As of 13 August 2026, AI can enrich your sales lead records.

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, power-user skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open your CRM export and create a copy containing only the fields needed for matching, such as record ID, name, company, website, role and existing contact details.
   2. Remove unnecessary sensitive personal data and confirm your organisation's lawful basis, privacy notice and direct-marketing rules before sending any record to an AI or enrichment provider.
   3. Write the fields you want completed, the approved sources, the matching rules and the fields that must never be inferred.
   4. Paste a small redacted sample into the prompt and ask the chatbot to produce the evidence-backed output format before processing the full export.
   5. Open Apollo and configure its enrichment workflow with the same fields, matching rules and approved data restrictions, or use the validated prompt for a controlled batch.
   6. Compare every new value and evidence URL against the source, your CRM and your current account ownership before importing; reject unresolved matches and duplicate records.
   7. Import only approved fields into a separate CRM column, record the source and review date, and have the data owner approve the records before they are used for outreach.

   Prompt

   Enrich this sales lead data without inventing or guessing anything.
   
   Objective: complete only the fields listed below for each record.
   Fields to complete: [for example: company website, company size band, industry, UK region, current job title, business email status]
   
   Data and rules:
   1. Use only the supplied records and the approved sources listed below: [approved sources].
   2. Treat a match as valid only when the company and person clearly correspond. If there is any ambiguity, write " needs review" and explain why.
   3. Do not infer personal characteristics, buying intent, seniority, employment status or contact details from weak evidence.
   4. Do not create or guess email addresses, phone numbers, company facts or job titles.
   5. Preserve the original values in separate columns and add an evidence URL, evidence date, match rationale and confidence of high, medium or low for every new value.
   6. Mark each value as verified, probable or unresolved. Use unresolved when the evidence is missing, conflicting or out of date.
   7. Flag duplicate records and conflicting sources without choosing between them.
   8. Do not process special category personal data. Highlight any field that may create a UK GDPR consent, lawful-basis or direct-marketing issue for human review.
   
   Return a table with: original record ID, original values, enriched values, evidence URL, evidence date, match rationale, confidence, status and review flag. Then provide a short list of records that must not be used for outreach until checked.
   
   Records:
   [paste a small, redacted sample or an approved export here]

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