Keywords, Audiences and Remarketing: Three Different Targeting Questions

AI-generated conceptual illustration; no real audience, advertising account or campaign result is shown.

“Reach the right people” hides several different questions. What is someone searching for now? Which audience description might apply to them? Have they previously interacted with the business? Those questions can inform the same advertising brief, but they do not describe interchangeable controls. This chapter separates them using Google Ads terminology and a fictional catalog.

A keyword connects an ad to a search request

A keyword match type describes how a Search keyword relates to a user’s query for auction eligibility. Google’s matching documentation distinguishes broad, phrase and exact match. Broad match can cover related searches; phrase match considers the keyword’s meaning; exact match can cover the same meaning or intent. Exact therefore does not simply mean identical typed characters. Eligibility is not a promise that an ad will appear.

Imagine a fictional catalog selling notebooks and planners. A proposed keyword describes planners, while a search query is the actual request someone enters. The reviewer asks whether the match rule fits the intended product question. This distinction concerns a present search, not proof that the person belongs to a particular audience or has visited the store before.

A negative keyword defines an exclusion

A negative keyword excludes matching Search queries under its negative match rule. Google’s negative-keyword explanation warns that these rules differ from positive keyword matching. Negative broad, phrase and exact have different conditions, and exclusions do not expand to close variants as positive matching does. Do not assume one negative covers every related intention or synonym.

The fictional catalog sells physical planners and does not provide downloadable planner templates. Its draft brief flags template-related requests for examination. That is a proposed exclusion question, not an implemented negative list or a prediction about which queries would be blocked. The reviewer still needs the intended words, negative match rule and campaign scope before approving any exclusion.

An audience describes a group, not a single query

An audience segment groups people using characteristics, interests, intent or prior interaction, as defined or estimated by the platform. Google describes options such as affinity, in-market and your-data segments, with availability varying by campaign type. A segment label is not a verified personal fact, an individual’s current search, or evidence that every member wants the same product.

The setting determines what adding that segment means. Google’s Targeting and Observation explanation distinguishes narrowing reach with Targeting from monitoring an additional criterion without narrowing reach through Observation, where those settings apply. An audience displayed in a report therefore does not establish that delivery was restricted to its members. Identify the campaign type and actual setting before interpreting it.

Performance Max uses audience signals differently again. Google’s audience-segment documentation states that those signals do not guarantee delivery only within the supplied audiences; ads can be served beyond them. A suggested starting audience should not be described as a closed boundary. This distinction does not imply that every campaign type has the same controls.

Four terms, four questions in the brief

Remarketing starts from a previous interaction

Remarketing concerns re-engaging people who previously interacted with a business. Google now uses “your data” terminology for this area. Its your-data explanation includes website visitors as one example. This is different from inferring a general interest: the proposed audience has a relationship to an earlier interaction. The term alone does not establish that a particular person is reachable.

For the fictional catalog, “people interested in stationery” and “eligible previous catalog visitors” are different proposed descriptions. Neither is an actual list in this example. A past visit would not establish purchase intent today, and showing a later ad would not by itself prove that the earlier visit caused a sale. Keep the relationship, the allowed use and the claimed result separate.

Permission and eligibility are part of the question

Google’s personalized-advertising policy restricts targeting in sensitive categories and the collection and use of data. It also requires compliance with Google’s EU user-consent policy where applicable. Having a technical way to identify a past interaction does not establish permission to use it for personalized advertising. Check the applicable category, data-use and consent requirements before treating a proposed remarketing audience as usable.

A generic diagram cannot certify an advertiser’s compliance or eligibility. The relevant product, campaign method, location and current platform requirements need their own review. This chapter provides no consent implementation or legal determination, and no customer information is collected or uploaded here. If an audience’s permitted use is unknown, record that as an unresolved prerequisite rather than assuming that prior interaction is sufficient.

Write a brief that preserves the distinctions

For the fictional catalog, the brief has three separate lines: a present product-search question, an audience-context question and a prior-interaction question. Each line identifies the intended campaign type and the role of the proposed control. A fourth line records the exclusion question. Nothing in the brief supplies a budget, creates an audience, installs tracking or authorizes an account change.

The reviewer can now ask a precise question: is the team trying to change which queries may qualify, remove an unwanted request, understand a group, or re-engage eligible previous visitors? If the answer changes halfway through the discussion, revise the brief before comparing options. A familiar word such as “targeting” should not hide a change in the decision being made.

For the organizational context, continue with Campaign, Ad Group or Performance Max: Where Does the Decision Belong?. For the objective assigned to automation, read Give Automated Bidding the Right Objective—and Read Its Limits. Those chapters address different questions from audience membership or keyword eligibility; a targeting proposal still needs its own scope.

To define what an outcome means, use From a Valuable Action to a Measured Conversion. Before calling any targeting idea successful, specify the business action, the evidence that would count and the limitations of that evidence. The fictional catalog has no observed delivery or sales results, so it supports an explanation of the choices, not a performance comparison.