A lead campaign reports a growing number of conversions, but the sales team sees fewer useful opportunities. The problem may be traffic quality. It may also be that the campaign is optimizing toward a different action from the one the business thinks it selected.
A conversion audit follows the signal from the real customer event to the campaign's bidding configuration. The output should explain what the campaign is rewarded for, how reliably that event is measured, and whether the reward matches the commercial objective.
Follow the campaign's actual goal path
Google's primary and secondary conversion documentation explains that primary actions are used for bidding when their standard goal is selected. Secondary actions are normally observational, but actions included in a custom goal are used for bidding regardless of the primary or secondary label.
That exception is why a spreadsheet containing only action names and labels is insufficient. Inspect each campaign's selected goals, account-default use, and custom-goal membership. A label can be accurate while the team's interpretation of its effect is wrong.
Save the campaign and conversion-action IDs. Names can change, and similar names can conceal different sources.
Define the business outcome before judging the setup
Write the campaign's intended result in business language. “Qualified consultation request within our service area” is more specific than “lead.” “Completed purchase with an order value” is more specific than “engagement.”
For lead generation, use the lead-stage map to distinguish submitted, contacted, qualified, booked, attended, and sold outcomes. These stages answer different questions and arrive at different times.
A higher-funnel action is not automatically useless. It may be a deliberate signal when deeper outcomes are sparse or delayed. The audit should reveal that tradeoff and test whether it is working, rather than treating every configuration as either universally correct or universally wrong.
Inventory the conversion actions
| Field | Audit question |
|---|---|
| Event definition | What must happen in the business for this event to exist? |
| Source | Website tag, analytics import, CRM import, call system, or another path? |
| Identity | Which stable action ID does the campaign use? |
| Goal membership | Which standard or custom goals contain it? |
| Value | Actual revenue, modeled value, fixed proxy, or no value? |
| Counting | Can one customer journey create repeated recorded outcomes? |
| Reliability | What sample proves the event and value are correct? |
Include inactive or legacy actions if they remain part of a live goal. An old action that still influences a campaign matters more than its last edit date suggests.
Look for duplicate business outcomes
Two conversion actions can represent the same purchase through different collection systems. That does not automatically mean the system has deduplicated them into one business outcome for every reporting or bidding purpose.
Trace representative orders or leads across the actions. Document which are intended for observation and which are intended to drive optimization. If both are deliberately used, explain the business rationale and how values are interpreted.
Avoid fixing suspected duplication by deleting historical actions without understanding dependencies. First establish the duplication, identify affected campaigns, and propose the smallest configuration change that produces the intended signal.
Review values separately from counts
An action can fire correctly but carry an inappropriate value. A form submission assigned the same value as a completed sale can distort a value-based objective if the business treats those events as economically equivalent without evidence.
Use a documented value policy. If values are proxies, say so. Record how they were estimated and when they should be recalibrated. Do not label a modeled lead value as collected revenue in the client report.
The metric dictionary should define these distinctions so the buyer, analyst, and business owner use the same terms.
Validate the deeper signal before switching
A proposal to optimize toward qualified leads depends on a working qualification feed. Check whether sales staff apply the stage consistently, whether the import arrives on time, and whether the platform accepts the records.
Use the offline lead import QA guide before relying on that signal. A semantically better event that rarely arrives may create a different operational problem from a frequent but weak event.
Document both the quality advantage and the reporting delay. The decision should acknowledge the available evidence, rather than assuming that moving down the funnel automatically improves every account.
Plan the configuration change as a migration
Save the existing goal path and recent mature performance. List the campaigns affected, the exact proposed settings, and the expected change in reported conversion counts or values. Obtain the normal business approval for the resulting objective.
Where practical, isolate the goal change from unrelated creative, destination, and budget changes. This makes subsequent interpretation easier. If several changes are necessary, record them as a combined operating transition rather than pretending the outcome isolates one setting.
After execution, read back the campaign's selected goals and action membership. Then verify that new valid events continue to arrive through the intended path.
Report the audit in plain language
The final report should state: “This campaign currently bids toward these events, using these values, with these known limitations.” Follow that with the proposed change and the evidence needed to judge it.
Success is not a larger conversion count. It is a campaign objective that matches the business, a measurement path that supports that objective, and a clear record of the tradeoffs the team accepted.