How to Separate a Company's Own Claims From Independent Evidence
Summary
- Separate what a company says about itself from what independent sources can confirm, using a consistent record format.
- Capture four layers for every claim: original text, paraphrase, your interpretation, and verification status.
- Classify evidence by independence and strength (e.g., company-controlled, third-party but paid, truly independent, primary records).
- Use a worked example to practice turning marketing language into checkable sub-claims.
- Maintain a reusable “claim log” so future readers can see exactly what is known, unknown, and why.
When you read a company website, pitch deck, press release, or product page, you are reading the company’s own narrative. That narrative can be useful, but it is not the same thing as independent evidence. The practical skill is not “distrust everything,” but to separate (1) what the company claims, (2) what you think it means, and (3) what outside sources can actually confirm.
This article gives you a concrete, repeatable method: a simple evidence-handling workflow, a worked example, and a reusable record format you can copy into a doc, spreadsheet, or notes system.
What counts as a company’s “own claims” vs independent evidence?
Company’s own claims (company-controlled)
These are statements the company controls or can edit at any time. They can be accurate, but they are not independent.
- Company website pages, blog posts, help docs, and landing pages
- Press releases and “newsroom” announcements
- Case studies hosted by the company (even if they quote a customer)
- Founder interviews on the company’s channels
- Testimonials selected and presented by the company
- Marketing decks, sales one-pagers, and webinars run by the company
Independent evidence (not controlled by the company)
Independent evidence is created and maintained by parties who are not under the company’s control. Independence is a spectrum, so it helps to record how independent something is.
- Primary records: filings, court records, regulatory actions, patents, procurement records, public datasets, or other official documents (where relevant)
- Independent reporting: journalism or analysis where the publisher controls the content and can challenge claims
- Third-party technical validation: reproducible tests, benchmarks, or audits published by an external party (when you can inspect methods)
- Independent customer evidence: reviews or write-ups not commissioned or edited by the company (still evaluate incentives and verification)
“Third-party” that is not fully independent
Some sources look independent but may still be influenced by the company. They can still be useful, but you should label them clearly.
- Sponsored content, paid placements, affiliate reviews
- Partner announcements and co-marketing pages
- Analyst notes commissioned by the company
- Influencer posts with compensation or free product
The core task: turn marketing statements into checkable sub-claims
Company claims are frequently bundled: one sentence can contain multiple assertions. Your job is to split them into smaller pieces that can be verified separately.
Example of a bundled claim (illustrative):
- “We are the industry leader, trusted by thousands, and our platform reduces costs by 30%.”
That single line contains at least three different claims:
- Market position: “industry leader” (requires a defined market and a measurable leadership metric)
- Adoption: “trusted by thousands” (requires a definition of “trusted” and a countable unit: customers? users?)
- Impact: “reduces costs by 30%” (requires a baseline, timeframe, and measurement method)
Once split, each sub-claim can be assigned its own evidence needs and verification status.
A reusable record format (copy/paste template)
Use a consistent “claim log” so you can show your work without turning it into a debate. The key is to keep four layers distinct:
- Original text (exact words you saw)
- Paraphrase (same meaning, simpler wording)
- Interpretation (what you think it implies in practice)
- Verification status (what is confirmed, by what kind of source, and what remains unknown)
Claim log template
| Field | What to record |
|---|---|
| Claim ID | A unique label (e.g., C-001) |
| Claim topic | Short category (e.g., “security,” “performance,” “customers,” “pricing,” “compliance”) |
| Original text (verbatim) | Copy the exact sentence(s) as written, including qualifiers |
| Source type | Company-controlled / Third-party (paid/partner) / Independent / Primary record |
| Where found | Page title + date accessed (and URL if you are storing links) |
| Paraphrase | Rewrite plainly without adding new meaning |
| Interpretation (what it implies) | What a reasonable reader might conclude; list assumptions explicitly |
| Sub-claims | Break into checkable parts (S1, S2, S3...) |
| Evidence needed | What would confirm or refute each sub-claim (documents, tests, records) |
| Evidence found | List independent items you located (with brief notes) |
| Verification status | Confirmed / Partially confirmed / Unconfirmed / Contradicted (and why) |
| Confidence notes | Limits, ambiguity, conflicts, and what would change your conclusion |
Tip: Keep “verification status” about the claim, not about whether you like the company. You can be neutral and still be rigorous.
Worked example: separating claim, paraphrase, interpretation, and verification
Below is a fully worked example using a fictional company and fictional wording. The goal is to demonstrate the method without relying on real-world sources or quotes.
Example claim record (fictional)
| Field | Example entry |
|---|---|
| Claim ID | C-014 |
| Claim topic | Security & compliance |
| Original text (verbatim) | “AcmeCloud is enterprise-grade and compliant with all major standards.” |
| Source type | Company-controlled |
| Where found | Security page (date accessed: YYYY-MM-DD) |
| Paraphrase | AcmeCloud says it meets enterprise expectations and follows widely used compliance standards. |
| Interpretation (what it implies) | A buyer might assume: (1) there is third-party validation, (2) specific standards are covered (not just internal policies), and (3) documentation exists that a procurement team can review. |
| Sub-claims |
S1: “Enterprise-grade” corresponds to specific controls (uptime targets, access controls, audit logs, incident response). S2: “Compliant” means verified against named standards (not just “aligned with”). S3: “All major standards” is a defined list (e.g., which ones?) and coverage is current. |
| Evidence needed |
For S1: a security whitepaper with concrete controls; customer security questionnaire responses; independent reliability data (if available). For S2: third-party attestation letters or certificates that name the standard and scope. For S3: a list of standards with dates, scope, and exclusions; evidence each is current. |
| Evidence found |
Independent evidence: none recorded yet. Company evidence: marketing page statement only. |
| Verification status | Unconfirmed (claim is broad; no independent documentation recorded) |
| Confidence notes | “Enterprise-grade” and “all major standards” are undefined. Next step is to request a list of standards and scope documents, then verify whether any attestations are third-party and current. |
What this example teaches
- Original text stays untouched, even if it is vague.
- Paraphrase removes hype words but does not add new facts.
- Interpretation makes assumptions explicit so they can be tested.
- Verification status is about evidence you have in hand, not what might exist.
Verification statuses that stay honest (and useful)
Use a small set of statuses so your log remains readable. Here is a practical set that works for most situations:
- Confirmed: independent evidence directly supports the claim as written (including scope and timeframe).
- Partially confirmed: some sub-claims are supported, but others are missing, narrower than stated, or time-limited.
- Unconfirmed: you have only company-controlled statements, or evidence is too weak to rely on.
- Contradicted: credible independent evidence conflicts with the claim.
To keep this rigorous, tie the status to the sub-claims. A broad claim can be “partially confirmed” when only one part checks out.
How to judge independence and strength without overcomplicating it
1) Independence checklist
- Control: Can the company edit or remove it?
- Incentives: Was the author paid, sponsored, or otherwise compensated?
- Method: Does it show how conclusions were reached, or is it just an opinion?
- Specificity: Does it name scope, dates, versions, and definitions?
- Reproducibility: Could another party repeat the test or verify the record?
2) Strength ladder (simple version)
- Weak: vague praise, anonymous testimonials, uncheckable superlatives (“best,” “leading,” “world-class”)
- Medium: specific claims with some documentation, but still company-provided or method-light
- Strong: independent documentation with clear scope, dates, and methods; primary records where applicable
This is not about being cynical. It is about making sure your conclusions match the quality of what you can verify.
Common traps (and how to rewrite them into checkable claims)
Trap: Superlatives without a defined comparison set
Claim: “#1 platform”
Rewrite into checkable sub-claims:
- #1 by what metric (revenue, users, deployments, satisfaction)?
- Within what market definition and geography?
- Measured when, and by whom?
Trap: “Trusted by” without a unit
Claim: “Trusted by thousands”
Rewrite:
- Thousands of what: paying customers, active users, newsletter subscribers, downloads?
- Over what timeframe?
- Is there independent confirmation (e.g., public customer list, procurement records, independent reviews)?
Trap: Percent improvements without baseline and method
Claim: “Cuts costs by 30%”
Rewrite:
- Costs for whom (which customer type) and which cost category?
- Compared to what baseline (previous tool, manual process)?
- Over what period, and how measured?
A lightweight workflow you can reuse in real work
- Capture the claim verbatim and label the source type (company-controlled vs independent).
- Split into sub-claims that can be verified independently.
- Write a neutral paraphrase that removes hype but keeps meaning.
- Write your interpretation as explicit assumptions (so you can test them).
- List evidence needed for each sub-claim (documents, records, tests).
- Record evidence found with short notes on independence and scope.
- Assign verification status (confirmed/partial/unconfirmed/contradicted) and explain in one sentence.
- Update over time when new evidence appears; keep old entries for traceability.
How to write conclusions that do not overreach
Once you have a claim log, your final write-up should reflect the evidence quality. A practical pattern is:
- What the company claims: one sentence, neutral.
- What you can independently confirm: one sentence, scoped (dates, versions, limits).
- What remains unknown: one sentence, with the next verification step.
Example conclusion sentence (template): “The company states X. Independent documentation confirms Y (scope/timeframe). The broader claim Z remains unconfirmed without A and B.”
Frequently Asked Questions
FAQ 1: What is the simplest way to tell a company claim from independent evidence?
Answer: Ask one question: “Can the company edit, curate, or remove this information?” If yes, treat it as a company-controlled claim. If no, and the publisher controls the content and method, it can qualify as independent evidence (still check incentives and scope).
Takeaway: Control is the fastest first filter.
FAQ 2: Are customer testimonials considered independent evidence?
Answer: Testimonials hosted on a company site are company-controlled, even if the words came from a customer, because selection and presentation are curated. A customer’s own public write-up can be more independent, but you should still note whether it was sponsored, whether the customer is identifiable, and whether the claims are specific and checkable.
Takeaway: Treat testimonials as signals, then look for confirmable details.
FAQ 3: How do I handle vague phrases like "enterprise-grade" or "industry-leading"?
Answer: Convert them into sub-claims with definitions. For “enterprise-grade,” list the concrete controls you would expect (for example: access management, auditability, incident response, uptime targets). For “industry-leading,” require a defined market and a measurable metric. If the company does not define terms, record the claim as broad and keep verification status narrow.
Takeaway: Replace vague labels with measurable requirements.
FAQ 4: What is the difference between paraphrase and interpretation in a claim log?
Answer: A paraphrase restates the claim in simpler language without adding meaning. Interpretation captures what a reader might infer in practice, including assumptions. If your sentence introduces a new detail not present in the original, it belongs in interpretation (and should be marked as an assumption to verify).
Takeaway: Paraphrase preserves meaning; interpretation exposes assumptions.
FAQ 5: How should I label evidence that is third-party but sponsored or affiliate-based?
Answer: Label it as “third-party (paid/partner)” rather than “independent,” and note the incentive (sponsored post, affiliate links, paid placement, free product). You can still extract checkable facts from it, but you should avoid treating its conclusions as neutral validation without additional support.
Takeaway: Keep “third-party” separate from “independent.”
FAQ 6: What verification status labels should I use to avoid overstating certainty?
Answer: A small set works well: Confirmed, Partially confirmed, Unconfirmed, and Contradicted. Tie the status to the sub-claims and scope (dates, versions, geography). If only one part is supported, mark it partially confirmed and specify which part.
Takeaway: Use few labels, but be precise about scope.
FAQ 7: How do I document contradictions when sources disagree?
Answer: Record each source separately, note what exactly conflicts (which sub-claim), and compare scope and timing (a claim can be true in one timeframe and false later). If you cannot resolve the conflict, keep the claim status as “Unconfirmed” or “Contradicted” depending on which evidence is stronger, and write what additional evidence would settle it.
Takeaway: Contradictions are manageable when you track scope and strength.
FAQ 8: How detailed should my evidence record be for internal decision-making?
Answer: Detailed enough that someone else can understand (a) what was claimed, (b) what you checked, and (c) why you reached your status. For high-stakes decisions, include dates, scope, and the exact wording of key claims. For low-stakes choices, a lighter log with sub-claims and a short evidence note may be sufficient.
Takeaway: Match the record depth to the decision risk.