Reach relevant people using behaviour, location and interests

In Chapter 6 you chose a first audience. Chapter 7 examined the decision that audience might face, and Chapter 8 asked whether your explanation helps them choose fairly. Now you need a practical way for relevant people to encounter that explanation. Reaching people is not the same as finding a hidden list of everyone who will buy. You choose where a message could be useful, use imperfect clues about relevance and check what happens.

This chapter explains three common clues: behaviour (what someone did), location (where an offer can be useful or where a person may be), and interests (topics a person may care about). It also introduces contextual placement, which uses the content around an advert rather than a profile of the viewer. You will learn why a platform label is only an estimate, when a narrow audience becomes unhelpful, and how to plan a small test with privacy and fairness in mind. The written lesson stands alone. Two optional videos appear near the end.

By the end, you should be able to: distinguish a market segment from an ad-platform audience; choose a relevant signal for a specific task; explain the limits of behaviour, location and interest signals; compare contextual and personalised delivery; and record a testable reach hypothesis without inventing results or collecting unnecessary personal data.
A09-01-where-this-fits
Figure 1 in words: Your first audience is a provisional group you want to serve. The decision is what they need to resolve. The message answers a relevant question. Reach is the route by which that answer might be encountered. A successful impression of an ad does not prove that the right person understood the message.

Start with the task, not a targeting menu

Segmentation means dividing a broad market into groups with meaningfully different needs or responses. Targeting means choosing which group a particular marketing effort will serve. Positioning means explaining why an offer is relevant compared with alternatives. These strategic choices are connected, but none is identical to ticking an audience box inside an ad account. OpenStax describes geographic, demographic, behavioural and psychographic bases for segmentation, and asks whether a segment can be reached and served in a practical way (Albrecht, Green and Hoffman, 2023a; 2023b).

An ad-platform audience is a delivery or reporting definition offered by a platform. It may use observed actions, inferred interests, search activity, location signals or your own appropriately sourced customer data. A platform can estimate that someone belongs to a category without knowing their private reason for acting. Some campaign types may use an audience as a broad signal, then deliver beyond it. Google Ads, for instance, says Performance Max audience signals do not guarantee delivery only to the people specified (Google Ads Help, n.d.-a). Always check what a chosen setting actually controls in the current campaign type.

A channel is a route such as search, social, email, a community, an event or a relevant publication. A placement is a specific place in that route where a message appears. You can reach people through useful public content and a good page without buying an ad. Paid targeting becomes a possible method once you know the question, the offer and the resources you can responsibly use.

Ask four questions before touching a platform: Who can the offer genuinely help? What question would bring them to look for an answer? Where could that answer naturally appear? What evidence would tell you that the chosen route was useful? “Everyone interested in marketing” is not a meaningful answer for a specific task. Equally, a very narrow label is not necessarily accurate. Start with the decision, then select only the signals that help reach it.
A09-02-four-levels-table
Figure 2 in words: A segment describes a meaningful group and need. A target is the group selected for one effort. A platform audience describes who may receive or be reported under a delivery setting. A placement describes where the message actually appears. The same person may fit several groups, and the platform's category does not prove that they need the offer.

What counts as a relevance signal?

A signal is an observable or inferred clue that a message might be useful. It is not a diagnosis of a person's motivation. The distinction matters because two people can take the same action for different reasons, and an algorithm may classify them with incomplete information.

Behavioural signals can be recent searches, a page visit, an interaction with your own content or a past transaction, depending on what you are allowed to use. Location signals can relate to the area where an offer is available or where a device seems to be, as well as a place someone searched about. Interest signals can reflect a broad topic affinity, a stated preference or an inferred category. Contextual signals come from the current search query or page topic and need not depend on a long-term profile of the person. The Information Commissioner's Office (ICO, n.d.-a) distinguishes contextual advertising based on page content from behavioural advertising based on known or inferred interests and activities over time.

Begin with the least intrusive clue that can answer the question. A search for a task can be more relevant than a broad demographic label. A clearly identified service area can be more useful than precise location tracking. A public explanation in a relevant place may be better than following a visitor across sites. These are planning judgments, not a universal ranking of channels.
A09-03-signals-table
Figure 3 in words: Behaviour tells you what was done, not why. Location can indicate service relevance but may be approximate or represent interest in a place. An interest category is an inference rather than a fixed identity. Context tells you about the content being viewed at that moment. The table asks what each signal can and cannot reasonably support.

Behaviour: an action is not an intention

Behavioural segmentation groups people by actions or patterns: use of a service, response to content, repeat visits or benefits sought (Albrecht, Green and Hoffman, 2023a). In digital work, you may see page views, search terms, completed tasks or requests for help. A first-party observation is recorded through a direct relationship with your own site or service. A platform may also provide its own audience estimate. Do not confuse the two or assume that a number in a dashboard explains the person's motive.

Ask what action happened, how recently and in what context. A page visit may be research, accidental navigation, an existing customer's support task or a comparison by someone who will never buy. A completed enquiry may be stronger evidence of interest in that action, but not proof of a future sale. Recency means how long ago an action happened; frequency means how often it happened within a defined period. Both can matter, but neither reveals a private state of mind on its own.

Retargeting or remarketing generally means showing an ad to a person after some previous interaction with a site, app or business. It can help make a previously seen answer easy to find, but it can also feel intrusive or repeatedly show a message after the task has ended. It often relies on identifiers and tracking technology, so the data and permission checks below are especially important. A customer list uploaded to a platform is not automatically available for any new purpose just because the business holds it.

Before creating a behaviour-based audience, write a sentence that does not pretend to know too much: “This action may suggest that the person is asking [question], but it could also mean [other explanation].” Then ask whether the proposed message genuinely answers the question. If you cannot name a reasonable alternative explanation, you are probably over-interpreting the action.
A09-04-behaviour-ladder
Figure 4 in words: Record the action first. Form a tentative question it might indicate, write at least one other explanation, then choose a way to check. An observed click is evidence of the click. A motive remains a hypothesis until other relevant evidence supports it.

Location: where you can serve and where someone may be

Location can be relevant when an offer can only be delivered, used or supported in certain places. Begin with the service area: the places where you can actually fulfil the promise. This is a property of the offer, not an assumption about the person. A location-specific message should make the area and conditions clear on the landing page as well as in the ad.

A platform's location targeting is more complicated than drawing a line around a map. It may infer where someone is, where they are regularly present or which place interests them. Google Ads Help (n.d.-b) explains that its location options can include physical presence and location of interest, with the default advanced option including both. A search about a city does not prove that the searcher is physically there. The available controls and defaults can change, so inspect the exact campaign setting and its help text when you use it.

Avoid treating a city-level ad setting as a guarantee that every impression falls inside a service boundary. A phone may travel, an IP-derived location may be imperfect, and someone elsewhere can research a place they plan to visit. If delivery outside the service area is a problem, verify the available presence setting and also make eligibility plain before someone commits. Do not imply that a business serves an area it cannot serve.

Exclusion is a setting intended to avoid delivery to certain places or groups. It can reduce clearly irrelevant delivery, but an exclusion is also an ethical decision. It should follow service availability and a fair reason, not an unsupported assumption that people in a neighbourhood are less valuable. Where an offer concerns essential opportunities or sensitive issues, platform and legal restrictions require extra care.
A09-05-location-layers
Figure 5 in words: The service area is where the offer works. Estimated presence is where a platform thinks someone may be or regularly be. Location of interest is a place they have signalled interest in. These can overlap, but they are not the same. Check the setting and describe the actual service boundary clearly.

Interests: useful topic clue, uncertain personal label

An interest can mean several things. A person may explicitly choose topics to follow. A platform may infer broad affinities from activity and context. Someone may be actively researching a purchase now. These are different strengths and time horizons. Google Ads Help (n.d.-a), for example, distinguishes affinity audiences concerned with interests and habits, in-market audiences based on recent purchasing intent, and data segments based on earlier interactions with a business. The labels are platform definitions, not direct statements from every person in the group.

Do not confuse interest in a topic with need for your offer. A person may read about a topic for work, learning, curiosity or someone else. An interest label is a starting hypothesis for placement or a message, not a personality test. If the message depends on a narrow claim about a person's private circumstances, pause and reconsider the approach.

Intent is a person's readiness or purpose in relation to a specific action. It can be suggested by the question they are asking now more strongly than by a general hobby, but intent is still inferred. A search query can be ambiguous. The best response is often a page that answers the query honestly, offers relevant alternatives and lets the reader decide. Do not write a message that announces a sensitive inference about the person.
A09-06-interest-intent
Figure 6
Figure 6 in words: A broad interest suggests a topic may be relevant; a current question narrows the information need; a completed action tells you that the action occurred. The arrows do not mean everyone progresses through all three. Each clue leaves uncertainty about the individual's reason.

Contextual versus personalised delivery

Contextual advertising uses the page or query currently being viewed as the basis for placing an ad. Personalised or behavioural advertising can use information about a person across time, sites, apps or devices to predict interests and decide delivery. Both can be targeted in the ordinary sense of making a message relevant, but they involve different data practices (ICO, n.d.-a).

This distinction prevents a common mistake: assuming every relevant ad needs a detailed profile. You can publish an accessible answer on a topic page, choose a suitable publication or use a search query without building a list of visitors. Context still needs review; a page topic alone may not match your intended meaning, and advertising tools may combine contextual and personal signals. Check the actual product and settings rather than relying on a label such as “interest” or “smart.”

Privacy rules also depend on the technology and purpose. The ICO's current guidance states that using storage or access technologies, such as relevant cookies or similar identifiers, for online advertising requires consent. This covers technical ad selection and delivery as well as associated tracking and profiling. It also says contextual advertising can more readily meet PECR and UK GDPR requirements in principle, while still warning that any storage or access technology used for the ad purpose needs consent (ICO, n.d.-a). That does not mean every contextual placement is exempt from all privacy rules, or that every personalised activity follows one identical consent rule.

If you use personal data for marketing, you also need a valid UK GDPR lawful basis and to consider other applicable rules. The ICO notes that online identifiers such as cookie IDs, IP addresses and advertising IDs can be personal information, and that PECR can also apply to electronic messages and storage or access technologies (ICO, n.d.-c). These details belong in a real campaign review, not a guessed statement on a teaching diagram. A platform's built-in audience button does not remove the advertiser's responsibilities.
A09-07-two-routes
Figure 7
Figure 7 in words: One route starts with the current page or question. The other uses known or inferred activity over time. Either may use additional technologies, so both routes lead to a check of the actual data, partners, permissions and controls before publication. The diagram is a planning distinction, not a legal classification of every product.

Direct email and messaging need a separate check. The ICO describes direct marketing broadly enough to include activities that lead up to a targeted message, such as profiling, as well as the message itself (ICO, n.d.-b). Rules for email, texts and calls can differ from those for a public page or a contextual ad. Do not export a list and send a message merely because a platform can technically accept the file. Start with the purpose, the relationship, the source of the details and the permissions for that specific use.

If you want an additional plain-language explanation of the basic groupings, read OpenStax's introduction to market segmentation. For the deeper question of data use in online ads, use the ICO's current guidance on online advertising. These are optional reading links. The first explains a marketing concept; the second is the relevant UK regulator's guidance. An older draft or an unverified internal article URL should not replace either link.

Reach fairly: avoid harmful inferences and exclusions

A narrow audience may appear efficient in a dashboard while excluding people who could benefit. It may also unintentionally reveal sensitive assumptions in the creative. Ask whether a targeting criterion is tied to genuine service availability or a relevant information need. Do not infer health, finances, identity or vulnerability from a casual browsing trace and then write copy as if you know it.

Platform policies add their own restrictions. Google Advertising Policies Help (n.d.) lists categories of sensitive interests and limits certain audience methods in those categories. The settings and restricted categories can change. Check the current policy for the product, location and campaign type before using a sensitive signal. Compliance with a platform setting alone does not establish that a message is fair or that the data use is appropriate.

People within any segment differ. A group can help you choose an initial message, but it should not replace the ability to read the same clear terms as everyone else. Accessibility, language, device constraints and service eligibility can be more useful planning questions than an inferred personality. If an exclusion would affect access to an important opportunity, scrutinise the reason and seek suitable specialist advice for the real campaign.

The simplest ethical test is: Could you explain the criterion to a person who saw the ad, without embarrassment or concealment? Would they reasonably understand why the message was relevant? Could they avoid further contact when they choose? This test is a practical prompt, not a substitute for the current rules.
A09-08-method-check-table
Figure 8
Figure 8 in words: A service-area criterion should reflect where the offer works. A current topic or query should match the answer. A past-site-visit audience needs an alternative explanation and a data-permission check. A broad interest category should not be treated as a personal fact. The table helps review one proposed method before it is enabled.

How narrow should the audience be?

There is no ideal audience size that works for every message. A broad public explanation may help many people at the start of a question. A narrow route may help where the service is available only in a defined area or the message answers a specialised task. Narrowing is useful only when the criterion improves the match between what you can offer and what a person needs to decide.

Every extra condition can also remove people. If you require a location, an interest category and a past action together, a person who needs the answer may be missed because one platform signal is absent or wrong. A tiny audience may produce too little response to learn from, even if each label seems precise. Ask which condition carries a genuine reason for relevance. Remove conditions that exist only because the interface makes them available.

Platform delivery can further complicate the picture. Some audience controls restrict delivery, some merely supply a signal or reporting layer, and automated systems may seek people outside the specified signal. Read the current setting's description and inspect where ads actually appeared. A high-level estimate of potential reach is not a count of eligible customers. Neither is a low cost per click proof that the people clicking can use the offer.

Start with the clearest task-related signal and service boundary. Observe questions and outcomes, then narrow or widen for a stated reason. Keep an accessible route to the same essential information for people who do not fall inside an advertising segment.

Plan one small reach test

You do not need to buy an ad to practise. Choose one relevant question from the previous chapters and one place where an answer could naturally be encountered. It could be a public article, a search result, a community resource where sharing is permitted, or a paid placement you are authorised and equipped to run. If you use a live platform, check its current options and policies rather than following old screenshots blindly.

Write a reach hypothesis in a form that can be checked: “People seeking [task or answer] may encounter [message] through [route] because [relevance reason]. We will look for [meaningful response] during [period], while checking [limitation].” The brackets are prompts, not fabricated data. Include a next-step page that actually answers the promise in the message.

Choose one primary question to learn from the test. If you change audience, copy, placement, price and page at the same time, you will not know which difference mattered. A small test cannot prove that your chosen group represents the entire market. It can reveal whether the route was feasible, whether the message was understood and where the next uncertainty lies.

Impressions are times an ad is shown, not necessarily unique people or attentive readers. Clicks show an interaction, not proof of comprehension or a purchase. Conversions are actions the platform or business has defined for measurement; check the exact definition before comparing them. Google Ads Help explains its reporting columns and conversion definition in separate guidance (Google Ads Help, n.d.-c; Google Ads Help, n.d.-d). More impressions from an irrelevant audience are not progress toward the customer task. Where measurement itself uses storage or access technology for online ads, include that purpose in the appropriate consent review (ICO, n.d.-a).
A09-09-reach-worksheet
Figure 9
Figure 9 in words: Fill one row per decision: the reader's question, chosen first audience, relevance signal, route and message, necessary data check, outcome worth watching and next check. The cells are blank. Write “unknown” where you lack evidence rather than inventing a response.

Exercise: choose one route and defend it

Allow 30–45 minutes. This is a planning exercise, not an instruction to collect personal data or launch a campaign. If you have no suitable real evidence, keep your inputs as questions and assumptions. Do not create a fictional customer or reuse a story from slides.

  1. Name the real task. Use one question from your decision map: what must a relevant person understand or do? State who can genuinely use the offer and where it is available.
  2. List two possible routes. One can be public or contextual, and the other can be an audience-based option. For each, say where the person would see the message and why that place relates to the question.
  3. Write the signal and its uncertainty. Is it a current query, past action, broad interest or location? Name at least one other reason the signal could appear.
  4. Check service and fairness. Can you fulfil the promise in the selected area? Would the criterion wrongly exclude people or suggest a sensitive private fact? Revise or drop it if necessary.
  5. Map the data flow. Would the route use a cookie, similar device access, an uploaded list or other personal data? Who handles it? What purpose and control would the person see? If you cannot answer, do not assume a platform tool has made it acceptable.
  6. Write one honest message. It should answer the question and lead to a page with matching information. Keep claims and limits consistent with Chapter 8.
  7. Choose one meaningful check. Define what an impression, click, enquiry or other action means in this context. Record a time period and one result that would make you reconsider the route.
  8. Record the decision. Mark each statement observed, reported, secondary, assumed or unknown. Decide what you would learn before spending more time or money.
The completed worksheet should let someone else follow your reasoning without needing access to a hidden targeting dashboard. A good answer may be “use a public explanation first” or “this signal is too uncertain.” Narrowness is not a goal in itself.

Learn and revise

After a real test, inspect both delivery and usefulness. Did the chosen route actually show the message where expected? Could eligible people access the page? Did their questions match the information you supplied? Were there complaints, confusion or signs that the targeting felt intrusive? You may need to change the message, the placement, the service-area setting or the offer itself.

Do not claim success merely because one metric rose. A lower click count can coexist with more useful enquiries; a broad campaign can show many impressions without reaching people who can use the service. Compare measures only when their definitions, periods and conditions are clear. Note other changes that could explain the result. If you have a very small sample, describe it honestly.
A09-10-learning-loop
Figure 10
Figure 10 in words: Start with the task, choose a signal and route, check the actual data use, publish a relevant answer, observe a meaningful response and revise. Return to the task rather than assuming a larger audience or more precise tracking will fix a weak offer.

Optional videos

The two videos are complementary, not compulsory. Selection is based on publicly available YouTube descriptions and source information; neither video has been played in full for this draft. Verify playback, English accessibility and current screen details before publication. The chapter's conceptual and privacy explanation stands on its written sources.

Start with the concept: Market Segmentation (With Real World Examples) | From A Business Professor, Business School 101. Public descriptions indicate an explanation of segmentation. It is an independent educational channel; its title alone does not verify the speaker's academic credentials. Its video examples are not reproduced in this article. Use the video to revisit the difference between a useful group and a platform setting.

See a current platform application: How To Apply Audience Segments in Google Ads (Display, Search & PMax), Loves Data (2026). The description covers targeting versus observation and audience signals in several campaign types, with an introduction and sections through 08:48. Settings may change; use the current official help for any live campaign. This is a professional tutorial, not UK privacy guidance.

For a deeper academic teaching option, the shortlist in the publishing packet identifies an NPTEL/IIT Kharagpur segmentation lecture. The article itself needs no video to be understood.

What to remember

  • Choose the task and first audience before a platform audience setting. A platform category is an estimate.
  • Behaviour shows an action, not a motive. Location may mean presence or interest in a place. An interest is a clue, not a fixed identity.
  • Contextual placement can make an answer relevant without a long-term personal profile, though the actual technologies still need checking.
  • Storage or access technologies used for online advertising require the relevant consent under the ICO's current guidance. Personal-data uses and direct messages need their own appropriate review.
  • Use the smallest relevant, fair and permitted route. Make the message and landing page match.
  • Define what you will learn before spending, then revise the signal or route when evidence disagrees.
Next in the series: Chapter 10, Map the customer journey from first question to repeat purchase, will connect these individual decisions across a longer relationship. Add its canonical public link only after publication and verification.

References

Albrecht, M.G., Green, M. and Hoffman, L. (2023a) ‘5.1 Market segmentation and consumer markets’, Principles of Marketing. Houston, TX: OpenStax. Available at: https://openstax.org/books/principles-marketing/pages/5-1-market-segmentation-and-consumer-markets (Accessed: 14 November 2026).
Albrecht, M.G., Green, M. and Hoffman, L. (2023b) ‘5.4 Essential factors in effective market segmentation’, Principles of Marketing. Houston, TX: OpenStax. Available at: https://openstax.org/books/principles-marketing/pages/5-4-essential-factors-in-effective-market-segmentation (Accessed: 14 November 2026).
Google Advertising Policies Help (n.d.) ‘Restricted targeting in personalised advertising’. Available at: https://support.google.com/adspolicy/answer/143465?hl=en-GB (Accessed: 14 November 2026).
Google Ads Help (n.d.-a) ‘About audience segments’. Available at: https://support.google.com/google-ads/answer/2497941?hl=en-GB (Accessed: 14 November 2026).
Google Ads Help (n.d.-b) ‘About targeting geographic locations’. Available at: https://support.google.com/google-ads/answer/2453995?hl=en-GB (Accessed: 14 November 2026).
Google Ads Help (n.d.-c) ‘Key event, conversion and purchase definition’. Available at: https://support.google.com/google-ads/answer/6365?hl=en-GB (Accessed: 14 November 2026).
Google Ads Help (n.d.-d) ‘About columns in your statistics table’. Available at: https://support.google.com/google-ads/answer/2454071?hl=en-GB (Accessed: 14 November 2026).
Information Commissioner's Office (ICO) (n.d.-a) ‘How do the rules apply to online advertising?’. Available at: https://ico.org.uk/for-organisations/direct-marketing-and-privacy-and-electronic-communications/guidance-on-the-use-of-storage-and-access-technologies/how-do-the-rules-apply-to-online-advertising/ (Accessed: 14 November 2026).
Information Commissioner's Office (ICO) (n.d.-b) ‘Identify direct marketing’. Available at: https://ico.org.uk/for-organisations/direct-marketing-and-privacy-and-electronic-communications/direct-marketing-guidance/identify-direct-marketing/ (Accessed: 14 November 2026).
Information Commissioner's Office (ICO) (n.d.-c) ‘Plan direct marketing’. Available at: https://ico.org.uk/for-organisations/direct-marketing-and-privacy-and-electronic-communications/direct-marketing-guidance/plan-direct-marketing/ (Accessed: 14 November 2026).

Supplementary videos: Business School 101 (n.d.) ‘Market Segmentation (With Real World Examples) | From A Business Professor’, YouTube. Available at: https://www.youtube.com/watch?v=IrJ1cNIfmsk (Accessed: 14 November 2026). Loves Data (2026) ‘How To Apply Audience Segments in Google Ads (Display, Search & PMax)’, YouTube, 17 March. Available at: https://www.youtube.com/watch?v=v6JNavMZkvg (Accessed: 14 November 2026).