Weighted Scoring for RFPs: How the Math Turns Ratings into a Ranking

A weight error that swaps the top two proposals

A grants manager at a grant-funded nonprofit finishes tabulating an evaluation late on a Thursday. The committee's scores are in, the spreadsheet shows a clear leader, and the award memo is half drafted. On Friday morning a colleague notices that the weights add up to 105, because a late-added criterion was never offset against the others.

She corrects the weights, and the top two offerors trade places. No evaluator changed a rating. The ranking moved because the arithmetic had been wrong from the start, and nobody can say which version the committee actually saw.

A funder's monitoring visit is three weeks away, and the file must show how the ranking was reached. A monitor will want the published weights, the rating definitions, the math and the order in which things happened. A sheet that lets weights drift cannot supply those on its own.

After reading, you will be able to show how each rating became points, why the weights were fixed when they were, and which factor moved the ranking.

The weighted scoring RFP formula, step by step

In a weighted scoring RFP model, each criterion carries a weight. Each evaluator rates the proposal on a defined scale, and the rating is converted to points. The usual conversion is weight × rating ÷ maximum rating. Add the points across criteria, and the total is out of 100 whenever the weights total 100.

Three conditions give that total any meaning. The weights must sum to exactly 100, and every rating point needs a written definition, called an anchor, fixed before anyone scores. The rule for combining several evaluators' ratings must also be stated in advance. The scoring matrix article shows how to lay out the anchors.

The worked example below uses illustrative numbers only and is not a recommendation of any weights. Four criteria carry weights of 35, 20, 15 and 30, and ratings run from 0 to 5. Price is converted with the lowest-price formula: lowest evaluated price ÷ this offer's evaluated price × maximum price points. Three offerors price at $100,000 (A), $80,000 (B) and $125,000 (C).

Worked example (illustrative numbers only)

Criterion	Weight	Offeror A	Offeror B	Offeror C
Technical approach	35	rating 5 → 35.0	rating 4 → 28.0	rating 3 → 21.0
Staff qualifications	20	rating 4 → 16.0	rating 4 → 16.0	rating 5 → 20.0
Past performance	15	rating 4 → 12.0	rating 3 → 9.0	rating 4 → 12.0
Price	30	$100,000 → 24.0	$80,000 → 30.0	$125,000 → 19.2
Total points	100	87.0	83.0	72.2

To check the method, read one cell. Offeror A's technical rating of 5 gives 35 × 5 ÷ 5 = 35.0 points, and Offeror C's price gives $80,000 ÷ $125,000 × 30 = 19.2 points. Offeror A leads with 87.0, followed by B at 83.0 and C at 72.2. For other ways to turn price into points, see the price-to-points formula article.

Why a total of 100? A 100-point scale lets any reader take each weight as a share of the score, so a 35-point criterion is plainly worth more than a 20-point one. Weights that total 105 or 97 still produce a ranking, but the percentages no longer match the points. A later reader then cannot tell whether the sheet or the solicitation is correct.

Two housekeeping choices belong in the written procedures. Decide how many decimal places you carry and where rounding happens, because rounding can create or hide a tie. Record whether the evaluated price includes alternates and options, because including them changes every offeror's price points.

Combining evaluators' ratings before the weights apply

When several evaluators score a proposal, the file needs one rating per criterion before the weights apply. The two common routes are an average of the individual ratings and a consensus meeting that agrees a single rating with a written reason. The solicitation's written procedures should name the route before scoring begins.

The conversion is linear, so averaging first and converting second gives the same total as converting each evaluator's ratings and averaging the totals. Rounding is the only place the two routes can differ. That is one more reason to fix the rounding rule in advance.

Averages can hide disagreement. In the example, Offeror B's technical rating of 4 could be the average of ratings of 5, 4 and 3, a spread of two rating points on the 0–5 scale. A variance check set at a threshold your organization chooses, such as two rating points, can prompt the committee to discuss that criterion. Such a check is informational only, and it should never change a score by itself.

Independence matters here as well. Evaluators who see one another's ratings before submitting their own may anchor on the initial number they read, or may be accused of it. Scoring first and meeting second, with the rationale for each final score written down, leaves a cleaner record.

What 2 CFR 200.320(b)(2) asks you to publish before proposals open

Under 2 CFR 200.320(b)(2)(i), as revised effective October 1, 2024 (89 FR 30046), a request for proposals requires public notice and must identify all evaluation factors and their relative importance. Proposals must also be solicited from multiple qualified entities. That makes rfp evaluation criteria weighting part of the published solicitation, not an internal choice made once proposals arrive.

Point or percent weights are a common way to state relative importance, and the awarding agency or pass-through entity can confirm what it expects to see. Choosing the factors themselves is covered in the RFP evaluation criteria article. Confirm the current text of the section on eCFR, which is authoritative but unofficial; the official edition is the annual CFR on govinfo.gov. State and local rules can be stricter, and when both apply, the stricter rule governs.

Section 200.320(b)(2)(ii) separately requires written procedures for conducting technical evaluations and making selections. A weighted scoring RFP method belongs in those procedures: the weights, the scale, the anchors, the combining rule and the tie rule. The award standard itself is in 2 CFR 200.320(b)(2)(iii), which says contracts must be awarded to:

"the responsible offeror whose proposal is most advantageous to the recipient or subrecipient considering price and other factors."

The weighted total organizes that judgment, and the award decision stays with the organization. Changing a weight after proposals are opened would leave the file showing a method different from the one offerors were told about. If a change is truly needed, local policy and the awarding agency govern how to amend the solicitation, and the record should show when and why.

Anchors follow the same logic. A rating of 4 should mean the same thing to every evaluator, so the definition for each point on the scale, for example what separates a 3 from a 4 on technical approach, is written into the file before any proposal is read.

Price is excluded as a selection factor only for architectural and engineering professional services, under 2 CFR 200.320(b)(2)(iv). There the entity evaluates qualifications, selects the most qualified offeror and then negotiates fair and reasonable compensation. State qualifications-based selection laws vary, so check your own state's statute.

E-Rate works differently. Under 47 CFR 54.503, bids are considered with price as the primary factor, so confirm with USAC and your state and local rules how to express that in the weights.

GAO's FY2025 bid protest annual report (GAO-26-900695, December 12, 2025) lists the top three grounds for sustained protests: unreasonable technical evaluation, unreasonable cost or price evaluation, and unreasonable rejection of a proposal. Those findings describe federal agency procurements, not grant-recipient or local procurements. They still show which parts of an evaluation record draw scrutiny when someone objects.

Testing which factor drives the ranking

A sensitivity test holds every rating fixed, moves weight from one factor to another in steps, and recomputes the totals. It answers a plain question about how much weight it would take to change the order. If a small shift reorders the top two, the ranking depends on the weights as much as on the evaluators' judgment.

The table continues the worked example, again with illustrative numbers only. Moving ten points from technical approach to price produces a tie between A and B. Moving fifteen points reverses their order, and C stays last in every row.

Weights (technical / qualifications / past performance / price)	Offeror A	Offeror B	Offeror C
35 / 20 / 15 / 30 (base case)	87.0	83.0	72.2
25 / 20 / 15 / 40	85.0	85.0	72.6
20 / 20 / 15 / 45	84.0	86.0	72.8

Offeror A's four-point lead at the base weights disappears once price carries 40 points. Offeror B is lowest priced, so every point moved from technical approach to price shifts the A–B gap by 0.4 points in B's favor. Seeing that figure helps the committee judge whether the published weights express what the program wants.

Run this test before the solicitation is released, using hypothetical proposal profiles, such as a strong technical offer at a high price against a weaker offer at a low price. If the weights produce an order the program would not accept, adjust them then. Do not tune weights once real proposals are in hand.

A table run after scoring can sit in the file, labeled informational, to show the committee understood what drove the result. It does not replace the published weights. The tie in the middle row is also a reminder to set a tie rule before scoring begins, and the tie-breaking procedure article covers the usual options.

Where spreadsheet formulas drift

Spreadsheets are free, familiar and flexible, and most small offices start there. A weighted scoring model in Excel needs only a few formulas. A common rfp evaluation spreadsheet formula for the non-price criteria is =SUMPRODUCT(weights, ratings)/max_rating, which returns the points earned on those criteria. Price usually gets its own row, such as =MIN(all_prices)/this_price*price_weight.

The limit is structural, not a fault in any template. The sheet is usually rebuilt for each solicitation, so weights, scales and formulas drift between files. Nothing forces the weights to total 100, and nothing technically stops a shared file from showing one evaluator's scores to another before everyone has submitted. The Excel scoring template article covers layout in more detail.

Whatever the tool, a few checks are worth building in:

• a cell that sums the weights and flags anything other than 100;
• weights, scale maximum and anchors locked once the solicitation is released;
• a price formula that names which prices are evaluated and which alternates and options are excluded;
• a stated rounding rule and a version date on the sheet;
• a saved copy of each evaluator's ratings made before any combined view is opened.

RFPManagement is built around those same checks. It is designed to require that weights total 100 and that every rating point is anchored before a round opens. It is also designed to hide scoresheets until the round is locked and to compute the ranking with a tie rule chosen in advance. Scoring stays with human evaluators; the software computes the arithmetic and keeps the record.

What to keep in the file, and where to go next

2 CFR 200.318(i), in the version effective October 1, 2024, requires records detailed enough to show the history of each procurement. They must include four elements: the rationale for the method, the contract type selection, the contractor selection or rejection, and the basis for the contract price. Scoring records speak mainly to the last two, and a file typically holds the following:

• the solicitation as published, with its date and the weights or relative importance it stated;
• the rating scale and the written anchor for every rating point;
• each evaluator's individual ratings, saved before any combined view;
• the combining rule, the rounding rule and the tie rule;
• the price formula, the evaluated prices, and how alternates and options were treated;
• the computed totals and ranking, with any sensitivity table labeled informational;
• the rationale for each final score and a note of who made the award decision.

Local public-records and retention rules may add requirements beyond the federal text, so confirm them with your pass-through entity or counsel. Keeping the sheet's version date next to the solicitation date lets a later reader reconstruct which weights applied at the time.

The RFP Evaluation Criteria Builder is a store workbook designed to hold criteria, weights that total 100 and a definition for every rating point in one reusable sheet. Check that page for its current status.

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