Research and evidence standards

Define the question and scope

Record the audience, market, time period and decision the research supports. Separate business capability claims from market-demand research. A country in a keyword export does not establish that a team has delivered work there.

Use a source hierarchy

Prefer product documentation for product behavior, original research for research findings, and current authorities for legal or regulatory matters. Keep retrieval dates where change matters. Competitor pages can show positioning and structure, but their claims should not be adopted as independently proven results.

Retain evidence and uncertainty

Keep raw exports and the settings that produced them. Preserve displayed ranges rather than converting them into precise demand claims. Record missing values, sampling limitations and inaccessible sources. If a source cannot be inspected, mark the gap instead of presenting it as verified.

Create one page per useful task

Compare query intent, existing coverage and the distinct value a proposed page would add. Do not create separate pages simply because phrases vary. A definition, an implementation guide and a service offer may have distinct reader tasks, but their relationships should be explicit.

Review claims and examples

• Separate client results from invented illustrative calculations.
• Attribute employment experience to the person who did the work.
• Confirm permissions and denominators before publishing results.
• Use real authors and reviewers, with accurately described roles.
• Review local-language copy with suitable language expertise.
• Record unresolved questions and a responsible owner.

Maintain the published record

Revisit pages after material product or guidance changes. Correct substantive errors and retain an editorial history. Sources explain the foundation of a claim; they do not transfer responsibility for every recommendation to the source. This page describes the standard we intend to apply, not a claim that every legacy article has already passed it.

Make the conclusion traceable to its evidence

A research note should distinguish what a source states, what the team infers and what remains uncertain. For example, a published account of revenue growth does not establish that one channel caused all of it. Record the period, business scope and measurement basis so the reader can assess the claim. Where those details are unavailable, narrow the conclusion instead of filling the gap with confident wording.

For practical recommendations, explain the reasoning and the conditions under which the advice may change. Prefer primary documentation for product behaviour and original accounts for reported experience. A credible source can still be outdated or describe a different setting. Keep a review trigger for material changes and a route for correcting the published explanation. Confidential supporting records can remain private; public evidence should be limited to material the business is authorised to share.

• Attach sources to the claims they support rather than to a page indiscriminately.
• Label illustrative figures and scenarios explicitly.
• Record missing definitions and avoid presenting assumptions as verified results.

Review team delivery experience →