One school advertised 86% job placement. Its own metrics said 30%.

A school’s chief executive announced a 100% job-placement rate

Then a federal regulator asked how many students that was.

One.

He had said it publicly. He acknowledged the sample privately. The rate was true and it measured nothing, and that is the whole problem with placement rates in one sentence.

Why a placement rate can be almost any number

A placement rate is a fraction. Everyone argues about the top of it. The number that decides the answer is the bottom.

Suppose 100 people enrol. 80 finish the course. Of those, 60 are looking for work when the survey runs — the rest took another path, went back to a previous job, or never replied. 48 of the 60 find work in the field.

The school can now advertise 80% placement without lying once. Eighty per cent of the people it counted found jobs. It counted 60 of the 100 who paid.

You are asking about the 100. The advertised number is answering about the 60, and nothing on the page tells you which question was answered.

Two rules follow, and they cost nothing to apply.

A rate without a denominator is not a rate. If a school will not tell you how many people started, the number it is showing you cannot be checked.

The survey window is part of the number. “Within six months” and “within a year” are different claims, and so is “of those seeking employment”.

What a regulator found when it checked one

On 17 April 2024 the United States Consumer Financial Protection Bureau took action against the coding school BloomTech and its chief executive. The record is cfpb-bloomtech-2024, docket 2024-CFPB-0001, and it is a grade 1 source on this project’s scale: findings of fact by a public body, published with the consent order attached.

The school advertised job placement of 71% to 86% within six months of graduation. The agency found its own internal metrics ran closer to 50%, and in some cases as low as 30%.

Sit with the size of that. The advertised figure and the measured one differ by more than twenty percentage points at the middle, and by more than fifty at the bottom. This is not a rounding argument or a definitional quibble. A student choosing on the advertised number was reading a different market.

What it cost the people who believed it

The same action covered how students paid.

The agreements took 17% of a graduate’s pre-tax income once they earned over $50,000, for 24 payments, up to a cap of $30,000. The school said these were not loans and created no debt. The agency found they were loans, carrying an average finance charge of about $4,000. One missed payment could trigger default, and the remainder of the cap became due at once.

BloomTech was permanently banned from consumer lending. Its chief executive was banned from student lending for ten years.

That is the price of the mistake this page is about: up to thirty thousand dollars, committed against a number nobody outside the school had checked.

One case is not a pattern. The absence is.

It would be easy to over-read this. One regulator examined one school and found the advertised rate was wrong. That does not prove every school’s rate is wrong, and anyone who tells you it does is doing the thing this page is complaining about.

The stronger claim is quieter, and it is about who is looking.

Almost nobody audits these numbers. There is one widely used standard for independently verified outcomes reporting in this sector. While researching Chapter 7 this project tried to establish how many schools currently submit to it and found no primary source giving a current membership count — that absence is recorded in the gap register rather than filled with a guess.

So the honest statement is this. Placement rates in this market are largely unaudited, and in the one case where a regulator did look, the advertised number was overstated by twenty to fifty percentage points. Those two facts together are worth more than either alone, and neither requires you to believe anything about schools in general.

What to ask before you pay

Ask for the funnel, not the rate. Five questions, in this order.

1. How many people started the cohort?
2. How many finished?
3. How many of those were counted as seeking work, and who decided?
4. How many were working in the field, and at what date?
5. Who checked, other than you?

A school that publishes its whole funnel is telling you something real, whatever the number turns out to be. A school that answers only the fifth question with silence has told you something too.

Where this came from

This is one record from a project that scores 46 computing careers against the same ten dimensions, with a source, a date and a reliability grade beside every figure. Chapter 7 is about checking a school before you give it money, and it is in the half that is free to read — no email box, no signup. So is everything the project could not find out.

If a figure here is wrong, the address is here, and the correction will be printed on this page.