Required original evidence
- [ADD ORIGINAL DATA] Eligible businesses, recruitment method and consent or other appropriate collection basis.
- [ADD ORIGINAL DATA] Observation dates, eligible sessions and device or language cohorts.
- [ADD ORIGINAL DATA] Conversation starts, voluntary contact requests and de-duplicated useful enquiries.
- [ADD ORIGINAL DATA] Staff-confirmed outcomes and the time allowed for follow-up.
- [ADD ORIGINAL DATA] Missing records, exclusions, uncertainty and reviewer approval.
Define the funnel before collecting records
An eligible session, a chat opened, a contact request and a sale are different units. Record exactly how each is counted and whether one person can appear more than once. Choose an observation period long enough for staff to complete the planned next step.
Avoid combining traffic from businesses with incompatible sales cycles into an unexplained headline average. Separate exploratory observations from a design that could support a causal claim.
Publish a method readers can inspect
Explain sample selection, dates, incomplete follow-up and known sources of bias. Prefer aggregated reporting with proportionate access to underlying information. Do not expose conversation transcripts or contact details as evidence of a marketing result.
[ADD ORIGINAL DATA] A reviewed summary table and reproducible calculations should replace these placeholders before any results article is published. A study outline must not be promoted as a completed industry report.