Discriminator tests: three worked examples

The APMP discriminator test is one paragraph. A capability is a discriminator if, and only if, your competitors cannot truthfully claim it. Anything else is a feature, a benefit, or boilerplate. Discriminators win bids; features and benefits do not.

The test is simple. The application is where the work is. Most claims that read like discriminators on a first draft are not — they are differentiators-in-degree (we are faster, more reliable, more experienced) rather than differentiators-in-kind (we are the only vendor who can do X, and here is the proof). Evaluators read the difference. The bid loses or wins on whether the claim survives the test.

Here are three sections from real-shape proposals (composited and anonymized — the patterns are real, the specific numbers are not), run through the discriminator filter.

Example 1 — the technical-approach claim

Before. “Our cloud-native architecture delivers superior reliability and performance through industry-leading auto-scaling and multi-region failover, ensuring your workloads remain available 24/7 with industry-leading uptime.”

Discriminator test, applied.

• “Cloud-native architecture” — true of every cloud vendor. Not a discriminator.
• “Superior reliability and performance” — adjective, not a measurement. Not a discriminator.
• “Industry-leading auto-scaling” — every competitor will claim auto-scaling. Not a discriminator.
• “Multi-region failover” — common feature. Not a discriminator.
• “Available 24/7” — table stakes. Not a discriminator.
• “Industry-leading uptime” — undefined claim. Not a discriminator.

Six claims, zero discriminators. The paragraph is technically true and competitively invisible.

After. “Your stated availability requirement is 99.99% (Section 3.2). Our production system has run at 99.997% measured uptime over the trailing 24 months across all customer-facing services, audited by [named third-party auditor]. Two of our four direct competitors do not publish measured uptime numbers; the third publishes a 99.9% target without a third-party audit. The audit report is appended as Attachment C.”

What changed. The before paragraph asserted reliability; the after paragraph proves it with a measured number, names the comparison frame, and provides verifiable evidence. The discriminator is not “we have uptime” — it is “we publish audited uptime that exceeds the buyer’s stated requirement, and our competitors do not.” That claim cannot be truthfully made by competitors. It is a discriminator.

The PropLibrary essay makes this point using a related test — the swap test — but the underlying logic is the same: if your claim survives a competitor swapping their name in, you have not differentiated.

Example 2 — the past-performance claim

Before. “With over 15 years of experience serving Fortune 500 customers, our team has the expertise and proven track record to deliver complex enterprise solutions on time and on budget.”

Discriminator test, applied.

• “Over 15 years of experience” — most competitors have more. Not a discriminator (and possibly anti-discriminating).
• “Fortune 500 customers” — true of every enterprise vendor in this space. Not a discriminator.
• “Expertise” — adjective. Not a discriminator.
• “Proven track record” — undefined. Not a discriminator.
• “On time and on budget” — assertion without evidence. Not a discriminator.

Five claims, zero discriminators, and the experience claim is actually weaker than the competitor’s likely claim.

After. “Three of our last five engagements at this scope and scale ($5M+ migrations from on-prem to cloud-native architectures) shipped on the original schedule with zero customer-reported regressions in the first 90 days post-cutover. The reference contacts for those three engagements are listed in Attachment D and have agreed to direct conversations with the evaluation panel. We are aware of two competitors who can credibly claim a similar pattern; their references are public knowledge through their case-study pages, and we encourage the panel to compare directly.”

What changed. The before paragraph asserted a track record; the after paragraph names a specific, narrow scope, gives a measurable outcome, provides reachable references, and concedes the competitive landscape honestly. The discriminator is not “we have past performance” — it is “we have past performance at this exact scope, with verifiable references, and we welcome direct comparison.” The acknowledgment that two competitors can match the claim is itself a credibility move; it tells the evaluator that the team has thought about the comparison, which is what an evaluator wants to see.

The narrowing matters most. “Fortune 500 customers” is not a discriminator. “Three of our last five engagements at $5M+ migrations from on-prem to cloud-native” is — because it is specific enough to be falsifiable. Discriminators are always more specific than the claims they replace.

Example 3 — the management-approach claim

Before. “Our seasoned project management team uses industry best practices to ensure smooth execution, proactive risk management, and timely delivery. We provide weekly status reports, monthly executive reviews, and a dedicated customer success manager to ensure your project stays on track.”

Discriminator test, applied.

• “Seasoned project management team” — every vendor will claim this. Not a discriminator.
• “Industry best practices” — undefined. Not a discriminator.
• “Smooth execution” — adjective. Not a discriminator.
• “Proactive risk management” — every PMO claims this. Not a discriminator.
• “Timely delivery” — assertion. Not a discriminator.
• “Weekly status reports / monthly executive reviews / dedicated CSM” — table stakes. Not a discriminator.

Six claims, zero discriminators. This is the kind of paragraph that loses three to five rubric points without the team noticing — every other vendor’s response says the same thing.

After. “Your stated risk for this engagement (Section 5.4) is data-migration timeline overrun. Our delivery model assigns a dedicated migration architect — not a project manager — as the named accountable owner for cutover dates. The architect has executive escalation authority directly to our VP of Engineering, named here as [name], whose calendar is committed to your project at 4 hours per week through cutover. We have made this commitment on six prior engagements; in five of those, cutover hit the original date. In the sixth, we extended cutover by two weeks at our cost. The reference for the sixth is listed in Attachment D and is willing to discuss what happened.”

What changed. The before paragraph offered a generic PMO model; the after paragraph names a specific risk from the buyer’s own RFP, names a specific role and a specific person responsible for that risk, makes a measurable commitment (four hours per week, escalation to a named VP), and proactively addresses the failure mode (the one cutover that did not hit the original date) with a reachable reference.

The discriminator is the named-VP-with-committed-calendar commitment. Most competitors will offer a CSM. Few will offer named executive accountability with committed weekly time. The credibility-extending move is the honest acknowledgment of the one cutover that overran — that is the move that makes the rest of the paragraph believable to an experienced evaluator.

What the three examples share

The before paragraphs are competently written. They contain no factual errors. A reader without an evaluator’s discipline reads them as professional and reasonable. They are also, in evaluator terms, score-neutral — they do not move the rubric in any direction because every responding vendor produces text in roughly the same shape.

The after paragraphs are aggressive in three ways the before paragraphs are not:

1. They name specific buyer-side language from the RFP (an availability requirement, a stated risk).
2. They make falsifiable claims with numbers, named references, and audit trails.
3. They concede what they cannot honestly claim, which builds credibility for the claims they do make.

The third point is the one that distinguishes a discriminator from an aggressive marketing claim. A claim that admits its limits is a claim an evaluator will trust. A claim that asserts superiority across every dimension is a claim that evaluators have learned to discount.

Applying the test

When you read your own draft, run the discriminator filter on every claim. For each sentence, ask:

1. Could a competitor truthfully make this claim?
2. If yes — delete the sentence or rewrite it as a discriminator.
3. If no — what is the evidence that proves the claim, and is it in the response?

If steps 1 through 3 leave you with fewer sentences than you started with, the draft is improving. Most discriminator-filter passes cut paragraph length by 30 to 50 percent. The remaining text does more work per word.

The Shipley method calls discriminators the “ghost in the machine” of a winning proposal. The discriminator is the thing that, by itself, can move the evaluator from “this vendor is competent” to “this vendor is the right choice.” It cannot do that work if it is buried in three paragraphs of features that every other vendor is also claiming. Cut the features. Surface the discriminators. Run the test on every sentence.