LearnGrokUse these prompts to turn mixed research and CRM data into a segment document that a paid media team can activate. They are for campaign planners who need to decide whom to reach, whom to avoid, what each group needs, and which channel roles are justified.
Start with the evidence, not a preferred channel or a familiar audience label. The final prompt produces the decision record you can take into a planning meeting.
Key point
Start with addressability
A useful segment has a distinct need or barrier, and a way to reach or measure it. A descriptive label alone is not a media audience.
1. Gather the inputs before segmenting
Put the material into one working document before you use the first prompt. You do not need perfect data, but you do need the source, date and limitation of each item.
Include:
- Customer interviews, survey answers, review themes and sales-call notes.
- CRM field names and their actual values. Include definitions if someone has documented them.
- Conversion events, such as lead submission, demo booking, purchase, renewal or product use.
- Existing campaign results, including what audience was actually used.
- Offer rules, markets, consent status and groups that must not receive the campaign.
Do not paste a CRM export containing names, email addresses, phone numbers or free-text notes that identify individuals. The planning task needs field definitions, value distributions and aggregate patterns, not personal records.
Watch out
Do not treat a CRM field name as proof
A field called industry, status or lead source can mean different things across teams. Confirm its definition before you build targeting around it.
Run Prepare the audience evidence sheet first. Its job is to separate facts from labels. For example, recent trial user may be a usable CRM group, but it is not yet a reason to give that group a separate message. The research may show that recent trial users include people comparing alternatives and people who simply ran out of time. Those are different barriers, and may require different treatment.
2. Build segments that change the plan
Use Draft priority audience segments once you have the evidence sheet. Limit the output to a primary and secondary segment where possible. A long list creates the appearance of precision while leaving no clear buying decision.
A segment belongs in the plan only if it changes at least one of these:
- The message or proof point.
- The conversion action.
- The channel role or audience method.
- The suppression rule.
- The measurement signal.
For each segment card, look hardest at Need or job to be done, Trigger, Barrier and Addressable definition. These fields connect customer evidence to media execution. Operations teams is rarely sufficient. Teams replacing a manual reporting process after a compliance review gives you a trigger and a message direction, provided your evidence supports it.
Check
A segment is ready to plan against when you can answer three questions
What makes this group act now? What stops it acting? Which field, behaviour or media context lets you reach or measure it?
3. Separate exclusions from low priority audiences
Run Set exclusions and suppression rules before choosing final channel tactics. This prevents an otherwise sound plan from spending against people who cannot convert, have already converted, or should receive a different message.
There are two decisions here. An exclusion says the audience is not eligible. A suppression says the audience may be eligible, but not for this campaign at this time. Keep them separate in the schedule. A recent purchaser might be suppressed from an acquisition message, while a non-serviceable postcode is excluded altogether.
Check that each rule names the data field and the point where it is applied. Exclude existing customers is not implementable. Suppress records where customer_status equals active is implementable only if the field is current and its meaning is confirmed.
4. Give channels a job, not a default place
Use Match segments to channels to create a channel-role table. Do not ask it to select every channel available to you. Ask whether a channel has a specific role for a particular segment and whether you can measure that role.
The table should make weak choices visible. If the only justification is that a channel is widely used, the recommendation is unsupported. If you cannot confirm targeting or measurement capability, record that as a dependency. Capabilities and product limits can change, so check the current documentation at xAI documentation overview when your workflow depends on a specific feature.
Note
Keep test plans narrow
Test one meaningful uncertainty at a time, such as message barrier or audience definition. A test that changes audience, creative and channel together rarely explains what caused the result.
5. Check the final decision for unsupported claims
Run Make the media planning decision last. Read the Why this is the decision section line by line. Every claim should trace back to a research source, CRM field, campaign result or explicit constraint. If it does not, move it to Evidence gaps and tests.
Use this table to review common problems before approval:
| If you see this | It usually means | Do this |
|---|---|---|
| Two segments have the same need and message | The split is descriptive, not strategic | Combine them and retain the usable targeting distinction as a sub-rule |
| A segment has no addressable definition | Research is ahead of your data | Plan a test or measurement approach, not scaled targeting |
| A channel appears without a role | The plan began with inventory | Remove it or define the segment-specific job |
| An exclusion lacks a field and value | The rule cannot be activated reliably | Assign an owner to define the data rule |
When the prompts do not work
Do not force a neat segment document from thin evidence. If the output repeats generic labels, produces unsupported channel claims or cannot distinguish primary from secondary audiences, return to the evidence sheet. Add the missing CRM definitions, identify the source of conflicting research, and ask the campaign owner to rank the commercial objective.
If the evidence remains weak, make a smaller decision: choose one clearly addressable group, suppress obvious ineligible groups, and frame the rest as tests. That is more useful than presenting assumptions as settled audience strategy.
