One person watched a video. Another viewed three products. A third bought yesterday. All might be called a “warm audience,” but the same ad will not necessarily help each of them.
Even an interaction close to purchase is an incomplete explanation. Someone may add to the basket just to see the delivery cost. They may then buy in store. They may be searching on someone else’s behalf. The audience becomes easier to manage when we keep these possibilities in the analysis.
A previous relationship is a starting point for communication, not a confirmation of current purchasing intention.
Retargeting or remarketing?
The terms are often used with overlapping meanings. Some teams use retargeting for ads aimed at people who have interacted before, and remarketing for a broader category of re-engagement, including email. This distinction is not a universal rule.
In some interfaces, Google has replaced the term “remarketing” with “your data.” The documentation describes segments built from previous interactions, including visits, app use or video activity. Google Ads: about your data segments
The definition matters more than the label: whom are we including, using which information, through which channel and for what purpose? Here, “retargeting” means paid ads based on previous interactions or relationships. Email and other communication are considered separately.
A warm audience includes different situations
| Observed situation | What we can safely assume | Question for the next message |
|---|---|---|
| Interacted with an asset | There was a measurable contact | What do they need to understand about the offer? |
| Viewed the category | Explored a theme or option | What differences deserve to be explained? |
| Started an action | There is a signal closer to the decision | Which condition might be unclear? |
| Submitted an inquiry | There is a recorded commercial contact | Is sales already handling it? |
| Made a purchase | There is a confirmed transaction | Do they need help, another product or time? |
These signals do not tell us someone’s income, vulnerabilities or private motives. Segment according to the offer, the information actually available and the applicable rules.
A “warm audience” includes different situations.
A previous interaction does not confirm current purchase intent.
INTERACTIONS
There was contact with an asset.
EXPLORATION
Someone explored a category or product.
ACTION STARTED
Someone started an action, with no confirmed outcome yet.
REQUEST
The sales team may already be handling it.
PURCHASE
The customer may need support or a different follow-up.
Illustrative framework. Application depends on the project’s context and data.
Three questions before we invest again
- First, what useful information can we add?
- Then, where does the message lead?
- Finally, why does this deserve investment compared with other priorities?
A fictional example: a store sells coffee machines. For a visitor to a comparison page, an explanation of maintenance and use may be useful. For someone who has already bought the model, the same invitation to buy it may be inappropriate. Information about setting it up may make more sense.
If the model is no longer available, a dynamic ad based on an old catalog can continue a promise that the store can no longer support. Product data, events and messages must be analyzed together.
Not every return visit calls for a discount. Someone may need clarity on the warranty, compatibility or delivery. The discount changes the economic conditions and may not answer the remaining question.
A defined segment does not mean exclusive delivery
In PMax, audience signals guide the system, but ads may reach people beyond them. Adding a visitor list as a signal does not automatically create a retargeting-only campaign. Google Ads: PMax audience signals
Customer Match uses contact data under the product’s conditions. Match rates and availability are not the same as the number of uploaded rows. Inventory and regional restrictions apply; Google documents limitations involving partner inventory and third-party exchanges in the EEA, UK and Switzerland. How Customer Match works
Check the actual setup: inclusions, exclusions, expansion, objective and network. A campaign name chosen by the team does not prove how it delivers.
How complete is the audience?
Not every visitor can be recorded or used for personalization. Consent choices, browser, identification, segment size and platform rules influence availability.
Server-side tracking can improve aspects of control and data quality. It does not create consent or recognize everyone automatically. We explain the architectures and limits in The button was clicked. Did the inquiry arrive?
Customer Match policy requires, among other things, data collected within a permitted relationship with users and compliance with applicable notice and consent obligations. A purchased list does not become acceptable simply by uploading it to the account.
Personalized ads in sensitive categories may face additional restrictions. Eligibility is checked for the actual category and platform before designing audiences.
Choose a lookback window that fits the product
An accessory for a weekend event and a design project do not have the same decision cycle. A lookback window chosen without regard to the product may include people whose moment has passed or exclude people who are still evaluating.
Base the window on purchase behavior and platform limits. No duration suits every business. More segments are not inherently better either: each should support a useful difference in message or decision.
Exclusions need updating. An online order, an offline purchase or an opportunity already being handled can change what communication is appropriate. Synchronization may have delays, and identification may be incomplete; we do not promise to eliminate every inappropriate impression.
The same person can meet several channels
Visitors, subscribers and customers may overlap. In a fictional example, group A has 600 identifiable people, group B has 400, and 150 belong to both. Together they contain 850 people, not 1,000. This assumes exact identification; aggregate platform reports do not always allow such deduplication.
600 + 400 doesn’t always mean 1,000.
The example assumes exact identification and the same period.
GROUP A
600 people in the group.
GROUP B
400 people in the group.
INTERSECTION
150 people are in both groups.
UNIQUE PEOPLE
600 + 400 - 150 = 850. Reports do not always provide the intersection.
Fictional example, not NOT MORE. results or forecasts.
Coordination means considering communication pressure and messages together. We do not assume a perfect frequency cap exists for the same person across all platforms. We use available controls and data that we can interpret correctly.
A good ROAS doesn’t end the conversation
People who already know the offer may be more likely to buy than new audiences. An order attributed to retargeting does not prove the order would not have happened without it.
We evaluate the cost, contribution, overlap and role of the campaign throughout the system. Where volume and configuration allow, a well-designed incrementality test can help. Without such a test, we present attribution as attribution, with its limits.
How we work at NOT MORE.
Start with the actual relationship and objective: clarification, inquiry, purchase, use or repeat buying. Connect segments with messages, pages and update rules, then choose a campaign structure that can deliver the plan.
Retargeting is a part of the promotion strategy. It needs people who come into contact with the business, a relevant offer and an experience that confirms the promise.
Segments need clear operating rules
The client confirms the source of order, inquiry and status-change data. NOT MORE. can build the segments, messages, materials and campaigns, and implement the data connections in scope. We define what needs updating and who acts when the commercial situation no longer matches the segment.
During ongoing management, we monitor distribution, observable overlaps and relevant results. We adjust messaging and investment within the approved limits, without assuming that every eligible person will be reached or that every attributed sale was generated exclusively by remarketing.