Before changing bids, check what counts as a conversion

A lower reported cost per conversion can look like progress while the number of useful enquiries stays unchanged. Before changing bids or expanding the budget, ask whether the conversion signal represents the outcome you actually want.

The aim is not to make every report match perfectly. It is to understand what each system counts, where the gaps are, and which signal is appropriate for a decision.

Start with a conversion definition

Write down the business event first: a completed order, a qualified enquiry, or a booked consultation. Then document the technical event that represents it and the conditions under which it fires.

A form button click is different from a successful submission. A successful submission is different from a qualified lead. Both can be useful, but they answer different questions.

For each conversion, record:

• The business outcome it represents.
• The event name and trigger conditions.
• The system that records it.
• Whether it is used for reporting, bidding, or both.
• How repeat actions and test submissions are handled.

This simple inventory often reveals that several teams have been using the same word for different outcomes.

Test the event through the full journey

Complete the journey on desktop and mobile. Test a successful action, a validation error, a repeated submission, and a page reload after success. Inspect the event payload rather than relying only on an interface showing that a tag fired.

For an order, verify that the transaction identifier and value are appropriate. For an enquiry, check whether the event fires only after the submission succeeds. Keep personal information out of analytics payloads that are not designed to receive it.

Document the test conditions, including consent state. Measurement gaps caused by consent choices should not be confused with implementation failures.

Separate volume from quality

Suppose an illustrative campaign produces 40 enquiries, of which eight meet the sales team’s qualification criteria. A report based only on the 40 submissions cannot explain the cost of acquiring those eight useful opportunities.

Agree on a qualification rule that the sales team can apply consistently. Where supported and appropriate, examine whether later outcomes can be connected to acquisition through offline conversion workflows. Begin with reliable recordkeeping and permissions, not an assumption that every lead can be matched.

If the quality data is incomplete, show that limitation. Do not replace a missing outcome with an invented conversion value.

Reconcile without forcing equality

Platform reports, analytics reports, and business records can differ because they use different attribution rules, reporting windows, time zones, and observation methods.

Compare like with like before interpreting a discrepancy. Distinguish event counts from attributed conversions. Check conversion lag and whether the date shown refers to the interaction or the outcome.

A useful reconciliation note explains the main sources of difference and identifies any implementation problem that can be fixed. It does not silently adjust one report to match another.

Change bidding after the signal is understood

Once conversion definitions and event behaviour are clear, review which actions inform bidding and which remain diagnostic. Then evaluate the campaign using an observation period appropriate to its conversion cycle.

If the current signal is weak, increasing spend may simply produce more weak data. The first useful change might be measurement, the landing page, or qualification—not the bid.

For implementation review, explore our Analysis service. For account structure and optimisation, see Google Ads.

Reference

Google Ads: About conversion tracking explains the platform’s conversion tracking framework. Validate the details against your own account configuration.