The summary is not the solicitation.

A notice came through our opportunity scan that read as though it had been written for us. A federal statistical agency wanted to pull operational data out of thousands of independently-run local systems, standardize it into uniform fields, and turn the result into something analysts could actually query. Ingestion, schema normalization, an analytic layer on top. That is not adjacent to what we do. That is the work.

We scored it as a pursue and filed the work to draft a response. Four hours later we withdrew it. Nothing about the opportunity had changed. We had opened the attachment.

What the summary said

The posted description on a federal notice is a scope teaser. It exists so that a firm scanning hundreds of notices can decide in fifteen seconds whether to keep reading. It describes the work — the problem, the desired end state, sometimes the scale. It is written to attract capable respondents, and it does that job well.

Read on its own, ours described a data engineering program in plain terms, and every sentence of it matched what we have spent a decade doing. We would score that capability match the same way again. It was real.

What the attachment said

The attached document was seven pages. Most of it was background. Then, under submission instructions, one line: provide three previous contracts where similar services were provided in the last five years — including the contract number, the period of performance, the total contract value, and the contracting officer and program manager, each with a telephone number and an email address.

Not a narrative. Not adjacent work described persuasively. Three references a government buyer can pick up the phone and call. Alongside it: demonstrated experience with the specific class of source system, and demonstrated experience building a shared data schema across the several competing vendors who supply it.

The description tells you what the work is. The attachment tells you who is allowed to claim it.

Those are different questions. They live in different files. And only the first one is on the page you were reading when you decided the opportunity looked good.

Why this failure is flattering rather than obvious

If the summary had described work we do not do, we would have moved on in seconds. The trap only springs when the fit is genuine. The description reads as an ideal match precisely when the work is squarely in your lane — so the better it looks, the more confident the wrong call, and the further you carry it before anything contradicts you.

That is worth naming as a shape rather than an anecdote, because the correction is cheap and mechanical: fetch the attachment before you record an opportunity as something you are pursuing, not after. Ours cost one afternoon and one PDF. Had we read it a week later, it would have cost a drafted response and the one thing a small firm cannot manufacture — the principal's attention.

The older mistake underneath it

Finding this notice at all exposed something worse, and it had been running quietly for months. Our scanner filtered every query to the set-aside category we qualify for. That sounds like hygiene. Why look at opportunities you are not eligible for?

Because a large share of federal notices are market research. The agency has not yet decided whether to restrict the eventual contract to small or veteran-owned firms — and the way it decides is by reading the responses. Filtering to notices that already carry the set-aside means you only ever see decisions that have already been made, in a process whose entire purpose is making them.

We had built a scanner that could only find doors already open, inside a process designed to decide which doors to open.

The filter was invisible because it felt like eligibility discipline rather than a blind spot. When we ran a keyword pass without it, one slice that had shown three notices showed ninety-seven. Three windows in our own capability area had already opened and closed without a response.

What we changed

Two things, both small. The attachment now gets read before an opportunity is recorded as a pursuit, and the qualification bar it carries — past performance references, demonstrated-experience clauses, required contract vehicles — is checked as its own gate, independent of whether we can do the work. And the scan now runs a second pass with no set-aside filter, keyed to what we actually do rather than to what we are already eligible for.

Neither is clever. Both were obvious in hindsight, which is the usual character of the things that cost you quietly. The reason we are writing it down is that the first mistake produced a decision we were confident in and wrong about, and the second produced no signal at all — just a smaller list, arriving on schedule, looking exactly like a complete one.