What should your campaign learn to value?

Write down the customer action you actually want before changing an audience. A conversion is the action you choose to measure, such as a completed order or a useful enquiry. For this exercise, give that action a plain definition your sales team can apply.

Consider a hypothetical home cleaning company. Its owner wants recurring customers, while the campaign report counts enquiry forms. I would ask the team to record which enquiries fit the service area and which become bookings before deciding whether an audience change helped.

Check the journey yourself

Submit a test enquiry and follow it into the team's records. Check whether the service requested, contact details and final status are available to the person reviewing results. Label your test so it does not get mistaken for a customer.

Google's setup sequence starts with conversion tracking, followed by creating and adding audience segments and applying an automated bid strategy. Use that order for your preparation. Google's setup guidance.

If one enquiry appears more than once, investigate before making targeting decisions. Use the guide to duplicate Google Ads conversions to structure that check. Keep a written example of what a correct record should look like.

What does Smart Bidding actually change?

Smart Bidding changes bids using predictions about the opportunity in an individual auction. Machine learning means using patterns in data to make those predictions. Google describes contextual inputs including device, location and time of day. How auction-time bidding works.

Separate eligibility from bidding

Ask two questions during your review: who could receive an ad, and what should the campaign bid for an eligible opportunity? Keep separate notes for these questions. Calling everything “AI targeting” makes it harder to explain which setting you intend to test.

In a hypothetical online stationery shop, the owner might assume every previous visitor deserves a larger bid. I would turn that assumption into a question: which previous interactions are relevant to the purchase goal? Avoid making the size of a visitor list your success measure.

Google's example uses Maximize Conversions to pursue newsletter registrations. That is a useful illustration of matching a strategy to a measured action, not a reason to make newsletter signups your own sales campaign's goal. The Maximize Conversions example.

Before selecting a strategy, write whether you want more completed actions or more value from those actions. If different orders matter differently to your business, ask your campaign manager how that difference is represented. Do not accept “the AI handles it” as the entire explanation.

Which audience segments deserve a place in your test?

Choose a segment because it represents a specific customer hypothesis. Google Ads audience segments group people by interests, purchase intent or previous interaction with a business, with some characteristics estimated by Google. Treat a segment label as useful information rather than proof about every individual. About audience segments.

The descriptions below follow Google's definitions. The final column contains suggested questions for your own planning. Check availability for your campaign type before building the test around a particular segment. Segment types and campaign availability.

Write a reason beside each audience

For a hypothetical furniture retailer, “people who explored dining tables” is a more useful starting hypothesis than “everyone who likes attractive homes.” This is a planning example, not a claim about the performance of either audience. Write the offer and desired next action beside each group.

Keep your initial selection small enough to explain. If two groups have the same purpose, ask why both are needed. For help grouping your own customers before campaign setup, use the customer segmentation guide.

When is an audience a signal rather than a boundary?

Check the campaign context before assuming a selected audience restricts every impression. In Performance Max, audience signals guide Google's AI, but ads can reach people outside those suggested audiences. Google's Performance Max explanation.

Write “suggested audience” in your test notes when that is what you are providing. If the business requires a specific restriction, have the campaign manager identify the applicable control separately. Use the Performance Max guide to prepare that discussion.

Observation can still inform bidding

For Search campaigns using Smart Bidding, Google says your own audience segments added under Observation are used as bidding signals. Do not describe Observation as though the information has no role in automated decisions. How Observation audiences support Smart Bidding.

Ask your manager to show the audience, its setting and the business reason for adding it. Record those together. A screenshot without the reason is not enough for the next person reviewing the experiment.

How can you prepare a useful test this week?

Use the first week to establish a reliable test, not to force a winning result. The schedule below is a suggested work plan, not a platform requirement. Adapt it to the time between an enquiry and a completed sale.

1. Define success. Write the exact action and any quality conditions.
2. Save the starting position. Record spend, qualified enquiries and completed sales for a consistent period.
3. Check measurement. Follow a test action through the records.
4. Choose one change. State the audience or bidding hypothesis being examined.
5. Assign an owner. Name the person checking campaign settings and the person confirming sales quality.
6. Schedule a review. Allow time for the customer decision before interpreting results.

A hypothetical catering business

Imagine a caterer receiving both wedding enquiries and requests for small office lunches. Before adding more audiences, I would ask the owner whether these requests should be evaluated by the same business rule. Write down the service requested, whether the job is suitable and what happened next.

Then select one clear campaign question. For example: does the proposed audience information support the type of booking this campaign is intended to attract? Do not change the offer, enquiry form and audience together if you want to understand the audience change.

Review the destination page as part of preparation. Check whether it explains the service, location and next step clearly. The conversion rate optimization guide can help structure that review; keep any resulting page changes separate from the audience test.

Also write a spending limit that you are comfortable testing. Treat it as a business decision, not a prediction of the amount required for success. For a broader operating routine, consult the ad budget management guide.

How should you judge whether the change helped?

Review customer quality alongside campaign activity. I recommend keeping spend, qualified enquiries and completed sales together in the same review sheet. Decide which measure takes priority before looking at the result.

• If tracking is broken, repair it and mark the affected period.
• If enquiries are unsuitable, inspect the offer and audience hypothesis.
• If suitable enquiries do not become customers, review the sales conversation.
• If there is too little evidence, record an undecided result.

Separate what you observed from what you believe caused it. “More unsuitable enquiries were recorded” is an observation. “The new audience caused them” is an explanation that needs further checking.

Ask the team whether anything else changed: availability, prices, response times or the page customers visited. Include those changes in the review note. Use the small business KPI guide to keep the scorecard focused; a KPI is simply a measure used to judge progress.

Finish the review with one decision: continue collecting evidence, revise the test, stop it or expand it. Give the reason in ordinary language. Another team member should be able to understand that decision without knowing every advertising term.