Research report · 2 November 2026
Customer-language evidence: keep the quote attached to its source
A bounded review of three portable skill sources found no reason to replace a maintained evidence table and named marketer review for a few supplied statements. No fifth skill card follows.
The bounded job
If a small set of public or consent-cleared, dated and attributable non-sensitive statements is supplied, a named marketer can answer one narrow brief question: which observed objection might this brief address? The output is a local record, not publishable customer copy. A reviewer checks quotation fidelity, rights and privacy, sample boundaries, disagreements, omissions and any proposed implication before use. No customer statements were available for this review; it produced no quotes, theme counts or customer findings.
| Keep with each row | Why it matters |
|---|---|
| Exact quote or marked extract; speaker and source locator; date and immediate context | Lets a reviewer verify wording and avoid turning a partial sentence into a broader claim. |
| Sample or cohort; permission and use restrictions; missingness and counterexamples | Shows whose views were available and where the record is silent. |
| Observed point, inferred interpretation and reviewer-owned brief implication in separate fields | Prevents a proposed message from being presented as a customer statement. |
What the three sources add—and what they do not
| Original selected source | Useful source guidance | Reason not to add a card |
|---|---|---|
| Anthropic customer-voice | Requires verbatim quotes with speaker, date, link and context; reports coverage gaps and distinguishes no searched calls from no mention. | Its default CRM, call, document and email search is wider than a local supplied-statement job. A files-only route is described, but no single-skill payload, host route or quote-use gate was established. |
| LeadMagic customer-research | Names consent, quote source and cohort, evidence versus interpretation, counterexamples and sample limits. | It covers recruiting and interviews as a wider research process. Five directly named supporting resources could not be opened in this review, so their actual template and checker remain unknown. The local table can carry the visible evidence fields. |
| Zorai user-research-synthesis | Visible text advises full-transcript reading, direct quotes, recruitment-bias notes and care with frequency versus importance. | The browser rendering was incomplete. It did not establish a per-quote locator or permission rule in the visible portions; this is an inspection limit, not proof the original file lacks one. Its sample-size assertion is not validation of representativeness. |
Limits and next comparison
This was an original-source review, not execution or a customer study. The selected files were mutable main renderings observed on 2 November; no selected-file commit pin, complete payload, installed bytes, host permissions, safety result or measured advantage was established. LeadMagic's frontmatter displayed version 1.0.0; no reliable selected-file version was established for the other two. Repository licenses and installation instructions are not a package audit.
Only compare a skill with the table on a later, authorized same-case task with real non-sensitive material, a named reviewer, an exact payload and host route, and a fixed rubric for traceability, omissions and review burden. If the table preserves the evidence with equal or less burden, the proposed skill adds no demonstrated value for this job. Connected sales-book searches, recruitment, scraping and downstream publishing need separate decisions. The four conditional cards and two historical nonrecommendations are unchanged.