At a glance
- Start by cleaning upBefore scoring anything, tidy the lead.
- Research and scoreA useful score looks at fit, intent and authority: does the company match your ideal customer, did they say what they need, and can this person decide?
- Let your rules decide the tierDo not let the model pick the outcome on its own.
- Route each tierLog the reasons on the contact as a note, so the next person can see why a lead was scored the way it was.
- Keep it reliableEvery CRM, calendar and email step should retry before giving up, and anything that still fails should tell someone with the lead and the reason.
Automation
- Clean up and drop spam
- Research the company
- Score it against your ideal customer
- Book, reply or nurture by your rules
- Lead scored
- Deal created
- Three times offered
Start by cleaning up
Before scoring anything, tidy the lead. Normalize the name and email, flag personal and throwaway addresses, spot sales pitches sent through your contact form, and handle the same person submitting twice as one lead.
Then check your CRM. An existing customer filling in the form is not a new lead; send them to their account owner instead.
Research and score
A useful score looks at fit, intent and authority: does the company match your ideal customer, did they say what they need, and can this person decide? A language model you connect (the template uses OpenAI) can read the company's website and your CRM history, then return a score with its reasons.
Keep the score's inputs in one place: a short description of what you sell and who your ideal customer is. Changing those two lines changes how every lead is judged.
Let your rules decide the tier
Do not let the model pick the outcome on its own. Turn the score into hot, warm or nurture with thresholds you set, and add your own rules on top, for example capping leads from personal email addresses at warm unless the score is very high.
Pick the owner by rules too, such as company size, and flag leads that arrive outside business hours so nobody promises a same-day reply.
Route each tier
- Hot: open a deal, offer three real free times from your calendar, hold the first one, and tell sales in Slack.
- Warm: reply with your booking link and tell sales.
- Nurture: a polite reply, and the lead stays in your CRM for later.
- Spam: dropped, with nothing written to the CRM.
Log the reasons on the contact as a note, so the next person can see why a lead was scored the way it was.
Keep it reliable
Every CRM, calendar and email step should retry before giving up, and anything that still fails should tell someone with the lead and the reason. Test it with a sample hot, warm and spam lead before you point your real form at it.
Updated 2026-10-28. General guidance, not advice for your situation; we scope each project in writing before any work starts.
Download the template
An n8n workflow that cleans up each new lead, researches the company, scores it against your ideal customer, and by your rules books hot leads into your calendar, replies to warm ones and logs everything in HubSpot and Slack. Free, with setup steps. You add your own accounts and settings.
Questions people ask first
Does the model write to my CRM on its own?
No. It returns a score and reasons. Your rules decide the tier, and the workflow writes to HubSpot and your calendar in the steps you can see and edit.
Can it quote prices or promise meeting times?
It is told never to quote prices. Meeting times come from your calendar's free and busy data, not from the model.
Which CRM does the template use?
HubSpot. The same approach works with other CRMs, which we can scope with you.
What does it need to run?
n8n 1.110 or later, an OpenAI API key, a HubSpot private app token, Google Calendar, an SMTP account and a Slack webhook.
Tell us what you need built.
Tell us the process and the tools involved, or call. We reply with questions or a written scope.
hello@pushbuild.com