Account-Level Optimization in Google Ads: Choose the Right Conversion Goals

Three settings answer three different questions

A conversion action describes an event you measure. A standard goal groups related actions by category, such as purchases. Account-default goals are the normal goal selection for campaigns that inherit them; campaign-specific goals let a campaign use a different selection.

Primary actions normally enter the Conversions column and guide bidding when the campaign uses their goal. Secondary actions normally stay in All conversions instead. There is an important exception: an action included in a custom goal is used for reporting and bidding when that custom goal is assigned, even if the action is marked secondary. Google's primary and secondary action guidance explains that distinction.

Do not confuse the name of an action with the event it records. “Qualified lead” could still fire when anyone submits a form if the implementation is wrong. The setting describes how Google Ads uses the event; it cannot certify the lead's quality for you.

Nor does changing a default create one shared budget or make the account a single campaign. Goal selection, bid strategy and budget are separate controls. A default should reduce unnecessary setup differences, not erase a real difference between selling a subscription and recruiting partners.

Google recommends a common set of goals to support learning across campaigns. It generally discourages custom goals, and a campaign can bid toward only one custom goal. Start with a sensible shared default; make exceptions for a real business reason.

A higher CPA can be the better result

Consider a hypothetical software advertiser spending $5,000. Its report shows 500 form submissions, so cost per submission is $10. Sales accepts 50 of those leads, making cost per accepted lead $100. Those are two descriptions of the same period—not evidence that one team is ten times less efficient.

Now suppose the advertiser changes the campaign to optimize for accepted leads. Comparing its new $100 CPA with the old $10 CPA would make the change look disastrous before anything meaningful had happened. The event changed. Keep both series visible and compare the same outcome across periods.

That does not mean every business should immediately switch to its latest sales stage. A useful event can arrive too slowly or too rarely to support the decision you want to make. A faster event may be practical if it predicts quality reasonably well; a deeper event may be better once it is reliable and sufficiently frequent. Make that choice from your own volume and delay, not a universal lead-stage ladder.

Before touching the setting, reconcile a few real records. Does the form create one lead or several? When does sales accept it? Is the accepted event imported once, with the correct identity and time? If those answers are unclear, a new goal will not repair the underlying measurement. The Data Manager diagnostics guide covers that investigation.

Choose the campaigns before changing the default

Imagine an account with separate campaigns for software subscriptions and implementation services. Subscription purchases and service inquiries may both be valuable, but making both goals the default could give a campaign an easier outcome than the one its budget was meant to buy. Review its actual selection rather than assuming the account label tells the whole story.

Write down the small set of facts needed to make the change understandable: the affected campaigns, their present goal selections, the actions involved and the intended replacement. This can be a short note beside the account review. It does not need to become a new reporting process.

If only one campaign needs a different outcome, consider a campaign-specific selection rather than changing the account default. If the common default is genuinely wrong, check which campaigns inherit it before correcting it. Also inspect any custom goals, because their action membership can defeat an assumption based only on primary/secondary labels.

Make the smallest change that answers the question. Changing goals, budgets, targets and creative together may be operationally necessary, but it makes the result harder to explain. Where an eligible experiment can isolate the goal change, use it; otherwise be candid about the limits of a staged comparison.

Keep the result comparable while the campaign adjusts

Keep tracking the business outcome you cared about before the edit. In the software example, that means accepted leads and their eventual value, not simply whichever conversion count now appears largest. Record when the settings changed so the comparison can use the same actions and allow for outcomes still arriving.

Separate an implementation failure from a disappointing business result. An import that stops, a duplicated event or an unintended goal selection deserves immediate investigation. A few expensive days after a strategy change require more context: conversion delay, available volume, traffic changes and the campaign's learning state.

Google's goal-transition guidance covers Search and Shopping, not Search Ads 360. It advises setting budgets to the intended amount before changing goals. Its learning guidance is not a promise that every account settles after a fixed number of days; allow the relevant events to arrive.

Restoring the previous goal settings can stop an unintended configuration from continuing. It cannot undo spend, recover missed auctions or instantly restore the old performance. Treat recovery as a new change with a clear reason, rather than a guarantee that the account returns to yesterday.

Finally, secondary does not mean Google can never learn from the event in any product. Google's announced journey-aware bidding beta lets Search campaigns on Target CPA learn from biddable and non-biddable conversion goals. That is separate from ordinary goal selection; do not assume every campaign has the beta.