Known, Inferred, or Unknown? A Buying-Role Evidence Worksheet

Who this worksheet is for

This worksheet is for account researchers, sellers, managers, and operations teams responsible for buying-group data that another person must be able to review. It is especially useful when sources differ, people hold several roles, or the map contains important unknowns. It supports transparent research; it does not authorize collection of unnecessary personal data or replace applicable privacy, security, legal, or retention controls.

Contents

1 Write one testable claim per row 2 Apply the evidence states 3 Separate observation from inference 4 Assess source quality and scope 5 Record contradiction and freshness 6 Set confidence without hiding unknowns 7 Complete and review the worksheet 8 Illustrative example 9 Blank copy-ready worksheet 10 Common mistakes 11 What this does not mean 12 Common questions 13 Sources and methodology

1. Write one testable claim per row

Avoid composite claims such as: Morgan is the engaged decision-maker and can approve the solution. That sentence contains at least four separate claims: 1 Morgan's identity and current title are accurate. 2 Morgan is attending to this decision. 3 Morgan performs one or more relevant functional roles. 4 Morgan has a specified form of authority within this scope. Create separate rows so each claim can have its own evidence state. A role can be Known while authority remains Unknown. Interest can be Known while sentiment is conditional. A public title can be Known while current employment is stale or contradicted. Use this claim syntax: For [account/site, application, purchase situation, stage] as of [date], [person/team] [performs role, holds authority, influences criterion, provides access, or controls condition]. The claim should be specific enough that new evidence could support, limit, or disprove it.

2. Apply the evidence states

Use five states: State Definition What not to do Known Direct, current, sufficiently reliable, and scope-matched evidence supports the exact claim Do not extend it to another role, site, stage, or authority State Definition What not to do Inferred Evidence makes the claim plausible, but direct confirmation is absent or incomplete Do not present it as fact or use it to bypass a gate Unknown The claim has not been established Do not convert to false, zero, no interest, or not applicable Contradicted Credible evidence materially conflicts on the same claim and scope Do not average the sources or select the convenient one silently Not applicable A documented rule or scope decision shows the claim or role does not apply Do not use this merely because no evidence was found The U.S. GAO's data-reliability guide assesses whether data are accurate, complete, and applicable for the intended purpose and uses a risk-based method. U.S. GAO, Assessing Data Reliability Its audit context differs from commercial account research, but its framing helps distinguish a true fact from information that is incomplete or inapplicable to the specific claim. States describe evidence, not people. Unknown does not mean the stakeholder lacks a role.

3. Separate observation from inference

Every Inferred row must include both:

• Observation: the source fact, statement, or action.

• Inference logic: why that observation could support the scoped claim. Example: Element Entry Observation Lee issued the dated technical acceptance checklist for the named retrofit Inference Issuing the checklist suggests Lee contributes to Technical Evaluation Limitation It does not establish final acceptance or contract authority Disconfirming evidence Current project matrix assigns checklist ownership and acceptance to another team The BLS explains that workers are classified by job duties rather than titles and that identical titles can refer to different occupations. BLS, How Do I Find My SOC Code? Because even occupational classification requires work context, a title alone should not be promoted to a Known buying-role claim. When direct action is observed, the role contribution may become Known while adjacent claims remain Inferred or Unknown.

4. Assess source quality and scope

Evaluate four dimensions: 1 Origin: first-party, official, direct statement, primary research, authorized document, or secondary description. 2 Specificity: named person, team, site, project, application, and decision condition versus general company information. 3 Freshness: publication, effective, observation, and access dates. 4 Applicability: whether the source actually supports the exact claim and scope. NIST's information-quality standards describe utility, integrity, and objectivity and state that limitations, assumptions, collection methods, and uncertainties in third-party information should be considered and disclosed. NIST, Information Quality Standards Those standards govern NIST information. The editorial translation here is to preserve provenance and limitations rather than stripping context from a source. SEC EDGAR provides public access to company filings and can support first-party facts about U.S. reporting companies, officers, segments, operations, facilities, or disclosed risks. SEC, Search Filings A filing's silence about a person or purchasing role does not prove absence. A parent-company officer list does not establish plant-level participation.

5. Record contradiction and freshness

A contradiction exists only when credible sources address the same claim and scope but materially disagree. Different scopes are not necessarily contradictions. Example:

• Source A says plant procurement manages purchase orders.

• Source B says corporate procurement signs contracts above a stated threshold. These can both be true because the actions and limits differ. Rewrite the claim before declaring conflict. For a true contradiction, record:

• both sources and dates;

• exact conflicting statements;

• possible scope, time, or terminology explanation;

• source-precedence rule, if legitimate;

• owner and next reconciliation action;

• Hold status for any dependent decision. Freshness is claim-specific. A certification may have an expiry. A public title can change. A directly confirmed process can be superseded by a new project stage. Set review_by or an event trigger rather than calling evidence permanently verified. NIST SP 1326 frames supplier due diligence as research into pertinent supplier and product information for informed decisions, within an ICT supply-chain scope. NIST SP 1326, NIST Cybersecurity Supply Chain Risk Management: Due Diligence Assessment Quick-Start Guide It reinforces purposeful research and source context, not a general buying-role standard.

6. Set confidence without hiding unknowns

Confidence can be Low, Moderate, or High for an Inferred claim, but it must never replace the evidence state. Use three transparent inputs:

• directness: how close the source is to the action or authority;

• scope match: how exactly it matches the account, site, application, stage, and claim;

• freshness: whether it remains current enough for the decision. Record the reasoning in words. Do not average Unknown into a numeric score. A high-confidence inference is still Inferred. A Known claim can have a narrow scope. A Contradicted claim cannot be rescued by a high average. Use a hard-gate rule: If a claim controls formal authority, a mandatory approval, a stop condition, suppression, or a critical dependency, an Inferred or Unknown state remains Hold regardless of confidence.

7. Complete and review the worksheet

Use these fields: Field Required entry Claim ID/version Stable reference and change number Exact claim One testable statement Decision scope Account/site, application, situation, stage, date Claim type Identity, title, role, influence, authority, access, interest, sentiment, condition, relationship Observation What the evidence directly shows Source and date Original source, issuer, publication/effective/access dates State Known, Inferred, Unknown, Contradicted, Not applicable Inference logic Required for Inferred claims Confidence and rationale Optional for Inferred; never a replacement state Limits Scope, freshness, authority boundary, missing context Contradicting evidence Source and exact conflict Disconfirming test Evidence that would weaken or change the claim Owner Accountable reviewer Next validation One bounded action Review trigger/date When the claim must be checked again Review dependent claims together. If title identity becomes Contradicted, role claims that rely only on that identity must also be held. Preserve version history rather than editing the earlier source out of view.

8. Illustrative example

Illustrative example - explicitly fictional. Lumen Process Controls, a fictional exporter, records ten claims about a buying group for one fictional plant upgrade. Claim type Known Inferred Unknown Contradicted Total Identity/title 2 0 0 1 3 Functional role 2 2 1 0 5 Authority/access 0 1 1 0 2 Total 4 3 2 1 10 Known claim share is 4 / 10 = 40%. Established or plausible claims, if reported together, are (4 Known + 3 Inferred) / 10 = 70%, but the two states remain separate. The Contradicted title claim and the Unknown authority claim are not scored as zero or folded into an average. The person with the contradicted title is also assigned an Inferred Risk & Process Validator role based only on that title. The dependency review moves that role claim to Hold pending identity reconciliation. A plant engineer is a Known Technical Evaluator because she issued the acceptance test, but her formal approval authority remains Unknown. Lumen's next actions are to verify the contradictory employment sources and ask the current project owner who signs technical acceptance. The fictional table describes evidence quality, not purchase intent or probability.

9. Blank copy-ready worksheet

Claim ID/ver sion Exact claim Decisi on scope Claim type Obser vation Sourc e/date State Infere nce logic Confid ence r ationa le Limits Contr adicti on/dis confir ming test Owner Next v alidati on Revie w trig ger/da te Review summary Check Result All claims have one scope and one statement Pass/Hold Every Known claim has direct, applicable evidence Pass/Hold Every Inferred claim exposes logic and limits Pass/Hold Unknowns remain visible and unscored Pass/Hold Contradictions preserve both sources Pass/Hold Hard-gate claims are Known or held Pass/Hold Owner and review trigger are present Pass/Hold

10. Common mistakes

• Combining title, role, support, and authority in one claim.

• Treating a public title as a Known buying role.

• Recording hearsay without the speaker, date, scope, or limitation.

• Converting Unknown into false, zero, or Not applicable.

• Calling different scopes contradictory.

• Silently deleting the inconvenient source in a true conflict.

• Letting confidence change an Inferred claim into Known.

• Keeping stale evidence without a review trigger.

• Using absence from a filing or website as proof of absence.

• Allowing a score to override an unknown authority or mandatory condition.

11. What this does not mean

This worksheet does not determine whether personal data collection is lawful, necessary, or proportionate. Teams must apply legitimate purpose, access, retention, security, and deletion controls. Do not collect sensitive or irrelevant personal information. Evidence quality does not establish intent, consent to contact, approval, qualified lead status, opportunity, or revenue. A Known role claim simply means the scoped function is supported by adequate current evidence.

12. Common questions

Can a direct statement make a role Known? Yes, when the speaker is positioned to know, the statement is clear, current, and scope-matched. Preserve exact context and limits. Is a company biography first-party evidence? It can establish a public title or responsibility as of its date. It does not prove participation in the purchase. How should confidence be scored? Prefer a written rationale based on directness, scope match, and freshness. Do not use confidence to hide the evidence state. When is Unknown changed to Not applicable? Only when evidence or a documented scope rule shows the role or claim does not apply, with the reason retained. Should every claim be rechecked on the same schedule? No. Set a review date or event based on volatility and consequence. Authority and suppression may require more immediate validation than a stable public product fact.

13. Sources and methodology

The evidence states, claim schema, confidence inputs, hard-gate rule, dependency review, and worksheet are SellersGPT editorial tools. Official sources provide scoped methods or records and do not define commercial buying-role truth. 1 BC10-GAO-RELIABILITY - U.S. Government Accountability Office, "Assessing Data Reliability." https://www.gao.gov/products/gao-20-283g Accessed August 24, 2026. 2 BC10-NIST-QUALITY - National Institute of Standards and Technology, "Information Quality Standards." https://www.nist.gov/director/nist-information-quality-standards Accessed August 24, 2026. 3 BC10-BLS-SOC - U.S. Bureau of Labor Statistics, "How Do I Find My SOC Code?" https://www.bls.gov/soc/finding_soc_code.htm Accessed August 24, 2026. 4 BC10-SEC-EDGAR - U.S. Securities and Exchange Commission, "Search Filings." https://www.sec.gov/search-filings Accessed August 24, 2026. 5 BC10-NIST-DILIGENCE - National Institute of Standards and Technology, "SP 1326, NIST Cybersecurity Supply Chain Risk Management: Due Diligence Assessment Quick-Start Guide." https://csrc.nist.gov/pubs/sp/1326/final Accessed August 24, 2026.