As of 13 August 2026, AI can only partly choose a CRM for email marketing.

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 1 hour until you can act on the result.

   How to actually do it

   1. Open a document and record your email purpose, contact count, expected growth, sending volume, budget, team skills, required integrations and the functions you need.
   2. Ask the people who handle sales, customer support, data protection and website technology to add their must-have requirements and unacceptable risks.
   3. Collect the current product pages, supplier quotes, integration documentation, data-processing information and contract terms for the CRM candidates you are considering.
   4. Paste the business requirements and source material into a chatbot with the copyable prompt, then ask it to identify missing information before it scores the candidates.
   5. Replace any unknown feature, price or compliance statement with a question to the relevant supplier, and request written answers rather than relying on the chatbot's summary.
   6. Run the strongest candidates through a small test using non-sensitive contacts, checking importing, segmentation, automation, unsubscribe handling, reporting, integration and data export.
   7. Compare the trial results and supplier answers against your weighted criteria, then record the final choice, approval owner and migration plan before signing a contract.

   Prompt

   Act as a procurement analyst helping a UK business choose software for email marketing. Recommend a shortlist and a final choice only from the products and source links I provide. Do not invent features, prices, integrations, compliance claims or customer results, and mark any item you cannot verify as unknown.
   
   Business context:
   - Business type: [business type]
   - Main email use: [newsletters, customer updates, marketing automation, B2B outreach or other]
   - Number of contacts now and expected growth: [figures]
   - Emails sent each month: [figure or estimate]
   - Budget and billing preference: [details]
   - Required integrations: [website, ecommerce, accounting, CRM or other systems]
   - Required functions: [segmentation, automation, forms, templates, reporting, journeys or other]
   - Team size and technical ability: [details]
   - Existing data and migration needs: [details]
   - UK requirements or constraints: [UK GDPR, consent records, data hosting, procurement or other requirements]
   - What matters most: [rank the criteria]
   - What I will not accept: [constraints]
   
   Candidates and sources:
   [paste supplier links, product pages, notes and any quotes here]
   
   Produce:
   1. A short statement of the requirements you inferred, separating facts from assumptions.
   2. A comparison table covering each requirement, evidence from the supplied sources, unknowns and likely gaps.
   3. A weighted score out of 100 using these weights: [insert weights that total 100]. Explain each score.
   4. The best-fit choice, the strongest alternative and the reasons not to choose each other candidate.
   5. A verification checklist of exact questions to ask each supplier, including consent and unsubscribe controls, data processing, export and deletion, integrations, deliverability, support, migration and contract terms.
   6. A practical trial plan using a small non-sensitive test list, with pass or fail criteria.
   7. A final recommendation that says whether to buy now, run a trial first or keep researching.
   
   Do not give legal advice. Flag anything that needs confirmation from a UK data-protection professional or the supplier. Keep the recommendation tied to the stated requirements, not general popularity.

   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.