Broken tracking, PII leaks, duplicate events, (not set) values. Ask the questions that keep technical teams up at night and get straight answers.
Tom Nakamura is responsible for Bloom & Measure's analytics implementation. Before anyone can trust a revenue number or a campaign metric, the underlying tracking has to be right, and so does any tool reporting on it. Tom doesn't need marketing insights. He needs to know whether the data is clean, whether the answers checking it can be trusted, and he needs to know now.
Every answer below came from Askalytics using verified Q1 2026 data from the Bloom & Measure demo property.
"Which pages have the highest percentage of (not set) dimension values?"
None. No (not set) dimension values detected across any of the dimensions checked.
pagePath: Clean across all 13 pages.
pageTitle: Fully populated.
sessionDefaultChannelGroup: All 7 channels cleanly attributed.
deviceCategory: All sessions attributed to a device type.
Traffic source attribution is clean, page tracking is complete, and device data is fully populated. No gaps.
Confidence: High | Data: Google Analytics 4
The Confidence badge is the Confidence Check, which adds a second AI from a different provider on AI Insights. The counts themselves come from queries run against your own data and trace back to those queries, not the AI's estimate.
"Are there any events firing with missing or undefined parameters?"
No missing or undefined parameters detected across the key dimensions checked.
All 26 distinct events fired cleanly. The e-commerce funnel events (view_item, add_to_cart, begin_checkout, purchase) are all recording with valid parameters. No events fired with an unresolved page path, missing item name, or unattributed channel.
257,212 total events across the period, all cleanly attributed to known channels.
Confidence: High | Data: Google Analytics 4
"Are there any duplicate events being recorded?"
No true duplicates. All 26 event types are distinct. Some names are structurally similar but serve different purposes:
upsell_card_offered (843 events) vs upsell_card_accepted (352)
upsell_vase_offered (672) vs upsell_vase_accepted (192)
video_start (422) vs video_complete (195)
These are offer-acceptance and start-complete pairs, not duplicates. Standard lifecycle events (page_view, session_start, scroll, click) all appear at expected relative volumes.
Confidence: Medium | Data: Google Analytics 4
"Is there any PII being captured in event parameters?"
No obvious PII signals detected.
All 13 page paths are clean, structured URL slugs (e.g. /checkout, /account/orders, /product/classic-red-rose-bouquet). No email patterns, no query parameters like ?email= or ?name=, no PII-style strings in URLs.
Pages to monitor: /checkout, /account, and /account/orders are present in the data. These are the highest-risk pages for PII leakage if form field values are inadvertently passed as URL or event parameters.
For a complete audit, review your GA4 Data Deletion settings and check the BigQuery raw event export for individual parameter values that the standard API cannot expose.
Confidence: Medium | Data: Google Analytics 4
All 13 pages, 26 event types, and 7 channels passed the integrity checks. No (not set) values, no broken parameters, no missing dimensions.
28,707 sessions attributed to campaign name "(all)" rather than a named campaign. This means UTM campaign parameters are missing from some traffic sources. Fixable with a quick link audit.
The full purchase flow (view_item through to purchase) fires correctly with valid product data at every step. No leaky events or broken handoffs.
Custom events like upsell tracking follow a clear naming convention (offered vs accepted). No redundant or overlapping event names detected.
The underlying counts come from queries run against your own data and trace back to those queries, so you can see where each figure came from. How the Confidence Check works.
Every new property gets a 2-stage analytics health check within two minutes of connecting.
Immediate issues and easy fixes. Broken tracking, missing tags, configuration gaps that can be resolved in a single session.
Medium-term improvements. UTM tagging gaps, event taxonomy recommendations, and data hygiene issues that affect reporting accuracy.
Structural recommendations. Cross-domain tracking, enhanced e-commerce setup, and measurement framework improvements for the road ahead.
This is how Askalytics works for operations and technical teams. Every count traces back to the query behind it. See how the Confidence Check works on AI Insights.
(not set) values, missing or undefined event parameters, duplicate events, broken attribution, and PII signals in page paths and event parameters. Ask one question per check and get an answer in under a minute.
It scans for email patterns, query strings like ?email= or ?name=, and PII-style strings in URLs, and flags the highest-risk pages to review. For a complete audit it points you to your GA4 Data Deletion settings and the BigQuery raw export.
Yes. The counts come from queries run against your own data and trace back to those queries, not the AI's estimate. You can see where each one came from.
Yes. A 2-stage analytics health check runs within about two minutes of connecting: quick wins, strategic priorities, and long-term goals. You can keep asking diagnostic questions any time after that.
Read-only access only. Plans run from Scout $19 to Scale $149 per month. Start with a 14-day free trial, no card needed. See full pricing.
Connect a property and get a full analytics health check automatically, with every count traced back to your own data. Then keep asking diagnostic questions whenever something looks off.
Get Started