Google Ads Conversion Tracking: Audit Actions, Goals and Results With AI

Inventory the actions before judging performance

List conversion actions with their IDs, names, categories, types, statuses, counting methods, lookback windows and value settings. Names alone are unreliable. Two actions called “Lead” can use different sources or count different steps in a form journey.

For each action, write a business definition and identify its collection route. Examples include a website tag, an Analytics-based conversion or an imported offline result. Ask who owns its implementation and which business record can validate it. A form submission and a qualified lead represent different stages.

SourcetoMCP’s conversion-actions read returns configuration, so treat it as an inventory. Campaign performance is a separate report. If you need results by action, have the assistant discover compatible reporting fields before querying them, or use a Google Ads export. Keep unsupported breakdowns explicitly unchecked.

Read the goals each campaign actually uses

Inspect account-level goals, campaign-level settings and custom goals. A primary action affects ordinary goal-based bidding when the campaign uses its associated goal. A secondary action generally appears in All conversions. Google documents an exception for custom goals: an included secondary action can still be used for bidding.

SourcetoMCP reads these three goal scopes separately. A single account-level response cannot prove the settings for every campaign. Record the campaign IDs and follow up on any response-limit warnings before declaring the audit complete.

Use Google’s conversion-goal explanation when resolving the relationship between an action and a campaign. Avoid changing a primary flag simply because a conversion count looks high. Establish which outcome the campaign is supposed to pursue.

Find duplicate outcomes without guessing from totals

Consider a hypothetical month with AUD 6,000 spend and 100 completed lead forms. A website-tag action records 100 leads and an Analytics-based action records the same 100 submissions. If both are used in the campaign’s conversion total, the report can show 200 conversions and an apparent AUD 30 CPA.

Relative to 100 actual submissions, spend per lead is AUD 60. The two ratios answer different questions because the report counts two actions for each business outcome. This example illustrates duplicate measurement. It is not evidence that any pair of similarly sized actions in your account is duplicated.

Compare action definitions, trigger evidence and business records before proposing which action should guide bidding. Secondary measurement can help compare implementations, but a custom goal can still use it. Preserve the independent measurement you need and check the complete goal configuration.

Do not fix duplicate firing by deleting historical context. Save the baseline, the action IDs and the correction date. If the website sends one action twice, repair the trigger as well as reviewing the bidding goal.

Check values, timing and collection separately

A default value is an assigned fallback. It does not establish transaction revenue, profit or lead quality. Check whether the action always uses that default, which currency applies and whether a purchase sends the actual order value.

Compare equivalent date definitions. Google Ads reports can attribute a conversion to the advertising interaction, while another report uses the conversion date. Lookback windows, attribution and processing delays also affect comparisons. Use complete periods and allow recent conversions to mature before deciding tracking has failed.

Use Tag Assistant to check the website’s tag behaviour. An account audit cannot prove a successful payment fires once, the correct destination receives it or identifiers survive redirects. Browser evidence and business records complete those checks.

If Analytics supplies the conversion, use the GA4 conversion-tracking guide to validate the event and key-event settings. For URL parameters and auto-tagging, use the tracking-template audit.

Ask AI to produce an evidence-led conversion audit

Audit conversion tracking for Google Ads account [ID] and campaigns [IDs]. Make no changes. List conversion actions with IDs, source/type, category, status, primary setting, counting method, windows and default value rules. Read account, campaign and custom goals. Compare complete-period campaign cost, conversions, all conversions and conversion value using supported metrics. Identify possible duplicate business outcomes as hypotheses. State missing action-level reporting or browser evidence. Propose the smallest exact correction for review.

Require separate columns for observed settings, observed results and unresolved questions. The assistant should explain the definition behind CPA and ROAS before calculating them. A conversion total without its action and goal definitions is insufficient evidence for a bidding change.

Attach the findings to your Google Ads audit checklist. Keep this focused investigation separate from creative, targeting and landing-page changes so the tracking issue remains traceable.

Verify the correction before changing budgets

SourcetoMCP supports selected conversion-action writes, including supported counting and value settings. Check the discovered inputs for the exact action type and field. The connector’s goal operations described here are reads, so do not assume the assistant can rewrite campaign goal selection.

Approve the precise account, action ID and before-and-after settings. Use the Google Ads interface for unsupported goal or implementation changes. Read the saved configuration again, run the relevant collection test and log the change date.

Review the next complete period using the corrected definition. If CPA rises because duplicated conversions were removed, explain the denominator change before calling it a performance decline. Use the bidding-strategy guide only after you trust the outcomes the strategy receives.