Sessions
87from 109
Internal performance diagnostic
Find the signal behind the drop.
Start with what the analytics prove. Then test the explanations that remain.
Reported comparison
The decline that triggered the review
These are the figures supplied in the diagnostic request. Confirm the dashboard date range and timezone before using them as the permanent baseline.
20.2%
Pageviews
87from 152
Unique visitors
48from 94
Observed
A consistent window confirms the direction.
The latest seven completed days, September 9 through 15, are compared with September 2 through 8. This avoids the partial current day.
Sessions107 to 60 -43.9%
Pageviews149 to 90 -39.6%
Unique visitors89 to 46 -48.3%
Daily pageviews
September 2 through 15
23
232
24
49
6
618
37
813
24
1018
9
124
6
1416
Prior periodCurrent periodSep 8 peaked at 37 pageviews
Verified drivers
What the available analytics actually say
The homepage explains most of the session loss
Homepage landings fell from 71 to 37 sessions, a loss of 34 sessions.
Direct or unattributed traffic had the largest source decline
Direct or unattributed sessions fell from 74 to 40, also a loss of 34 sessions.
Google referrals also declined
Sessions referred by google.com fell from 15 to 7, a loss of 8 sessions.
The decline affected both major device groups
Desktop fell from 72 to 36 sessions. Mobile fell from 25 to 14.
Engagement depth improved among the traffic that remained
Desktop pages per session rose from 1.65 to 1.83. Mobile rose from 1.20 to 1.71.
Diagnostic assistant
Choose the question you need to answer.
ObservedCurrent answer
The strongest signal is lower acquisition volume, especially direct or unattributed traffic.
Evidence
- Direct or unattributed sessions fell by 34.
- Google referrals fell by 8.
- Pages per session improved, which argues against a broad engagement collapse.
Next checks
Hypotheses to test
Plausible explanations, not conclusions
Each explanation below includes the evidence that makes it plausible and the check that could confirm or reject it.
01
A direct-traffic burst ended
Why it is plausible. Direct or unattributed traffic accounts for 34 of the 47 lost sessions in the verified comparison.
How to test it. Compare campaign sends, social posts, offline promotions, and staff traffic around September 2 through 8.
02
Some campaign attribution was lost
Why it is plausible. A large direct or unattributed change can include links that arrived without durable UTM parameters.
How to test it. Audit recent campaign links and redirects. Confirm source, medium, and campaign parameters survive the full path.
03
Search demand or visibility softened
Why it is plausible. Google-referred sessions fell by 8, but this dataset alone cannot separate rankings, demand, and result click-through rate.
How to test it. Review Search Console clicks, impressions, average position, and queries for the same two completed weeks.
04
The prior week contained a short-lived spike
Why it is plausible. September 8 produced 37 pageviews, the highest day in the 14-day comparison.
How to test it. Identify what drove September 8 before treating the prior period as the normal baseline.