GA4 Mastery runs three diagnostics on your Google Analytics 4 data. Anomaly Engine finds the source, country or device behind a traffic drop. Revenue Leak finds the checkout step that loses buyers. Spam Purger finds bot traffic and writes the filter. Each one explains the result in plain language.
Read-only Google sign-in. Signup includes 100 free credits, no card.
fieldnote.app4,571 creditsFN
GA4 Mastery · Anomaly Engine
fieldnote.app · GA4 property · Tuesday
Illustrative data
Sessions today
171
−68%
Same-weekday baseline
534
Threshold
−15%
Cost
50 cr
Daily sessions15 days
Where the drop isshare of the drop
google / organic84%
United States79%
desktop61%
Explanation: almost all of the loss is Google organic traffic from the United States. Check Search Console for a ranking change before looking at your site.
Reads your data fromGoogle Analytics 4
What is GA4 anomaly detection?
GA4 anomaly detection flags a day when traffic falls well below normal and tells you where the loss came from. In AI Rank Lab, Anomaly Engine compares the most recent day of GA4 sessions with the same weekday earlier in a 15-day window. A drop of more than 15% counts as an anomaly, and the day is then split by source / medium, country and device to name the rows behind it.
GA4 Mastery adds two more checks on the same read-only connection: Revenue Leak for the checkout step that loses buyers, and Spam Purger for bot traffic. Each run costs 50 credits.
Three GA4 diagnostics, one connection
Each tool does the arithmetic on live GA4 data first, then asks Gemini to explain what it found and what to do.
Anomaly Engine: why did traffic drop?
It reads the last 15 days of sessions and compares the latest day with the same weekday earlier in that window, so a quiet Sunday is never compared with a busy Monday. A drop of more than 15% counts as an anomaly. It then slices that day by source / medium, country and device to find the rows that hold most of the loss.
Same-weekday baseline removes weekly seasonality
Source, country and device slices in parallel
A plain-language explanation you can send to a client
It counts view_item, add_to_cart, begin_checkout and purchase events for the last 7 days and for the 28 days before that, works out the abandonment rate at every step and flags the step that got worst. A device and browser breakdown shows whether the leak is, say, mobile Safari or desktop Chrome.
Worst step: begin_checkout to purchase. Abandonment is 18% higher than the baseline, mostly on mobile Safari.
Spam Purger: clean the numbers you report
It scans 30 days of traffic by source and hostname and flags sources with more than 50 sessions, an average session under 1 second and engagement below 5%. You get the spam share of your sessions, the clean session count and a regex filter for the flagged sources.
50 credits per analysis, charged when you run it. Signup includes 100 free credits.
Refresh
On demand. Nothing runs on a schedule; each run reads GA4 at that moment.
Limits
GA4 properties only (no Universal Analytics). The anomaly check reads the most recent day GA4 returns, which includes today's partial data, so a run early in the day can flag a day that has not finished yet. Revenue Leak needs ecommerce events.
How does GA4 anomaly detection work in AI Rank Lab?
Anomaly Engine reads 15 days of GA4 sessions and compares the latest day with the average of the same weekday in the previous weeks. If sessions are more than 15% lower, it slices that day by source / medium, country and device against the baseline days and names the rows that account for most of the drop. Gemini then writes a short, client-ready explanation.
What counts as an anomaly in the Anomaly Engine?
A day whose sessions are more than 15% below the same-weekday baseline. Comparing a Tuesday with earlier Tuesdays removes the normal weekly rise and fall, so a quiet weekend is not flagged as a problem.
How does the Revenue Leak Diagnostician work?
It pulls view_item, add_to_cart, begin_checkout and purchase event counts for the last 7 days and for the 28 days before, calculates the abandonment rate at each step and how much it changed, and flags the step that got worst. It adds a device and browser breakdown and asks Gemini for a specific diagnosis and fixes.
How does the Spam Purger identify bot traffic?
It flags traffic sources with more than 50 sessions, an average session duration under 1 second and an engagement rate below 5% over the last 30 days. Real visitors almost never fail all three at once. It then reports the spam share of your sessions and writes a regex for the flagged sources.
How do I use the spam filter regex in GA4?
GA4 has no admin filter that drops traffic by source, so apply the regex where you read the data: as an exclude condition on session source in a report filter, a comparison or an Exploration segment, or as a filter in Looker Studio. Check the excluded rows first so you do not remove real visitors. Unwanted referrals in GA4 Admin only re-attribute sessions; they do not remove spam.
What does each analysis cost?
Each of the three tools costs 50 credits per run, taken from your plan’s credit pool. Signup includes 100 free credits with no card, enough for two runs.
Is my GA4 data safe?
AI Rank Lab connects with Google OAuth and asks only for read-only analytics access, so it cannot change your property. Data is fetched when you run an analysis. You can revoke access at any time from your Google Account.
How is GA4 Mastery different from Analytics chat?
Analytics chat answers any question you type about your GA4 data. GA4 Mastery runs three fixed diagnostics with set windows and thresholds, so the result is the same kind of answer every time: the cause of a drop, the leaking checkout step, or the spam sources. Use chat to explore and GA4 Mastery to explain a change.
Can I use GA4 Mastery without the Search Console tools?
Yes. GA4 and Search Console are separate connections in the same Analytics workspace. Connect only GA4 and the three GA4 Mastery tools, the Overview and Chat work on their own.
What GA4 properties are supported?
Any Google Analytics 4 property your Google account can read. Universal Analytics is not supported. If you manage several properties, switch between them with the property selector.
Google Analytics 4