Skip to main content

Writing RFP Requirements Vendors Can Actually Answer

Part 4 of The Solicitation Author's Handbook. Why strong vendors skip weak solicitations, and how mandatory-versus-scored discipline changes what you receive.

August 14, 2026

This is Part 4 of The Solicitation Author’s Handbook, a vendor-neutral series for people in offices of government who write solicitations. Part 3 built the statement of work. This part is about the requirements apparatus around it, seen from the chair you never sit in: the vendor deciding whether to respond at all. Educational information, not legal advice.

The no-bid decision you never see

Here is the uncomfortable economics of your solicitation: a serious proposal costs a good vendor real money, often tens of thousands of dollars of senior people’s time. Vendors ration that spend. Our team reads public solicitations at scale, and the decision to skip one is usually made in the first twenty minutes, on signals authors rarely realize they are sending:

  • A requirements matrix where everything is mandatory, which reads as “we have not decided what matters.”
  • Requirements that contradict each other across sections, which reads as “stakeholders are still fighting, and the winner inherits the fight.”
  • A response window shorter than the document’s own page count seems to justify, which reads as “the winner is already chosen.”
  • No budget signal of any kind, which reads as “this may never be funded.”
  • Evaluation criteria vaguer than the requirements, which reads as “award will be defensible against nothing.”

None of these mean your agency intended anything of the sort. They mean the document is doing the talking, and the strongest vendors, the ones with the least need to gamble, quietly pass. What remains is whoever responds to everything. Writing answerable requirements is how you keep the strong ones at the table.

Mandatory, scored, informational

The single most useful discipline in requirements writing is sorting every requirement into one of three bins, and saying so in the document.

  • Mandatory means pass or fail: a response that lacks it is eliminated regardless of merit. Reserve this bin for the few things that are genuinely disqualifying, legal obligations, hard integrations, non-negotiable security posture. Every mandatory line is a door you are closing; count the doors.
  • Scored means better answers earn more points, and this is where competition actually happens. State what “better” means so vendors compete on the right axis.
  • Informational means you want to know but will not score it. Labeling these honestly stops vendors from padding thirty pages against a question worth zero points.

When everything is mandatory, nothing is, and vendors respond with a column of “comply” that tells you nothing. When the bins are honest, the responses sort themselves and your evaluators’ time goes where the decision lives.

The compliance matrix you would want to score

If you require a compliance matrix, design it as the instrument you will actually score with: numbered requirements in rows, the three bins marked, and a response column that demands substance (“describe how, with references to your proposal section”) rather than a checkbox. Cap narrative lengths per item. A matrix like this compresses evaluation from weeks to days and gives every debrief a paper spine. A matrix of eight hundred unweighted checkbox rows does the opposite, and everyone involved knows it.

Structure the response, kindly

Tell vendors exactly what to submit, in what order, with what page limits: an executive summary, a technical approach mapped to your numbered requirements, staffing with named key personnel, past performance with references you may actually call, and price in a separate sealed volume if your process requires it. Structure is a kindness in both directions: vendors stop guessing what you want, and your evaluators compare like with like instead of hunting through forty differently-shaped documents. Page limits are part of the kindness; they force prioritization and they respect your committee’s evenings.

Two structural habits worth adopting from the best solicitations we see: publish your scoring rubric alongside the requirements (Part 5 makes the case), and provide response templates as editable documents rather than locked PDFs, because you want vendors spending effort on substance, not on reconstructing your tables.

Past performance that predicts

Past performance questions predict best when they are specific and verifiable: engagements of similar size and kind in the last few years, with the outcome achieved, the reference’s name, and the permission to call. Generic “describe your experience” prompts collect marketing. And decide in advance what a new-but-capable vendor may substitute, key personnel history, for example, unless your genuine intent is to restrict to incumbents, in which case Part 3’s honesty rule applies: say so and defend it.

Questions worth asking about AI, since everyone is buying some

Technology solicitations now routinely involve AI components, and the requirements that matter are refreshingly old-fashioned: What decisions does the system make or recommend, and where does a human review them? What data trains or tunes it, who owns that data, and does the vendor use your data beyond your service? How are errors measured, reported, and corrected? Can the vendor explain a specific output when an affected member of the public asks? These are outcome requirements like any other, and they belong in the scored bin, where the quality of the answer tells you who you are dealing with.

The requirements checklist

  • Every requirement sits in a declared bin: mandatory, scored, or informational, and the mandatory bin is short.
  • The compliance matrix is the scoring instrument, not an afterthought.
  • Response structure, order, and page limits are specified, with editable templates.
  • Past performance questions are specific, verifiable, and fair to non-incumbents.
  • Someone has read the document end to end hunting contradictions between sections.
  • The response window matches the effort the document demands.

Part 5 crosses to the other side of publication day: evaluation criteria, scoring rubrics, and running a committee whose award survives both protest and the debrief conversation.

The Solicitation Author’s Handbook is published by GovSoft as a public resource for offices of government. It is educational information, not legal advice.

Let's Talk