A CRM run automatically
Automation 11 min
0/21
Business problem
A CRM dies because updating it is administrative work: enter a lead, add a note, change a stage, plan a follow-up. Salespeople don't do it (because it doesn't sell), so the CRM is empty, the pipeline doesn't reflect reality, and decisions are based on memory instead of data. Automation solves this at the source: the CRM updates when the work happens.
Proven method
An automatic CRM in 9 functions:
- LEAD CREATION: a new lead (form, import, reply) enters the CRM automatically (with dedup from module
- ,
- ENRICHMENT: data completes itself (company, contact, technology),
- TAGGING: a lead gets tags by rules (segment: e-commerce, size: 40 people, source: campaign X, status: qualified),
- SCORING: the score from module 5 saves automatically (87 → the 'score' field),
- STAGE UPDATES: the deal stage changes by action (a reply → 'conversation', a demo → 'demo', an offer → 'quote', a signature → 'won'),
- NOTES: call and email notes land in the deal automatically,
- TASKS: AI creates tasks with deadlines (book the demo, send the offer, follow-up on Friday),
- NEXT ACTIONS: the 'next step' field updates after every action (the system knows what's next),
- DEAL SUMMARIES: a 3-sentence summary of every deal (what we know / what we agreed / what's next). Rules:
- CRM automation works on data from other automations (enrichment, classification, inbox) — there's no 'separate CRM automation',
- a human corrects, doesn't type (AI creates, you edit on errors),
- a weekly pipeline review — automation doesn't replace a review,
- history: every change has a trace (who/what/when), so it can be debugged.
Market example
- BAD
- Manual, inconsistent execution or chaotic automation lacking data validation.
- GOOD
- A lead from a form: the CRM creates a deal automatically (lead creation + dedup), enrichment adds the company and contact, tagging: 'e-commerce, 40 people, june-campaign', scoring: 87 (module 5), stage: 'new'. The client replies to an email: stage → 'conversation', notes: 'wants a demo, asks about the price', tasks: 'book the demo by Wednesday', next action: 'demo Wednesday 11:00'. After the demo: stage → 'demo', notes completed, task: 'quote by Friday'. The pipeline review on Friday shows the current state — without typing anything by hand.
Ready-to-use prompt
You are a CRM automation architect. Context: CRM: [CRM]. Process: [PROCESS — from lead to deal]. I type by hand: [WHAT YOU TYPE]. Integrations: [MAKE/N8N + EMAIL TOOL + AI]. Task: Design the automatic CRM: 1. Lead creation: where leads come from and what happens on entry (dedup, fields) 2. Enrichment + tagging + scoring: how they automate 3. Stage updates: a stage map + what actions change a stage (a table: action → new stage) 4. Notes: where notes come from (calls, emails) and what lands in the deal 5. Tasks + next actions: how AI creates tasks and updates the next step (a template) 6. Deal summaries: a 3-sentence summary template + when it updates 7. Pipeline review: a weekly report (what it shows, how long) Format: 7 sections.
Model rekomendowany: Claude 3.5 Sonnet / GPT-4o Copy and paste into ChatGPT, Claude or Perplexity
ALTERNATYWNE SCENARIUSZE
Variations & alternative scenarios
Basic version: lead creation + scoring + tags (the rest manually).
Full version: 9 functions + inbox/classification integration.
Practical exercise
Design the stage map in your CRM + the actions that change a stage (a table). Write a deal summary template (3 sentences). List 3 functions to automate first.
Zapisano w przeglądarce