What Should You Optimize in Google Ads? A Guide to Choosing the Next Change

A single account may surface many recommendations at once: add assets, alter bids, widen reach, revise keywords, or raise a budget. If you accept them all together, it becomes much harder to see which change produced a benefit and which created a new issue.

What evidence shows this change is worth doing next?

Begin with one outcome and a usable baseline

Specify the result you want to improve. In a lead campaign, that might be the cost of a qualified enquiry or a customer gained. In a shop, it could include contribution after advertising. In a reach campaign, rely on the objective it was funded to deliver, along with a separate justification for that spend.

Capture the present result, the date span, spend, relevant volume, and any known reporting lag. Confirm that the conversion setup did not alter during the comparison. Our conversions and profitability guide covers the commercial measures that a conversion count leaves out.

Do not optimize a report you do not yet trust. If purchase values are inaccurate, improving reported return on ad spend may simply reward the measurement mistake.

Sort findings into repairs, controls and experiments

Fixes correct something proven to be wrong: a broken destination, an incorrect price, or an event that triggers before submission succeeds. Safeguards limit exposure that falls outside the plan, such as confirmed irrelevant searches or an unintended location. Tests examine a still-uncertain explanation, such as whether a different offer would bring in more suitable enquiries.

These categories shape the next action. You do not need an advertising experiment to show that checkout returns an error. You do need evidence before saying that a new headline drove higher sales.

Ask four practical questions: How solid is the evidence? How much relevant spend or business outcome is at stake? How easy is it to reverse? What else might it affect? A large estimated uplift backed by weak evidence can merit less urgent attention than a small, verified fault.

Read recommendations as proposals

Google's recommendations documentation explains suggestions based on account history and an optimization score that estimates how the account is expected to perform. Judge the proposal against your own goal and limits.

For a budget suggestion, consider whether the campaign is delivering an affordable outcome and whether the extra exposure suits the business. For an expansion suggestion, identify what new traffic or inventory becomes eligible. For a bidding change, verify the conversion signal and the economic foundation of the target.

Similarly, Google's Quality Score guidance describes Quality Score as a diagnostic aid, not the account's performance target. Use a weak component to look for relevant work. Do not replace evidence of better customer outcomes with a higher diagnostic score.

Match the change to the campaign's structure

These are examples of questions to explore, not directions to make those edits in every account. Some reporting is partial or rolled up. Google's search terms guidance, for instance, explains why not every query appears.

Also look for interactions. Excluding a term at a broad scope can influence multiple campaigns. Changing an account-default goal can affect campaigns that rely on it. Check the impacted scope before applying a change meant for a single issue.

Design a test that can answer its question

Draft a hypothesis that links a change to an outcome. For example: “Stating the minimum project size on this page may lower unsuitable enquiries without reducing suitable enquiries enough to make acquisition more expensive.” This is a hypothetical test idea, not a guaranteed effect.

Decide on the main measure ahead of time, plus one guardrail. In that example, suitable-enquiry cost could be the primary measure, with suitable-enquiry volume and total spend as guardrails. Record the lead qualification rule so it is used consistently.

When an appropriate Google Ads experiment is available, use it to organize the comparison. Google's experiment guidance advises isolating one variable and selecting success metrics before testing. Verify support for the campaign type and the proposed change instead of assuming every experiment works in every case.

Before-and-after reporting can still help, though seasonality, promotions, and other changes may influence it. State that limitation plainly. A rise after an edit does not, by itself, prove the edit caused the rise.

Keep a change log and let results mature

Log the exact change, the affected campaigns, the date, the owner, and the intended result. Note any other events, such as a promotion, site outage, or price change, that could shape interpretation.

Wait for enough completed outcomes to make a useful decision, taking into account the business's sales cycle and Google's conversion delay. Set the observation window and spending limits before the test begins. There is no helpful universal promise that every campaign can be judged after the same number of days.

At review time, decide whether to keep, reverse, continue within the agreed limit, or mark the evidence as inconclusive. A test that cannot separate the options has not demonstrated success or failure.

Use AdPilot to act on a finding you understand

The AdPilot Optimize walkthrough explains findings from your account, including wasted search terms, weak ad copy, and asset-group work. Examine the supporting data, scope, and suggested action before relying on a recommendation.

That workflow can make the next task easier to identify. It does not provide the business's margins or prove that a lead turned into a customer. Bring those records into the decision when the objective calls for them.

When Search traffic is plainly irrelevant, use the negative-keyword guide. Rely on the audit checklist to plan continued work. If the campaign has not delivered anything, begin with delivery troubleshooting before you try to optimize performance.

Pick the change that you can justify, define, and assess. If you cannot describe the evidence and the desired result, the next step is generally to refine the diagnosis.

Common questions

Should I accept every recommendation to reach a perfect score?

No. Judge each recommendation by the business goal, the evidence, and budget limits. A score is not the same as a profitability calculation.

How many changes should I make at once?

Fix confirmed problems as necessary and document them. If you are running an experiment, keep the change you want to learn about isolated; when multiple changes happen together, attribution becomes more difficult.

What if a test produces no clear winner?

Log it as inconclusive. Determine whether additional observation is affordable and likely to resolve the question. Do not force certainty from a small difference.

Download the campaign worksheet