Peak monthly active users
5
−50% vs peak 10
Zero-Click Authority · Product analytics
The analysis loop is active, but volume is well below the previous three-month period.
A read-only PostHog snapshot of the last three months, compared with the preceding three months.
Executive summary
Current window compared with 8 Mar – 7 Jun 2026. The sample is directional; the funnel includes five unique users.
Pageviews
65
−39% vs 106
Analyses started
85
−91% vs 999
Analyses completed
84
−88% vs 714
Results copied
200
+223% vs 62
AI generations
156
−81% vs 815
Analysis activity
Weekly event totals across the reporting window. The final bucket covers 7–8 September.
Weekly analyses
85 starts · 84 completionsStartedCompleted
Starts peaked at 12 in the weeks of 15–21 June and 6–12 July; completions peaked at 14 in the week of 15–21 June.
Analysis funnel
event totals · 90 daysAnalysis started85
Analysis completed84
100% unique-user conversion. Five users started an analysis and all five reached completion within 30 days; median time between stages was approximately 28.9 seconds.
Use with care: the conversion rate is based on five users, while the bars show event totals. It is a directional signal, not a stable benchmark.
Workflow engagement
Downstream value signals are present, but should be interpreted at event level.
Workflow milestones
current windowResults viewed
66
event total
Results copied
200
event total
Credits confirmed
105
event total
Previews requested
92
event total
Analysis failures
8
event total
Credit cancellations
10
event total
Value signals
interpretation200 result copies and 105 confirmed credit reservations show downstream activity beyond pageviews. Copying and credit events are repeatable events, not user-based conversion rates.
Credit path: the observed reservation → confirmation funnel converted 5 of 5 unique users (100%), with a median conversion time of approximately 27.0 seconds.
Recommendation: agree whether completed analyses, copied results, or confirmed credits is the primary activation measure, then report its user-based rate.
AI usage and analysis quality
AI activity is lower than the comparison period; performance values are monthly bucket measures.
AI usage
current vs previousGenerations
156
−81% vs 815
AI cost
$3.89
−79% vs $18.28
Total tokens
631k
−83% vs 3.76M
The lower AI volume broadly follows the lower analysis volume; cost and token totals are aggregate estimates from captured AI-generation properties.
Completed-analysis quality
84 completionsAverage response
20.7s
Weighted current average
P90 response
50.2s
Sep bucket · 2 completions
Average SEO score
75.5
Weighted current average
The September bucket reached a 50.2s P90 response time across only two completions; investigate before treating it as a regression.
Reach and acquisition context
65 regular pageviews across the three-month window; the peak monthly active-user value was 5.
Top paths by pageviews
pageviewsThe top three paths contributed 60 of 65 pageviews; six pageviews were spread across four other paths.
Traffic context
current pageviewsDirect traffic
50
76.9% of pageviews
Desktop
63
96.9% of pageviews
United Kingdom
65
100% of pageviews
Regular traffic
65
No bot traffic observed
Experience health
Performance and friction signals are available; no captured exceptions were observed.
Web vitals
168 observationsP90 LCP
4.98s
August bucket
P90 INP
144ms
August bucket
Average CLS
0.049
August bucket
August is the clearest full-month reference; the September bucket has only two observations and no LCP or CLS sample.
Friction signals
current vs previousExceptions observed
0
No events
Rageclick events
32
+167% vs 12
Dead clicks
0
No events
Zero captured exceptions does not prove zero application errors; it means no PostHog exception events were observed.
Key findings and recommended actions
Evidence-backed, with no causal claims inferred from this snapshot.
1. Workflow volume fell sharply
There were 85 analysis starts and 84 completions, versus 999 and 714 in the preceding three-month period.
Action: confirm the traffic mix, deployment history, and event delivery before using the comparison as a product trend.
2. Downstream value events held up
Result copies rose to 200 from 62 even as starts declined; that pattern may reflect a change in user behavior or instrumentation.
Action: define one primary activation event and validate its identity and deduplication logic.
3. Performance needs follow-up
The September P90 response time reached 50.2s on two completions, and 32 rageclicks were captured across the window.
Action: review the affected URL-analysis and onboarding sessions, then segment latency by route, provider, and model.