As of 13 August 2026, AI can only partly rank your leads by likelihood to buy.

What to actually do

1. Hand it to a person

   The route this page recommends

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

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   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 and export the current lead list with lead ID, company, role, source, creation date, recent activity, stated need, budget, decision authority, next step and last contact date.
   2. Export a separate set of comparable past leads with the same fields and the eventual outcome, such as became a customer, did not proceed or outcome unknown.
   3. Remove unnecessary personal data, reconcile duplicate lead IDs and replace missing values with clear labels such as unknown rather than guessing.
   4. Write down your product, target customer, sales process and definition of a successful outcome, then paste that context, the current leads and the historical outcomes into the prompt.
   5. Ask the chatbot to return the ranked table, scoring method, missing-data list, possible proxy risks and recommended next actions in separate sections.
   6. Compare the ranked lead facts against the CRM records, correct any mistaken activity or outcome, and ask the model to recalculate without changing the underlying evidence.
   7. Have a sales colleague compare the top and bottom groups with recent conversations, then record the final priority and next action in the CRM rather than treating the score as a forecast.

   Prompt

   Rank the leads in the data below by likelihood to buy, using only the information provided and making no invented assumptions. Treat the result as prioritisation, not as a factual probability or a sales forecast.
   
   Business context:
   - Product or service: [describe it]
   - Target customer: [describe the best-fit customer]
   - Sales cycle and buying process: [describe them]
   - Definition of a successful outcome: [for example, became a paying customer within a stated period]
   - Constraints or exclusions: [state any]
   
   Lead data:
   [Paste a CSV or table with one lead per row. Include fields such as company, role, sector, company size, source, date created, pages or materials viewed, email replies, calls or meetings, stated need, budget, decision authority, proposed solution, next step, last contact date, and outcome where known. Remove unnecessary personal data.]
   
   Past outcomes, if available:
   [Paste comparable historical leads and whether they became customers, including the time period and the fields available at the time of ranking.]
   
   Return:
   1. A ranked table with lead name or ID, rank, score from 0 to 100 labelled as a prioritisation score rather than a probability, evidence used, missing information, confidence level, and recommended next action.
   2. A short explanation of the scoring method and the strongest positive and negative signals.
   3. A separate list of leads that cannot be ranked fairly because key data is missing or contradictory.
   4. Any signs that the data may favour or disadvantage a type of lead because of source, sector, company size, location or another proxy.
   5. Three practical checks I should run against our CRM before acting on the ranking.
   Do not infer buying intent from a person’s name, protected characteristic or an unsupported demographic assumption. Do not claim that a lead will buy. If the historical data is too thin to support a meaningful comparison, say so and give a rules-based prioritisation instead.

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