Clear inputs. Defensible decisions.

• Industry and category. Matches the approved industry, subcategory, or operating context in the ICP.
• Business model. Has the customer, revenue, and delivery model relevant to the proposed product.
• Channel and market. Operates in the geography, channel mix, and market scope the profile permits.
• Account identity. The company is distinct, tenant-scoped, and not a duplicate product evaluation.

Evidence standard: Use a retained account record or a source-backed profile fact. Do not infer a profile match from a similar-looking brand.

Hold when: Hold when a mandatory profile field is absent, conflicts with the ICP, or cannot be attributed to a source.

• Cohort definition. Peer group, geography, channel, time window, and metric definitions are specified before comparison.
• Economic metric fit. AOV, purchase frequency, retention, or another metric is relevant to the selected product mechanism.
• Comparator quality. Ranges are treated as comparators, with observed account facts kept separate from estimates.
• Range treatment. Threshold decisions use the defined range and do not cherry-pick a midpoint.

Evidence standard: Record the cohort, observation window, and whether each input is observed, benchmark-assisted, or estimated.

Hold when: Hold when there is no comparable cohort, the metric is undefined, or the account result cannot be distinguished from a benchmark assumption.

• Value mechanism. States how the product could affect revenue, retention, conversion, efficiency, or another explicit commercial lever.
• Reach and adoption. Identifies the relevant eligible population and a bounded adoption or activation assumption.
• Contribution and cost. Includes margin, implementation, operating cost, and horizon where material to the decision.
• Sensitivity. Shows which uncertain inputs most affect the outcome rather than presenting a single unsupported ROI number.

Evidence standard: Use an attributable assumption set and label estimates. A preliminary model is not a customer-validated business case.

Hold when: Hold when the value mechanism is vague, material assumptions are hidden, or a required input is outside an approved estimate policy.

• Stack compatibility. Required systems, integrations, data flows, or ecosystem dependencies appear feasible.
• Channel coverage. The account operates in the channels where the product can create the proposed value.
• Implementation conditions. Operational ownership, capabilities, and likely delivery constraints are considered before recommendation.
• Access and rights. Data access, permissions, and customer rights are sufficient for the intended use case.

Evidence standard: Keep technical observations separate from confirmed implementation commitments and cite the source or customer-provided context.

Hold when: Hold when a required integration, access right, channel, or operating capability is absent or unknown.

• Signal provenance. Each material signal has a source, capture date, and concise description of what was actually observed.
• Relevant change. The event connects plausibly to the selected product mechanism rather than a generic activity metric.
• Recency and context. Timing, scope, and whether the event is active, historical, or speculative are stated.
• Conflict review. Conflicting evidence, uncertainty, and alternate explanations are recorded instead of being silently discarded.

Evidence standard: Treat signals as research inputs. They may prioritize a question for buyer discovery but do not establish purchase intent.

Hold when: Hold when the signal is unattributed, stale, unrelated to the use case, or contradicted by stronger evidence.

• Business need. A specific operational or commercial problem is plausible for the account and relevant to the product.
• Compelling event. An expansion, launch, budget cycle, leadership change, project phase, or other event may create a time-bound reason to explore.
• Planning horizon. The possible action window is stated with uncertainty—such as a next-six-month research hypothesis—not asserted as a committed purchase date.
• Actionability. The team can name a useful next question or research step rather than routing an unexplained timing score to sales.

Evidence standard: Retain the source and limitation for both the need hypothesis and the event. They must be distinguishable from intent claims.

Hold when: Hold when there is no relevant need, no plausible timing context, or no responsible next action beyond generic outreach.

• Need and timing. The evidence-led need hypothesis and timing context are documented for a buyer conversation.
• Budget status. Budget is labeled unknown, inferred, or confirmed; it is not represented as confirmed before buyer engagement.
• Authority status. Likely stakeholders may be mapped, while actual authority remains subject to confirmation with the account.
• Next action. A recommended research or conversation step is specific, appropriate, and reviewable before outreach.

Evidence standard: Preserve the distinction between research context and buyer-validated qualification. Route only with a visible evidence and limitation trail.

Hold when: Hold when the recommended action is unsupported, the evidence trail is incomplete, or commercial facts are presented as confirmed without buyer validation.