RFP Scoring Scale Definitions: Writing Anchors for a 1-to-5 or Adjectival Scale

When three evaluators read the same 3 three ways

A committee chair at a small nonprofit is preparing for the consensus meeting. Three evaluators scored the same proposal on technical approach. One gave it a 2, one a 4 and one a 5. When the chair asks about the gap, the evaluator who gave the 2 says a 3 should be "average compared with the others," and this proposal was below that. The evaluator who gave the 5 says a 3 means "meets the requirement," and this one clearly did more.

Neither is wrong. Nobody wrote down what a 3 meant, so each evaluator used a private definition. The meeting now spends its time arguing about the scale instead of the proposal.

Months later, a monitor or an unsuccessful offeror may ask how the ranking was reached. The scoresheets show numbers, but the file cannot show what those numbers meant. This article explains how to write a definition for every point on a numeric or adjectival scale. It also shows how to record that the definitions were fixed before anyone scored.

Why an undefined rating weakens the evaluation file

Under a federal award, a request for proposals must identify all evaluation factors and their relative importance (2 CFR 200.320(b)(2)(i)). An entity using the proposals method must also have written procedures for conducting technical evaluations and making selections (2 CFR 200.320(b)(2)(ii)). Weights tell evaluators how much a criterion matters. They do not tell evaluators what a given score on that criterion means.

Another provision of the Uniform Guidance, as revised effective October 1, 2024, also bears on the record. 2 CFR 200.318(i) asks for records detailed enough to show the history of each procurement, including the rationale for contractor selection or rejection. A selection rationale built on numbers with no shared meaning is hard to explain after the fact. Confirm the current text of both sections at eCFR.

GAO's Bid Protest Annual Report to Congress for fiscal year 2025 (GAO-26-900695, December 12, 2025) names unreasonable technical evaluation as the most prevalent reason for sustaining protests that year. Those figures describe federal agency procurements, not grant-recipient or local procurements. They do show where evaluation records tend to get challenged: at the point where a rating needs a reason.

Defined anchors do not make an evaluation correct. They give every evaluator the same yardstick and give the file a way to show what that yardstick was. For how the criteria and weights fit around the scale, see how to build RFP evaluation criteria.

Writing RFP scoring scale 1 to 5 definitions that evaluators can apply

An anchor is a written statement of what a proposal must show to earn a given rating on a given criterion. The following principles keep anchors usable.

• Describe the proposal, not the evaluator's impression. "Names every key role" can be checked. "Strong staffing" cannot.
• Define the middle first. Decide what "meets the requirement" looks like. Then write the points above and below it as steps away from that line.
• Make adjacent points distinguishable. If an evaluator cannot say why a proposal is a 4 and not a 3, the two anchors overlap.
• Write per criterion, not once for the whole scale. A generic "3 = acceptable" invites the same drift. The scale can share a structure, but each criterion needs its own wording.
• Keep each anchor to one or two sentences. Evaluators will read them again at the scoresheet.

Worked example: a staffing plan criterion

This is an illustration with invented wording, not a recommended standard. Assume the solicitation sets a minimum experience requirement for each key role.

Rating	Anchor
5	Names every key role. Each person exceeds the minimum experience, and the plan covers absences of key staff.
4	Names every key role. Each person meets the minimum, and the proposal shows at least one specific strength.
3	Names every key role. Each person meets the minimum, with no notable strengths or weaknesses.
2	One key role is unnamed or below the minimum, or no coverage plan is given.
1	Several key roles are unnamed or below the minimum, and the plan would not support the scope of work.

Two evaluators using this table can still disagree. They will be disagreeing about whether a résumé meets the minimum, which is a question the proposal can answer. They will not be disagreeing about what a 3 is.

Decide separately how to treat a proposal that does not respond at all. Many files handle that through a responsiveness check rather than a zero on the scale. Whichever route you pick, write it down before scoring.

Anchoring an adjectival scale

Some solicitations use words such as Exceptional, Good, Acceptable, Marginal and Unacceptable instead of numbers. The same problem applies. "Good" is a label, not a definition, and two evaluators can attach different meanings to it.

Treat an adjectival scale in the same way as an RFP scoring scale 1 to 5 definitions exercise:

1. Choose the labels and their order.
2. Write a definition for each label, per criterion where the criterion calls for it.
3. If the ratings will be converted to points for ranking, set the mapping in advance. For example, decide whether Good converts to 4 of 5 points, and write that down.
4. State whether evaluators may use only the listed labels or may score between them.

The conversion step matters more than it looks. If the mapping is chosen after the labels are in, the choice can change the ranking, and it looks like a choice made with the results in view. Fixing it first removes that question. The tie rule works the same way: choose it before scoring, not after.

Neither format is required by the Uniform Guidance, and this article does not treat any agency's scale as a standard. Pick the format your solicitation, funder and local policy allow, and define it fully.

Fixing anchors before scoring opens

Anchors need to be final before any evaluator sees a proposal. Fixing them early is also the best defense against evaluator bias in RFP scoring, because the definitions cannot be shaped by the proposals in front of the committee.

Several familiar biases show up when anchors are loose:

• Anchoring on a first impression. A strong first proposal becomes the unspoken standard for every one after it.
• Leniency or severity. Some evaluators rate high across the board, others low, and the spread reflects the evaluator rather than the offer.
• Central tendency. Without clear definitions, evaluators drift to the middle of the scale to avoid defending an extreme.
• Rescaling after the fact. An evaluator reads several proposals, then quietly changes what each number means.

Clear anchors do not remove these tendencies, but they give the committee something to check each rating against. A short calibration session before scoring helps. Have the committee read the anchors together and, if time allows, apply them to a sample response written for the purpose. Do not use a real offeror's proposal for this. That session also belongs in the file.

If an anchor turns out to be unworkable once scoring has started, do not edit it quietly. Record what changed, who decided, when, and why. Then have every evaluator rescore any affected criterion under the same revised wording. Changing the yardstick for some proposals and not others is the version of this problem that is hardest to defend.

Anchors also support independence. When each evaluator scores alone against the same written definitions, differences in ratings point to real differences in judgment about the proposal. They are easier to discuss in the consensus meeting that follows. For how that stage works, see evaluator scoring independence and consensus scoring for RFPs. For how anchors sit inside the scoresheet itself, see the RFP scoring matrix.

What to record, and where a workbook fits

Whatever tool you use, the file should let someone reconstruct the scale without asking the evaluators. For each criterion, record:

• the weight, with all weights totaling 100;
• the anchor for every rating point, in the final wording;
• the date and version when the anchors were fixed, and the date scoring opened;
• any later change, with who made it, why, and whether scores were redone;
• the mapping from adjectival labels to points, if one is used;
• the tie rule chosen in advance.

Spreadsheets and free templates can hold all of this. Their limit is structural: nothing forces every rating point to be defined before scoring, and the layout tends to be rebuilt for each solicitation. The RFP Evaluation Criteria Builder workbook is designed to record criteria, weights that total 100, and an anchor for each rating point in one consistent layout. It is available now; see the workbook page. If you want to hear when the evaluation workspace opens, you can join the waitlist. Plan details are on the pricing page. Scoring in either case stays with your evaluators. There is no automated scoring and no suggested ratings.

Your committee's roles and conflict-of-interest steps sit alongside this record. They are covered in the article on the RFP evaluation committee. Local policy, your funder and your state procurement rules may set additional requirements for scales and documentation. When state or local rules are stricter than the federal ones, follow the stricter rule, and confirm specifics with your awarding agency or pass-through entity.

RFPManagement records and computes. It does not give legal or procurement advice. The award decision stays with your organization.