As of 13 August 2026, AI can segment your email list for better UK campaigns.

What to actually do

1. Use a tool built for this

   The route this page recommends

2. Do it yourself

   Second choice

   A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

   How to actually do it

   1. Open your CRM or email platform and export the list with only the fields needed for segmentation, such as consent status, customer status, purchase history, location, engagement and relevant preferences.
   2. Remove names, email addresses, phone numbers and other unnecessary identifiers, then record what each remaining column means and how missing values are represented.
   3. Write the campaign objective, offer, intended audience, existing exclusion rules and the email platform you will use.
   4. Paste the campaign brief, column definitions and an anonymised sample or aggregate counts into the prompt, then ask the chatbot to return the segment rules, counts, overlaps and manual review list.
   5. Compare every proposed rule with the source column definitions, recalculate the segment counts in your spreadsheet, and check unsubscribed or otherwise excluded contacts are absent.
   6. Create the segments in your email platform, import a small test audience, compare several records with the source export, and send only after a colleague or campaign owner approves the audience and message.

   Prompt

   You are helping me plan email marketing segments for a UK business. Use the data structure and campaign brief below to propose practical segments, but do not send anything or make legal conclusions.
   
   Campaign objective: [for example, promote a product, re-engage customers or invite event attendees]
   Audience and offer: [describe who the campaign is for and what they will receive]
   Available columns: [list the column names and what each means]
   Segment rules I already use: [paste them, or write none]
   Exclusions: [for example, unsubscribed contacts, recent purchasers or existing customers]
   Email platform and supported fields: [name the platform and any limits]
   
   I will provide either an anonymised sample of rows or aggregate counts. Do not ask for names, email addresses, phone numbers or other unnecessary personal data. Do not infer sensitive personal data or protected characteristics, and do not use proxies for them. Do not include anyone whose consent or permission status fails the exclusion rules I provide. If a required field is missing, say so instead of guessing.
   
   Return:
   1. A table of segment names, exact inclusion rules, exact exclusions, intended message and the fields needed.
   2. A proposed priority order for the segments, with a short reason tied to the campaign objective.
   3. The number of records in each segment if counts are available, plus the number excluded and any overlap between segments.
   4. A list of records or fields that need manual review, without reproducing unnecessary personal data.
   5. A short validation checklist I can complete before importing the segments into my email platform.
   
   Keep every rule reproducible from the supplied columns. Separate observed facts from recommendations, and flag uncertainty rather than inventing it.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

3. Hand it to a person

   The distant third

   A person who owns the outcome does this end to end, worth it when the failure is dear.