Discriminator vs. feature: which one moves the score
Two real RFP sections, same underlying win theme, different framings. Which framing evaluators picked, and the rule I extracted from comparing them.
A discriminator and a feature are not the same thing. Most proposal teams know this. Most proposal drafts still blur the two, because under deadline the list of features the team wants to mention is long and the discipline of narrowing to discriminators is hard. This post is about what the difference actually costs on the scoreboard.
Two real RFP sections, from two different bids I was involved with. Same underlying capability. Different framing. Different outcomes. The identifying details have been changed; the framings are unchanged.
The capability
Both bids asked about the vendor’s ability to handle a specific regulated workflow. The vendor in question — call them Vendor A — had built a compliance posture for this workflow that most competitors had not built yet. The capability was real. The question was how to write about it.
Framing 1 — the feature version
Bid 1 (earlier in the year) wrote the section as a feature list:
“Our platform supports the [regulated workflow]. Features include: automated audit logging, role-based approvals, retention policies configurable per tenant, encryption at rest and in transit, SSO integration. The solution is built on a compliance-first architecture with SOC 2 Type II attestation.”
Every claim in that paragraph is true. Every claim is also something a technically-capable competitor could plausibly make. The evaluator reads the paragraph and mentally files the vendor in a bucket of “vendors who have this.”
Score on this section, per the debrief: middle of the pack. No penalty, no differentiation.
Framing 2 — the discriminator version
Bid 2 (three months later, different buyer, same underlying capability) wrote the section differently:
“The regulation changed in June 2024. It required a specific audit-trail structure that most SaaS vendors have not yet implemented — our analysis of the vendor landscape (see Appendix C) shows that of the [named set of comparable vendors], two have shipped the required structure and four have not. We shipped it in Q3 2024 and have been operating under it since. This section describes the specific structure, the audit evidence we can furnish, and the operational cadence that maintains it.”
Same capability. Different framing. The claim is specific and auditable. The comparison to the competitive set is named and sourced. The evaluator cannot file this into the “has it” bucket — the paragraph is telling them most of the competitive set does not have it, which is the point.
Score on this section, per the debrief: top of the pack, specifically cited as a differentiator.
What changed between the two
Three things.
Specificity about time. “Since Q3 2024” is verifiable. “Compliance-first architecture” is not. Evaluators filter unverifiable claims as boilerplate; they credit specific ones.
A comparison to the named competitive set. PropLibrary’s swap test is the diagnostic for this. Framing 1 passes the swap test — any competitor could write it about themselves. Framing 2 doesn’t — a competitor who shipped the structure in 2025 could write a similar paragraph, but the vendors who haven’t shipped it cannot write it at all.
Evidence in the appendix. Appendix C — a real appendix in Bid 2 — named the vendors and the status. An evaluator could verify the claim without taking the vendor’s word. Shipley’s guidance on discriminators emphasizes the evidence chain: a discriminator without evidence is a claim; a discriminator with evidence is a score mover.
The rule
A feature describes something your product has. A discriminator describes something your product has that the competitive set does not have, or has to a lesser degree, with evidence. If you can’t articulate the “that the competitive set does not have” part, you have a feature, not a discriminator. If you can articulate it but can’t evidence it, you have a claim, not a discriminator.
The rule is not “stop writing about features.” Features belong in the response — the compliance matrix needs them, the technical-approach section needs them, past-performance entries describe them. The rule is: the win-theme framing of the section needs to be a discriminator. The feature language supports the discriminator; it does not replace it.
The pressure that breaks this rule
The rule breaks under three pressures.
First, SME pressure. The engineer who owns the capability is proud of all its features. Asked to contribute a section, the engineer produces a feature list. The proposal writer who should rewrite that into discriminator framing is, under deadline, more likely to light-edit the feature list and ship it.
Second, review-committee pressure. A color-team reviewer who also happens to be a product marketer asks “why didn’t you mention [feature]?” during pink team. The writer adds the feature. Three reviewers add three features. The section bloats from a discriminator paragraph into a feature list paragraph and the discriminator gets diluted.
Third, legal pressure. Legal removes comparative language (“two of five named competitors have not shipped this”) because it’s adjacent to competitive claims that legal doesn’t have time to vet. The comparative framing is what makes the paragraph a discriminator; without it, the paragraph is a feature list with an extra date on it.
What to do about the pressure
The discriminator framing is protected by the capture plan, not by the draft. If the capture plan names three discriminators with evidence pointers, the draft’s job is to faithfully translate them into the response. The SMEs, reviewers, and legal can push back on specifics but they cannot easily argue the discriminators shouldn’t be in the response — the capture plan said they should.
Without the capture plan, the draft is where the discriminator work happens, and the draft under deadline is where the discriminator work fails. See the 8-stage RFP pipeline on why Stage 3 (capture) is the leverage point for Stage 5 (draft).
One more thing
Three months separates the two bids in this post. The capability did not change in those three months. Only the framing did. The framing change was the product of a capture plan in Bid 2 that wasn’t written in Bid 1. The discipline is where the score lives.
For the library-side work of keeping discriminator themes sharp across a book of pursuits, see win-themes-the-swap-test and win-themes-not-value-props.