Audience segments for campaign targeting

Use one segment brief for each campaign decision. It turns scattered research, approved CRM signals and commercial objectives into a targeting recommendation that media, creative and landing-page teams can act on.

This method is for campaign managers and growth teams deciding who to reach, who to exclude and what evidence supports the choice. The result is not a long list of personas. It is a short, testable set of segments with clear boundaries.

1. Set the campaign decision first

Write the decision before you collect audience material. Put it at the top of a working document called Audience segment brief.

Complete these fields:

• Campaign objective: the action you need, such as qualified demo requests, trial starts or repeat purchases.
• Offer: the product, plan, message or incentive being promoted.
• Decision date: when targeting must be signed off.
• Channel constraints: the channels, locations, formats or account lists available to the campaign.
• Success signal: the outcome that will show whether the segment was useful.

A useful objective is specific about the action and the audience stage. Increase awareness among operations leaders is too broad for targeting. Generate qualified demo requests from operations leaders at firms using manual reporting gives you a condition to investigate.

2. Collect signals into one evidence sheet

Create a table named Audience evidence. Add only material that can support or challenge a targeting decision. Do not paste whole reports, meeting transcripts or CRM exports into the brief.

Use a row for each signal. Include the source, date, audience it refers to and the exact observation. Separate what someone said from what you inferred.

Evidence source	Exact signal	What it may indicate	Confidence
Customer interviews	"Month-end reporting takes two days"	A time-sensitive reporting problem	Medium
CRM opportunity notes	Prospects ask about integrations before pricing	Existing systems shape evaluation	Medium
Campaign results	A role-based message received qualified responses	The role may be reachable with this message	High
Competitor scan	Competitors lead with implementation speed	Setup effort is likely part of the category conversation	Low

Use approved, relevant data only. Remove direct identifiers unless they are necessary for your internal review. Do not treat a CRM field as fact just because it exists. Job title, industry and lead source are often incomplete or inconsistently entered.

If you use the model to sort notes or propose segment hypotheses, give it a bounded input and a fixed output format. Product behaviour and available capabilities can vary, so check the current xAI documentation before building a repeated workflow around a particular feature.

Try this prompt:

Using only the evidence below, propose no more than four audience segments.
For each segment, provide: segment name, inclusion rules, exclusions,
job to be done, evidence references, confidence, message angle and unknowns.
Do not invent firmographic, behavioural or intent signals.

Campaign objective: [paste objective]
Channel constraints: [paste constraints]
Evidence: [paste evidence rows]

3. Build segments from conditions, not personas

Take the proposed segments and rewrite them as a combination of observable conditions. A usable segment has three parts:

1. Who: a role, account type, customer stage or other approved audience marker.
2. Situation: the trigger, problem or operating condition that makes the offer relevant now.
3. Boundary: the condition that makes someone a poor fit for this campaign.

For example, avoid busy marketing leaders. Write: Demand generation leads at accounts with an active lead-routing problem, where existing CRM notes show hand-off delays, excluding accounts already in an active sales opportunity.

This definition tells the media team what to select, tells sales which records to suppress and gives creative a real tension to address. It also reveals missing data. If you cannot identify the situation from your available signals, do not pretend you can target it precisely.

Keep each segment in the same format:

• Segment name: short and functional, such as Reporting bottleneck teams.
• Inclusion rules: conditions that must be present.
• Exclusions: people or accounts you must not include.
• Evidence: links or references to rows in Audience evidence.
• Message tension: the problem in the audience's own terms.
• Targeting recommendation: the available channel criteria or account-list method.
• Confidence: high, medium or low, with one reason.
• Test: what response would support or weaken the segment.

4. Add exclusions before recommending reach

Exclusions protect budget and trust. Write them before you estimate audience size or draft ads. Check three kinds.

Exclusion type	Example	Action
Commercial	Existing customers on the promoted plan	Suppress from acquisition activity
Sales process	Accounts with an active opportunity	Confirm the hand-off with sales, then suppress or use separate messaging
Relevance	Accounts that lack the required operating condition	Remove from the targeting rule
Data quality	Records missing the field used for targeting	Keep separate, do not silently assign them to a segment

State who owns each exclusion list and when it was last checked. A segment definition is incomplete if nobody can apply its exclusions in the campaign platform or CRM process.

5. Rank segments for this campaign

Do not launch every plausible segment. Score each one against the current objective. Use a simple one-to-three score for evidence strength, reachability, message fit and commercial value. Add a short note for every score.

The highest total is not automatically the right choice. A smaller segment with strong evidence and a clear message may be the best first test. A broad segment with weak evidence can be useful only when the campaign objective is exploratory and the measurement plan can separate its results.

Write a one-line recommendation beneath each segment: Test first, Hold for more evidence or Exclude from this campaign. Name the reason. For example: Test first because CRM notes and interview evidence point to the same problem, and the condition can be identified in the account list.

6. Check whether the output is wrong

Read the brief looking for claims that exceed the evidence. The common failure is false precision: a neat segment description that assumes motives, budgets or buying authority that your sources do not establish.

The output is likely wrong when:

• different sources refer to different people, but you have combined them into one segment;
• the segment can only be described with vague adjectives such as innovative or growing;
• there is no practical way to apply the inclusion rules in the selected channel;
• an exclusion cannot be checked before launch;
• every segment receives the same message angle;
• the recommendation depends on one anecdote, one stale report or an unverified CRM field.

Mark these as unknown, not as facts. Then decide whether to gather more evidence, run a narrow test or remove the segment.

7. Make this a campaign habit

Save the completed Audience segment brief with the campaign plan. At the campaign review, add actual results beside the original recommendation: who responded, who qualified, who converted and which exclusions prevented wasted contact.

Carry forward only evidence that changed a decision. Archive segments that were too broad, unreachable or unsupported. Over time, this gives your team a usable record of which signals predict campaign response, rather than a growing folder of generic personas.

When the process does not work, return to the evidence sheet. Usually one of three things is missing: a specific campaign decision, a signal that can actually be used for targeting, or a clear exclusion rule. Narrow the objective, separate uncertain records and rerun the segment prompt with only verified evidence. If the model adds facts that are not in the sheet, reject the draft and require evidence references for every recommendation.