What do you want to know?
Ask about a company, a sector or a trend. Signals turns it into measurable evidence: what’s happening, what it suggests, and what to watch.
Questions Redflag data answers
Worked examples, one per research question, showing the kind of answer each dataset gives.
Live sample: US travel panel
This is what the travel question above looks like as delivered data: one collection run on 2026-10-01, 72 questions put to OpenAI. It is a single cross-section, so it shows levels rather than a trend.
Where AI answers send the customer
Share of every valid answer, by the destination it leads with. The four buckets sum to one.
Issuer reach: share of answers mentioning any brand the issuer owns
| issuer | reach | rate | n / d | 95% CI | destination share |
|---|
Deduplicated per answer: an answer naming Expedia, Hotels.com and Travelocity counts once for Expedia Group, not three times.
Substitution: who gains when a company is absent
A peer's recommendation rate when the subject is absent, minus its rate when present. Pairs with fewer than 10 observations on either side are not published.
What this is and is not
Start a collection
Runs execute in the background. This returns a run id immediately; the table below polls for progress.
Runs
| run | as-of | status | phases owed | done | planned | ok | failed | coverage | published |
|---|
Query
Known-at is the knowledge cutoff. Leave blank for now. Changing it shows the feed as it stood then — corrections published later are excluded.
Results
| entity | name | tickers | value | n / d | 95% CI | rev | known at |
|---|
Observations
| cell | question | engine | outcome | recs | srcs | first observed |
|---|
Entity registry
| entity | name | kind | role | tickers | domains | parent links |
|---|