The program side comes from TrainingProviderResults.gov, which publishes state-reported outcomes for programs on eligible training provider (ETP) lists. The wage side comes from BLS Occupational Employment and Wage Statistics (OEWS), republished by occupation and area on O*NET OnLine's Local Wages page.

The program record

TrainingProviderResults.gov shows training cost, completion rate, credential attainment, employment and earnings. Its Data Dictionary lists the fields you need:

The earnings field is a quarterly figure. It is the most common misreading on the site, and it is covered under failure modes below. The value -1 is a code: for program length it means competency-based. What -1 means in the other fields is not established, so do not do arithmetic with it.

Expect gaps. Data are suppressed when samples are too small to protect personal information, when nothing was reported, or when the Department found significant data quality problems. On screen, suppressed records show asterisks and the note "Data Suppressed." Programs with fewer than 10 students are suppressed, and they cover 22 percent of all participants. In one spot check of the first 30 programs near the Labor Department's zip code, only 6 had Q2 earnings. The search page also flags states with data quality concerns, and some states are missing whole program years. Check the flag for your state before you trust a number. State ETP reports have to carry the current year plus the three preceding years, so you can usually find more than one cohort.

The entry wage

OEWS publishes 10th, 25th, 50th, 75th and 90th percentile wages at national, state and metro level. For entry pay, use the 10th to 25th percentile band.

For IT support, P10 is 66.2% of the median and P25 is 79.2%. For medical records nationally, the figures are 72.4% and 85.0%. I did not retrieve a primary-source OEWS figure for data analytics, and the nearest SOC code is ambiguous. If that is your target, settle which code you mean before pulling wages.

How to weight the uplift by completion and placement odds

Multiply the uplift by the probability that you end up in the job that pays it. Which probabilities you multiply by depends on where your uplift figure came from. The joint probability of finishing and getting placed is the completion rate times the placement rate.

Even at the Workforce Pell floor, 51% of enrollees end up outside "completed and placed". Codesmith's CIRR-audited report, covering 865 graduates, shows how much the time window matters: 43.6% employed in-field within 180 days and 70.1% within 360. CIRR's own site quotes about 71% of coding bootcamp graduates finding jobs within 180 days, but that is an undated press quote. Prefer a program-level audited report.

There are two ways to estimate the uplift, and they take different discounts:

1. Occupation-wage route. Uplift = target entry wage minus your current wage. You only get this if you complete and get placed, so multiply by completion × placement.
2. Peer-gain route. Uplift = the gain measured for comparable credentials against comparable non-earners, from the Burning Glass Institute's Credential Value Index. That gain was measured among people who earned the credential, and placement outcomes are already inside it. Multiply by completion only.

One public figure looks like a payback signal and is not one. Workforce Pell's value-added earnings (VAE) test takes graduates' median earnings, adjusts them by regional price parity, and subtracts 150% of the federal poverty guideline. Tuition and fees must not exceed the result. The bar is a multiple of the poverty line, not the target occupation's wage. In one analysis of Census earnings data on 10,025 short-term certificate programs, 97% cleared it, and two-thirds were too small to measure. Passing VAE tells you a program is not a disaster. It tells you nothing about your break-even month.

The retraining payback procedure, step by step

The method runs in seven steps. The first two take about an hour and a half, and the rest is arithmetic on the worksheet below. No public source sets out a standard method that combines tuition, exam retakes, study time and lost wages, so this one is derived from the parts that are published.

Target occupation (SOC):
Metro or state for wages:
Program and provider:
Record source: state ETP data / provider audited report / none
Completion rate = completed ÷ exited =
Placement rate = employed Q2 ÷ exited =
Q2 median earnings (quarterly):        × 4 = annual        ÷ 3 = monthly
OEWS P10 / P25 / median:
Cross-check: Q2 × 4 lands between P10 and P25?   yes / no / suppressed
Current wage (annual):
Route A uplift = P25 - current wage =        per year ÷ 12 =        per month
Route B uplift = peer-measured gain =        per year ÷ 12 =        per month
Weighted uplift: Route A × completion × placement | Route B × completion only =
Total cost = tuition + exam fee × (1 + retake probability) + lost wages while studying + unpaid search months × monthly pay =
Base break-even (months after starting the new job) = total cost ÷ weighted uplift =
Downside break-even = (total cost + 6 unpaid months) ÷ (P10 uplift × odds) =
No-placement case: sunk cost =

The A+ teardown: three scenarios and the wrong turns

Done properly, the A+ case gives a base break-even of about 3 months after starting the job if the credential gets you the move, and about 43 months if it performs like the average credential. The gap between those two numbers is the actual decision. I show the wrong turns first, because each one shrinks the break-even.

Wrong turn 1: the calculator answer

Plug the occupation median into a calculator and assume certain success: $61,860 minus $24,000 is $37,860 a year, or $3,155 a month. Divide that into $3,000 and the credential pays back in under one month. It looks decisive. It is also wrong on two counts: no new hire earns the median, and nobody has a 100% chance of finishing and getting placed. Using the median instead of P10 alone adds about $20,880 to first-year pay.

Wrong turn 2: the marketing salary

If the provider advertises a salary, check where it came from. Codesmith's $110,000 median salary comes from the 57.1% of job-getters who disclosed one. That is a median of a subset of a subset. Without the response rate and the denominator, an advertised salary cannot go on the worksheet.

Wrong turn 3: the quarterly figure read as monthly

If the state record for an IT support program sat right at P25, its Q2 median would be about $12,250. Read that as a monthly figure and you get $147,000 a year, three times the real number. The cross-check catches this: Q2 × 4 has to land between $40,980 and $49,000, which means a Q2 median of roughly $10,245 to $12,250.

Wrong turn 4: discounting twice

On the peer-gain route, I first multiplied the average $1,200 gain by the 49% joint odds. That gives $49 a month and a break-even of about 61 months. But the peer-measured gain already contains the placement outcome. Weighted by completion only, it is $70 a month and about 43 months. Double discounting makes a credential look worse than it is, which matters when you are comparing two programs.

The scenarios, done properly

Occupation-wage route, base case: P25 of $49,000 minus $24,000 is $25,000 a year, or $2,083 a month. Weighted by 49%, that is $1,021 a month. The person keeps their job while studying, so total cost is $3,000 plus exam fees. Break-even is about 3 months after starting the new job. On the calendar, that means 3 months of study, a search of up to 180 days while still employed, then 3 months of new pay: about month 12 from enrollment.

Occupation-wage route, downside: P10 of $40,980 minus $24,000 is $16,980 a year, or $1,415 a month, and $693 after weighting. The search runs to 360 days. The first six months are spent searching while still employed, and the last six with no income, which costs $12,000. Total cost is $15,000. Break-even is about 22 months after starting, or about month 37 from enrollment.

Before weighting, the top-decile gain of about $5,000 a year pays $3,000 back in about 7 months, and the average $1,200 takes about 30. Weighting by 70% completion moves those to about 10 and 43 months. Weighting stretches every case. The real fork is which route describes you.

A credential that only pays back if it gets you the job is a bet on the job, not the credential.

Reading the fork

Route A says 3 months and Route B says 43. Route A credits the certificate with the whole jump from $24,000 to an IT support entry wage. Route B credits it with what credentials typically add. Refolk's index points toward Route B for A+ specifically, since 99.48% of entry-level holders of these titles do not list it. The field-entry research points the other way for career changers, since entering a new field triples the success rate. Your own evidence decides which applies to you: count how many local postings require A+, and look at whether people with your background got hired without it.

The search months are part of the cost, which is why the downside case is so sensitive to them. Once you are credentialed, Refolk tailors your resume to each IT support posting from your own history, so the troubleshooting you did in a previous job reads as support experience instead of getting lost in a generic resume.

How the payback calculation goes wrong

Most retraining cost vs salary increase estimates fail in one of eight predictable ways. Each one makes the break-even look sooner or later than it really is. Each comes with a check you can run in minutes.

1. Quarterly read as monthly. The d125 earnings field covers a quarter. Reading it as monthly overstates pay three times over. Check: Q2 × 4 should land near the OEWS P10 to P25 band for the occupation.
2. Suppressed read as zero, or zero read as data. Virginia's list once showed zeros for every outcome. A 0 or -1 can mean suppressed, unreported, or a code. Check: read the suppression note and the Data Dictionary before using any 0 or -1.
3. Marketing salary. Codesmith's $110,000 comes from the 57.1% who disclosed a salary. Check: ask for the response rate and the denominator; with neither, the figure stays off the worksheet.
4. Median used as the entry wage. For IT support, that inflates year-one pay by about $20,880. Check: use P10 for the downside and P25 for the base.
5. Double discounting. Peer-measured gains already include placement outcomes. Check: if your uplift came from peer comparisons, multiply by completion only.
6. 100% odds assumed. Even the 70/70 floor leaves 51% of enrollees outside completed-and-placed. Check: every Route A uplift on the sheet must show two multipliers.
7. Title-filter noise. When you sample profiles to test whether a credential is the gate, 12% of a 25-profile sample from Refolk's index was off-target, because trainees list their training provider as their employer. Check: read the employer names before counting anyone as hired.
8. Waiting for VAE data. No program-level value-added earnings figure exists, and the test does not affect eligibility for several award years. Check: do not postpone the decision for it, and remember that 97% of measured programs pass anyway.

Two disagreements between sources are worth knowing about. On how long after completion Workforce Pell measures earnings, one description says three years after completing and another says the first full tax year after completion. Placement is measured at 180 days in one account and in the second quarter after exit in another. When you compare programs, make sure their figures use the same window, or a 180-day figure will look worse than a 360-day one for no real reason.

Before you enroll or quit: final checks and keeping the figure current

You are done when the worksheet has no blanks, the three scenarios are written down, and you have decided which route describes you. Then re-check the inputs whenever a new cohort is published or your situation changes.

When the state record is suppressed, ask the provider directly. Its answers fill the same worksheet lines.

1. Of everyone who enrolled in your most recent cohort with full data, how many completed?
2. Of those completers, how many were employed in the target occupation at 180 days, and at 360 days?
3. Is your salary figure the median of all employed graduates, or only of those who disclosed? What was the disclosure rate?
4. Are your outcomes audited by a third party such as CIRR, or reported to the state eligible training provider list? Under which program name and code?
5. What did completers earn in the second quarter after exit, and is that figure quarterly or annual?
6. How many graduates counted as employed were hired by you or a partner training program as staff, instructors or apprentices?

To keep the figure current:

• Program records. State ETP reports roll forward and carry the current year plus the three preceding years. Re-pull the record when a new program year appears, and check whether your state's data quality flag has changed.
• Wages. OEWS percentiles are re-released on a regular cycle. Re-pull P10 and P25 from O*NET Local Wages for your metro before you commit, not just when you start researching.
• Peer gains. Check the Credential Value Index for your specific credential rather than relying on the all-credential averages used here.
• VAE. The Department of Education will publish value-added earnings at least three months before the award year it applies to. Treat it as a floor check when it arrives, not as your payback figure.

If the base case pays back in under a year but the downside runs to several years, the inputs that matter most are the ones you control: keep your income, pick the cheaper route to the same credential, and confirm the credential is the gate before you pay for it.