Set one decision and a hard research boundary
Choose a single question such as whether an invoicing tool belongs in a guide for solo consultants who bill internationally. Define geography, device, budget range, frequency, required outcome, evidence cutoff, responsible editor, and the maximum time allowed for the review. Do not begin with a commission rate or a favored merchant. The stopping rule is equally important: publish only when essential requirements are supported, exclusions are explicit, and the best available alternative has been considered; otherwise record “not ready to recommend.”
Evidence: Government Digital Service
Describe the recent task before naming products
Collect evidence about what the intended reader is trying to complete, the trigger, current workaround, costly friction, and acceptable outcome. GOV.UK recommends researching actual or likely users and treating non-user opinions as assumptions. Write the need without embedding a feature or brand. “I need to collect payment from overseas clients without unpredictable conversion costs” is researchable; “I need Brand X’s global checkout” has already selected the answer. Include an edge case that challenges the apparent mainstream situation.
Evidence: Government Digital Service
Build a requirement hierarchy
Separate must-have outcomes from useful conveniences and preferences. For each requirement record why it matters, how failure harms the reader, and what evidence could verify it. A must-have for the invoicing example could be support for the reader’s business country; a preference could be customizable colors. Add prerequisites such as compatible accounts, identity checks, tax configuration, or minimum volume. This hierarchy prevents an impressive secondary feature from compensating for a missing capability that blocks the real job.
Evidence: Christensen Institute; Government Digital Service
Interrogate the offer through current primary material
Read the product page, help center, pricing schedule, customer agreement, privacy information, cancellation path, and support documentation. Save URLs, access date, quoted claim fragment only when necessary, and your paraphrased interpretation. Test whether each requirement is explicitly supported, merely implied, or absent. A screenshot or cached copy can preserve the reviewed state, but publication should link to the current public source. Escalate contradictions between marketing copy and governing terms rather than choosing the more favorable wording.
Evidence: Federal Trade Commission
Search for failure evidence and adequate substitutes
Seek independent reports about setup friction, support outcomes, limitations, and use in circumstances similar to the target row. Do not average star ratings or treat repeated wording as independent confirmation. Compare at least one non-commercial workaround, one competing offer where relevant, and doing nothing. The purpose is not to discredit the candidate; it is to identify conditions under which another path produces the same progress with less cost, commitment, or risk. Record the sampling limitations of every review source.
Evidence: Government Digital Service; Christensen Institute
Complete the decision table without a score shortcut
Create rows for the primary situation, a nearby variant, and at least two exclusion cases. Columns should include desired progress, existing method, mandatory requirement, source-backed capability, total commitment, support dependency, known failure, alternative, result, and revisit trigger. Use fit, conditional fit, exclude, or unknown rather than one blended percentage. A high numerical score can hide a fatal incompatibility. The decision must trace to the requirement that matters, not to the number of favorable boxes.
Evidence: Government Digital Service
Draft the recommendation from the table outward
Lead with who the offer may help and the condition that makes it relevant. Put major prerequisites and who should skip it before minor benefits. State what was documented, what independent evidence suggests, and what remains an inference. If an affiliate relationship is later introduced, disclose it near the recommendation in language ordinary readers understand. FTC guidance makes clear that an endorser cannot claim unperformed experience or use a testimonial as permission to repeat an unsupported product promise.
Evidence: Federal Trade Commission
Verify, date, and stop
A second editor should trace every material sentence to the table and confirm that the exclusion cases survived drafting. Recheck links, currencies, countries, version limits, refund conditions, and alternative availability. Stop if a required fact is unavailable, the intended reader cannot be distinguished from excluded readers, or a high-consequence claim needs specialist review. When the article passes, schedule a refresh around the most volatile dependency. The workflow produces a defensible snapshot, not a permanent certification of the product.
Evidence: Government Digital Service; Federal Trade Commission
Sources and further reading
These references informed this article. A source supports a claim; it does not imply endorsement of TenMultigure or any future product reference.
- Learning about users and their needsGovernment Digital Service · Accessed August 10, 2026
Supplies the recent-task research method and assumption discipline used before any candidate product enters the workflow.
- Understand users and their needsGovernment Digital Service · Accessed August 10, 2026
Grounds the bounded decision, early hypothesis testing, and use of multiple evidence streams for verifying the proposed match.
- FTC's Endorsement Guides: What People Are AskingFederal Trade Commission · Accessed August 10, 2026
Controls how experience, material connections, product claims, and atypical outcomes are represented in the drafted recommendation.
- Jobs to Be Done TheoryChristensen Institute · Accessed August 10, 2026
Informs the hierarchy from reader progress and circumstance to capabilities, rather than reversing the process from product features.
Reviewed by TenMultigure Editorial Review. See an error or a source that has changed? Tell the editorial team.
Review method: AI-assisted desk research with editorial checks. Reviewed ; next scheduled review . Turned fit research into an eight-step stop-go workflow, including a requirements hierarchy, adverse evidence search, invoicing scenario, unblended table outcomes, and second-editor traceability.