Where Sessions Started
Where readers came from when they opened a Signal Check or a dashboard: Slack, Teams, email, one of our own links, no link recorded, or another site.
The source belongs to the session, not the page. Google Analytics records where a session started, then credits every page in it. So “Slack, 20 sessions” on a page means 20 sessions that started from a Slack link and opened this page at some point. It does not mean 20 people clicked a Slack link to this page: a reader who came from Slack to a Signal Check and then followed a link to a dashboard counts as a Slack session on the dashboard too.
Calculation
From meta_page_traffic, sum sessions by page and source group over the reporting window. The groups map sessionsource:
| Group | sessionsource |
|---|---|
| Slack | slack |
| Teams | teams, or anything starting teams. |
email |
|
| Our own links | latest-link, report_link, dashboard_link |
| No link recorded | (direct), (not set) |
| Other sites | anything else |
The Readership dashboard’s source_group() holds the same map; change both together.
Our own links are the link to the latest Signal Check and the links inside a Signal Check or a dashboard. No link recorded is a bookmark, a typed address, or an app that drops the link’s source.
SELECT
pagepath_redacted,
CASE
WHEN sessionsource = 'slack' THEN 'Slack'
WHEN sessionsource = 'teams' OR starts_with(sessionsource, 'teams.') THEN 'Teams'
WHEN sessionsource = 'email' THEN 'Email'
WHEN sessionsource IN ('latest-link', 'report_link', 'dashboard_link') THEN 'Our own links'
WHEN sessionsource IN ('(direct)', '(not set)') THEN 'No link recorded'
ELSE 'Other sites'
END AS source_group,
SUM(sessions) AS sessions
FROM google_analytics.meta_page_traffic
WHERE page_kind IN ('signal_check', 'dashboard')
-- set the reporting window; see the timespans note in cookbook/notes.qmd
AND CAST(date AS DATE) BETWEEN DATE '2026-09-01' AND DATE '2026-09-30'
GROUP BY 1, 2
ORDER BY pagepath_redacted, sessions DESC, source_group;tbl(con, in_schema("google_analytics", "meta_page_traffic")) |>
filter(
page_kind %in% c("signal_check", "dashboard"),
# set the reporting window; see the timespans note in cookbook/notes.qmd
as.Date(date) >= as.Date("2026-09-01"),
as.Date(date) <= as.Date("2026-09-30")
) |>
mutate(
source_group = case_when(
sessionsource == "slack" ~ "Slack",
sessionsource == "teams" | substr(sessionsource, 1, 6) == "teams." ~ "Teams",
sessionsource == "email" ~ "Email",
sessionsource %in% c("latest-link", "report_link", "dashboard_link") ~ "Our own links",
sessionsource %in% c("(direct)", "(not set)") ~ "No link recorded",
TRUE ~ "Other sites"
)
) |>
group_by(pagepath_redacted, source_group) |>
summarise(sessions = sum(sessions, na.rm = TRUE), .groups = "drop") |>
arrange(pagepath_redacted, desc(sessions), source_group)Interpretation notes
Sessions add up within a page, not across pages. One session that opened two pages counts once on each, so the sessions on two pages together can exceed the sessions that reached either of them. For a whole product, sum per page and say “sessions that included a page”, or count each session once in the GA4 interface.
For a Signal Check, use its first 14 days. That is when the announcement posts send their readers. See Signal Check Views for the window.
Alternative names
- traffic sources
- referral sources
- where readers came from
- session source