Build the Source and Evidence Register

Why this decision matters

A homepage supports what a company currently says on that page; it does not independently verify product performance. A pricing page supports the displayed offer when captured, subject to geography, cookies, account state, and taxes. Reviews support attributed customer statements and repeated themes only when the sample is described. Community posts support public discussion, not market prevalence or purchase intent. The register prevents one weak source from silently becoming a strong conclusion.

This lesson advances the same capstone used throughout the course: the Competitor Intelligence Monitor Pack. The learner is not collecting ideas for later. The workbook record created here becomes an input to the next module and must be specific enough that another operator could review it. Where evidence is incomplete, the artifact should show the gap instead of smoothing it over.

Source evidence and limits

C6-S1 distinguishes website, pricing, launch, review, and community lanes. C6-S2 requires source lineage and dated raw records. C6-S3 demonstrates source tiers and warnings against treating discovery as proof.

The sources support the operating method stated above. They do not prove universal performance, guaranteed savings, complete market coverage, causal impact, or a client result. Any date-sensitive product, platform, competitor, legal, pricing, or policy fact must be checked against a current primary source before publication or operational use.

Continuing fictional example

Example: HarbourOps is a fictional Toronto workflow consultancy monitoring three invented competitors: BeaconFlow, TaskHarbor, and RelayDesk.

HarbourOps selects each competitor's homepage, pricing page, and changelog. It records that pricing is volatile, changelogs are first-party claims, and homepages are positioning evidence. Reviews are deferred so the first version stays inside three lanes.

Example boundary: Every company, page, event, and result in this example is fictional. It demonstrates the method and is not observed market evidence.

Workbook application

Use Workbook Section 2: Build the Source and Evidence Register. Complete Workbook Section 2. For every URL, record source type, allowed retrieval method, expected fields, authority, retrieval cadence, cost, and limitation. Mark volatile claims that require rechecking before use.

1. Write the current evidence or input before adding interpretation.
2. Apply the lesson's decision rule and state the reason for the classification.
3. Mark uncertainty, missing information, and the human owner for the next decision.
4. Check that the result stays inside the course boundaries and can be tested.

Failure modes to inspect

Using search snippets as full-page evidence; comparing one competitor's pricing with another's reviews; omitting retrieval dates; retaining only summaries; and failing to distinguish publication date from access date.

A polished output can still fail if its evidence, permissions, identity, denominator, source date, or action boundary is wrong. Review the underlying record rather than grading tone alone. The correct repair may be to narrow the scope, gather a permitted source, label an unknown, or stop the proposed action.

Decision rules

A source supports only what was actually read. Self-reported claims remain attributed. Search and community results are discovery until corroborated. Preserve raw or normalized records before interpretation.

Record the rule in the workbook in language two reviewers can apply consistently. A rule that depends on intuition alone cannot support a deterministic fixture. If reviewers disagree, preserve both readings, identify the missing evidence, and revise the rule before automation.

Three application checks