Run a conversion tracking audit with Claude or ChatGPT

What you are actually auditing

A conversion tracking audit covers three things: what is configured (actions defined in the platform), what is firing (recent conversion counts per action), and what is missing (actions with zero recorded conversions despite active spend in their associated campaigns).

The third category is where money leaks. If a campaign targets a "Purchase" conversion that has not fired in 30 days, the smart bidding algorithm has been flying blind for 30 days.

Connect SpendSignoff and pull conversion data

With SpendSignoff connected to your AI client, start with a broad read: "List all conversion actions in my Google Ads account with their status and 30-day conversion count." The query_entities tool returns each defined action, its category, its tag-based or import source, and recent volume.

Do the same for Meta: "Show me all Meta pixel events with conversion counts for the last 30 days." SpendSignoff normalizes the response format so you can compare across platforms in the same conversation.

Spotting the common failure modes

• Zero-fire actions with active campaigns — An action defined but not receiving traffic is either mis-tagged or targeting the wrong event name. Ask: "Which conversion actions have active campaigns but fewer than 5 conversions in the last 30 days?"
• Duplicate purchase events — Two actions recording the same transaction inflate reported conversions. Ask: "Do I have multiple purchase-category actions with overlapping counts?"
• Paused primary actions — The "Primary" action tag is sometimes accidentally paused during tag manager cleanup. Read its status explicitly.
• Cross-account mismatches — If you use Google Ads manager account imports, check that child accounts are not defining local conversion actions that shadow the manager-level ones.

Staging fixes as drafts

Once the audit surfaces actionable gaps — say, a primary conversion action that has been paused — you can ask the AI to stage a fix: "Propose a change to set 'Purchase (primary)' back to primary status." SpendSignoff calls propose_change and the fix lands in your approval queue as a before→after diff. You review and approve; no change goes live unreviewed.

For issues that require tag changes (the conversion tag itself is broken), SpendSignoff cannot fix the tag — that lives in Tag Manager or your site code. The AI can document the finding clearly: which campaigns are at risk, what the conversion action ID is, and what the tag container should reference.

Meta-specific checks

Meta's conversion tracking has a second layer: Conversions API (CAPI) vs pixel-only. Ask: "Which of my Meta conversion events have CAPI deduplication enabled?" Events that fire only through the browser pixel are vulnerable to iOS 14 signal loss. Events with both pixel and CAPI firing get better match rates.

SpendSignoff reads the event match quality scores Meta surfaces per pixel event. A score below 6 typically means missing customer information parameters — email, phone, or click ID — that improve attribution match.