Common Google Analytics mistakes
Every part in this series flagged one mistake in its own Callout, setting up tracking for two different businesses. This part collects all 12 in one place, each with where to check it in the GA4 UI and what the fix actually looks like — a periodic audit checklist, not new material.
Mistake 1: partial e-commerce tracking (part 9)
Tracking only purchase, skipping view_item and add_to_cart.
Check: Reports → Monetization → Ecommerce purchases. If view_item and add_to_cart rows show zero or near-zero events while purchase shows real volume, only the final step is instrumented.
Cost: accurate total revenue, but zero diagnostic power over why a specific product underperforms — no way to see whether visitors are viewing it and not adding it to cart, or adding it and abandoning at checkout.
Mistake 2: lead-gen tracking that never closes the loop (part 10)
generate_lead fires with an estimated value, but no offline conversion import ever connects it to a real, eventual closed deal.
Check: Admin → Data display → Offline conversion imports. An empty or never-updated import here means every lead is valued the same regardless of what actually happened to it in the CRM.
Cost: optimizing toward cheap, plentiful leads rather than leads that actually become paying customers — the two are not the same thing without this connection in place.
Mistake 3: rebuilding the same audience three times (part 11)
"Cart abandoners" defined separately and slightly differently in GA4, Google Ads, and Meta Ads.
Check: Compare the exact condition logic in Admin → Audiences against the equivalent audience definition in Google Ads' and Meta Ads' own audience managers, not just the name.
Cost: three audiences that don't represent the same actual group, making any cross-platform comparison built on them unreliable from the start.
Mistake 4: never opening Explorations at all (part 12)
Relying only on GA4's standard, built-in reports for every question.
Check: Explore tab usage — if it's never been opened, every question has been forced through a report built for a general audience rather than a real, specific one.
Cost: a specific question — exactly where this checkout flow leaks, exactly what content precedes this business's leads — goes unanswered when a funnel exploration would have answered it directly.
Mistake 5: comparing platforms' own last-click numbers directly (part 13)
"Meta Ads costs more per purchase than Google Ads" — compared using each platform's own isolated, self-reported last-click data.
Check: Advertising → Attribution → Model comparison, looking at the same conversion under a data-driven or linear model instead of each platform's own last-click default.
Cost: defunding a channel that was quietly contributing real, uncredited assist value earlier in the customer journey — a mistake that only shows up once cross-channel attribution is actually compared.
Mistake 6: treating Meta traffic in GA4 as equally complete as Google Ads (part 14)
Google Ads ↔ GA4 is a native, first-party link with full campaign detail; Meta Ads ↔ GA4 is UTM-tagged sessions only, structurally less complete.
Check: Advertising → Advertising snapshot — Google Ads campaign names populate automatically; Meta campaign names only appear correctly if every ad's URL was UTM-tagged consistently.
Cost: weighing both rows in an Acquisition report as equally authoritative when one has a structural completeness advantage the other never had.
Mistake 7: never linking Search Console (part 15)
Check: Admin → Product links → Search Console links. If this shows no linked property, organic query data isn't available inside GA4 at all.
Cost: no way to distinguish a real ranking drop, a seasonal demand shift, or a SERP feature change behind a traffic move — GA4 alone shows the traffic changed, not why.
Mistake 8: sending parameters that are never registered (part 16)
company_size sent with every generate_lead event, but never registered as a custom dimension in GA4's admin settings.
Check: Admin → Custom definitions — if a parameter appears in DebugView's real-time event stream but has no matching row here, it's arriving correctly but isn't reportable anywhere.
Cost: weeks spent concluding "the tracking must be broken" over data that was arriving correctly the entire time, just invisible in every standard report.
Mistake 9: assuming Consent Mode means tracking continues quietly regardless (part 17)
Check: Admin → Data Settings → Data Collection, and a real test in an incognito window with consent declined — confirm event volume genuinely drops rather than continuing unchanged.
Cost: misreading every downstream report, since a real decline in consent genuinely reduces individual-level tracking — the aggregate is statistically modeled to compensate, not a hidden continuation of full tracking.
Mistake 10: rebuilding GA4's own standard reports in Looker Studio (part 18)
Check: if a Looker Studio dashboard's tiles map one-to-one onto GA4's own Acquisition or Engagement reports with no added dimension or calculation, it's duplicating work GA4 already does natively.
Cost: ongoing, unnecessary maintenance duplicating a report GA4 already performs, for no genuine analytical gain over just opening GA4 directly.
Mistake 11: deciding from a single headline metric (part 19)
Check: any decision citing one number — "sessions are down 12%" — without a second report (by channel, by device, by landing page) breaking down where that change actually concentrated.
Cost: reactive decisions made on attribution-biased, statistically noisy, single-platform numbers instead of the fuller cross-channel picture GA4 makes available one click deeper.
Mistake 12: confirming an event fires without checking its parameters (part 20)
Check: DebugView — click into the specific purchase event and expand its parameters, not just confirm the event name appears in the stream.
Cost: a purchase event with an empty items array looks identical to a correct one at a glance in the event stream — silently zeroing out every product-level report until someone actually inspects the parameter panel.
The thread connecting all twelve
Every mistake above shares one root cause: trusting a single, isolated number or event without checking the fuller context behind it — the attribution model, the registration status, the parameter values, the linked data source.
Reading a list like this once and never checking it against a real, live property again. The value here is as a periodic audit checklist — worth revisiting whenever a new event, dimension, or linked platform is added to a real GA4 setup, not just once at initial setup.
FAQ
Which of these mistakes is hardest to detect after the fact? Mistake 12 (unchecked parameters) and Mistake 8 (unregistered dimensions) are the two that look completely correct from a distance — the event fires, the count looks right, and the actual problem is only visible one level deeper than most people check.
How often should this checklist actually be revisited? Whenever a new event, custom dimension, or ad platform link is added — not on a fixed calendar. Each of these mistakes gets introduced by a specific change, not by time passing on its own.
Do these mistakes apply the same way to a small site as a large one? The mechanics are identical regardless of traffic volume — a small site just has less data to reveal the gap, meaning a broken parameter or missing link can go unnoticed for far longer before it's actually consequential enough to investigate.
Is there a faster way to check all of these at once? Not reliably — each one lives in a different part of the GA4 interface (Admin, Explore, DebugView, Advertising) by design, since they're different categories of problem. A recurring calendar reminder to walk this list is more realistic than a single dashboard catching all twelve.
Next, and last: building a complete measurement plan — the upfront document that prevents most of these twelve mistakes before any tracking code is even written.