DARK SOCIAL REPORTING CHECKLIST
Worked case: a fictional report for a travel-lens guide. These are not real data.

COMPLETED REPORT · APRIL
Page: guide to choosing a travel lens.
Analytics unit: sessions on that page.
Total: 620 sessions.
Known campaign tag: 210 sessions with utm_source=instagram and utm_campaign=lente_abril.
No clear source: 95 sessions in (direct)/(none).
Other recorded sources: 315 sessions.
Optional question: “How did you find this guide?”
Answers: 19 people. Eight said someone sent the link privately; five named WhatsApp, two email and one another app.

HOW TO READ THIS EXAMPLE
The 210 sessions identify the tagged link, not its last sender. If someone forwarded the URL, its original campaign label may remain.
The 95 sessions belong to a no-clear-source category. They may include directly typed visits, untagged links, private messages, redirects, offline documents or other causes the system did not record.
The 19 answers are voluntary responses, not sessions. The eight private-share mentions cannot be assigned to the 95 sessions or generalised to all visitors. The five WhatsApp mentions do not estimate how much traffic came from WhatsApp.
The three session groups are disjoint, use the same definition and reporting period, and make up the total: 210 + 95 + 315 = 620. The 19 answers are people and stay separate; they are not added to sessions. The evidence supports partial signals while the rest remains unknown.

CHECK BEFORE THE NEXT REPORT
[ ] I set the page, dates, time zone and analytics tool.
[ ] I checked that totals and source rows use the same period and session definition.
[ ] I use UTMs only on campaign links my team controls.
[ ] I opened a test link after redirects and confirmed the tags reached the destination.
[ ] I kept (direct)/(none) labelled “no clear source”; I did not rename it WhatsApp or dark social.
[ ] If I ask an optional question, I record how many people saw it and how many answered.
[ ] I show answers, sessions and clicks separately; I do not add an answer to a session count.
[ ] I describe what the system recorded as an observation and a possible explanation as a hypothesis.
[ ] I leave any source I cannot verify unknown.
[ ] I do not request access to private messages or put personal data in UTM labels.

REPORT SENTENCE TO ADAPT
“In April, the guide recorded 620 sessions: 210 with the Instagram UTM campaign, 95 classified as (direct)/(none) and 315 with other recorded sources. Of 19 voluntary answers, eight people remembered receiving the link privately. These answers add context, but they do not identify the 95 sessions without a clear source or estimate where all visits came from.”
Replace the date, page, counts, tag names and tool before sharing. Do not present this sentence as a real result.

SOURCES AND LIMITS
Google Analytics, understanding (direct)/(none): https://support.google.com/analytics/answer/15258820?hl=en
It explains that direct/(none) indicates no clear source and lists several possible causes; it does not identify a private channel.
Google Analytics, campaigns and traffic sources: https://support.google.com/analytics/answer/11242841?hl=en
It documents campaign parameters. A UTM labels a controlled link, not necessarily the person who forwarded it or the reason for a visit.
Alexis C. Madrigal, The Atlantic, October 12, 2012: https://www.theatlantic.com/technology/archive/2012/10/dark-social-we-have-the-whole-history-of-the-web-wrong/263523/
Historical source for the term's use with email, messaging and app links without referrers. Its 2012 data are not a current benchmark.

The case, counts and report sentence are original editorial work and hypothetical. This checklist helps report uncertainty; it cannot reconstruct all private sharing.