How FedScope decides what to show you

Why an answer looks the way it does — how records become results, why one company ranks above another, and what stops FedScope from making things up.

FedScope is a research tool built on official federal records. It connects records you would otherwise have to look up one search at a time — then shows you exactly which records produced the answer.

This page explains why an answer looks the way it does.

Your question

Plain language becomes supported filters: agency, NAICS, geography, set-aside, window.

Retrieve

Search indexed copies of official opportunity, award, and contractor records.

Connect

Associate records by stable identifiers; label inferences as inferences.

Rank

Order by evidence relevant to your stated job — not a universal “best.”

Inspectable answer

Every claim links back to the records — and the official source — behind it.

The system can only assert what it can retrieve; anything derived across records is labeled a research lead.

From your question to your answer

It reads your question. FedScope turns your plain-English question into the things it can actually search on: agency, type of work, location, small-business set-aside, keywords, and how soon a response is due.

It pulls the matching records. It searches its own organized copy of federal opportunities, awards, and contractor registrations.

It links the records together. Where an official ID exists, FedScope matches on the ID. Where one does not, it matches on cleaned-up names and details — which is less certain, and labeled that way.

It ranks for your question. Results are ordered by what matters for the job you asked about — not by some general claim that one company is "the best."

It clears the hallucination shield. Before the answer reaches you, every claim in it is checked against the records that were actually pulled. Anything unsupported is removed.

It shows its work. You can open the records behind the answer and follow anything important back to the official government source.

The hallucination shield

Ask a general-purpose AI about a federal contractor and it will give you a confident, readable answer whether or not it knows anything. In federal contracting, a confidently wrong claim about a company's set-aside status or past performance is the expensive kind of wrong.

So FedScope puts a check between the records and the answer. It is ordinary code, not a polite instruction to the AI, which means it does not depend on the AI choosing to behave:

  • Every claim in a draft answer must trace back to a record that was actually retrieved.
  • An answer that traces back to nothing is blocked — you never see it.
  • An answer that is only partly supported is trimmed — the unsupported parts come out, the evidenced parts stay.
  • When the record simply cannot answer your question, FedScope tells you so instead of guessing.

You will notice this. FedScope will sometimes say it has no evidence, and that is the system working, not breaking. "No record found for this company" is more useful than an invented profile — and it is the reason the answers you do get are worth something, because you can check them.

What "verified" means here

"Verified" always points at one specific fact. It is never a judgment about how good a company is.

Depending on where you see it, verified can mean:

Where you see itWhat was verified
A company identityIt is tied to an official federal identifier
An awardIt appears in an official government record
Shared performance evidenceIt passed the access and ownership checks
The Verified badge on a profileFedScope confirmed control of the profile through the company's official government-record channel — see claiming your company

Verified does not mean FedScope endorses the company, guarantees it is eligible today, vouches for anything the company says about itself, or predicts how it will perform.

Why one company ranks above another

Five principles decide the order of results:

  • Related work beats a big portfolio. Closely matching work counts for more than a large pile of unrelated contracts.
  • Proof beats labels. A real award history is stronger evidence than a company describing itself as capable.
  • Official IDs beat similar names. Two companies can share a name. FedScope prefers the official identifier when deciding which records belong to whom.
  • Recent, but with judgment. Newer work usually matters more, though older work can still be the best evidence for a specialized skill.
  • Show the basis, never fake precision. FedScope would rather show you what an answer rests on than imply a certainty the record cannot support.

You are still the one deciding

FedScope narrows the research and lets you inspect the evidence. It does not make the call. Confirming what a solicitation requires, contacting companies, judging conflicts and capacity, and deciding whether to bid, team, price, or sign — those stay with you and your advisers.

If an answer looks wrong or incomplete, tell us, and include the record or result you are looking at. Fixing it at the source helps everyone, which is better than working around it on your own.

Beta: the method is settled, the coverage is still growing

How FedScope reasons about evidence is not going to loosen. What is still expanding is how much of the record it covers and how many surfaces it answers on. See what FedScope cannot tell you.

On this page