01 / START HERE
A useful way to approach it.
Match each source to a research question. Population data describes an area; venue terms explain room access; customer conversations can test a proposed workshop schedule. Those are different evidence jobs. A larger bibliography does not fix a mismatch between a source and the decision.
Use o1 to classify supplied material and identify gaps. Current discovery depends on enabled web search in signed-in chat; otherwise supply URLs and excerpts yourself. Mark suggested sources separately from material you reviewed, and retain access limitations.
02 / THE PROCESS
Go from question to next step.
- 01
Turn the decision into evidence questions
Write the decision first, then list what you need to know before choosing. Separate customer context, market counts, product behavior, terms, and reported outcomes. This prevents a broad search such as 'business research' from becoming your method. A good question names a population, period, or specific capability that evidence could address.
- 02
Choose appropriate source types
Look for official statistical portals for defined counts, original publications for research methods, company documentation for described product behavior, and published terms for offer conditions. Commentary can help discover those sources, but follow its references when the claim matters. A company page establishes what the company publishes; independent performance requires different evidence.
- 03
Find and inspect the original item
Record a stable URL, title, publisher, publication or update date, and the relevant section. Open the material rather than relying solely on a search snippet. Check access limitations and whether a summary points to a report you have not read. If only part of the document is available, note the boundary.
- 04
Evaluate fit and provenance
Ask who collected the information, how it was obtained, which population it covers, and what the author is trying to establish. Label older evidence and commercial claims. Do not reject a source simply because it has a perspective; explain which questions it can answer and which require corroboration or additional material.
- 05
Maintain a usable register
Group sources by question and mark each reviewed, partially reviewed, or still to inspect. Add a short note about the exact evidence you found and the remaining gap. Keep the register small enough to revisit. When your decision changes, reassess source relevance instead of carrying every old link into the next report.
03 / YOUR STARTING POINT
A prompt you can make your own.
Replace the bracketed fields. Add only information you have permission to share. This is a reusable text prompt; it does not run a task from this page.
Build a source register for this decision: {{decision}}. Research questions: {{questions}}. Geography and period: {{scope}}. Available URLs or excerpts: {{material}}. For each source list publisher, title, date, source type, relevant question, useful evidence, and limitations. If enabled web search finds more material, link original sources; otherwise suggest where I should look without claiming retrieval. Distinguish reviewed material from suggested sources. Identify the biggest evidence gap and an appropriate source type for resolving it.04 / SEE THE SHAPE OF IT
Illustrative source register for a workshop
Fictional research question and authored source descriptions; these rows are not citations to retrieved publications.
| Question | Source to inspect | Boundary |
|---|---|---|
| Who lives nearby? | Official demographic table for the area | Population does not prove interest |
| Can the room be used evenings? | Venue's current published booking terms | Availability still needs confirmation |
| What does a rival offer? | Rival's dated workshop page | A published offer is not sales volume |
| Would people attend? | Interviews you arrange with prospective attendees | Record who was asked and how |
A register makes different evidence jobs visible. It also prevents a source about one question from being stretched into proof of another.
05 / KEEP YOUR JUDGMENT
What to watch before you use it.
Treating summaries as original evidence
A summary may omit qualifications. Find the original item when your conclusion depends on a number, condition, or exact statement.
Ignoring publication context
Record the population, date, and collection method. A source can be credible yet poorly matched to the question you need answered.
Keeping an unreviewed bibliography
Mark suggested sources clearly. A list produced by an assistant is not evidence that its entries were opened or found to support your claim.
06 / THE FINAL PASS
A little review goes a long way.
Your checks stay on this page. They do not approve or run a task.
07 / GOOD QUESTIONS
Before you begin.
Are official sources always the best choice?
Choose the source suited to the question. Official material can establish defined statistics or published rules, while product documentation may describe a feature more directly. You still need to inspect scope, date, and any limitation before using the evidence.
Can I use customer reviews as a source?
Yes, for examining what those reviewers report. Record platform, selection method, dates, and sample limitations. Reviews should not be presented as a representative measurement of all customers or a complete account of a competitor's performance.
What should I do with an old source?
Keep it if historical context answers your question, and label its period. For current claims, find an appropriate newer source or state that current evidence is missing. Do not silently change an old figure's date to make it appear fresh.