How AI Lead Scoring Works (And Why It's Better Than Keywords)

Most lead monitoring tools work like this: you type in a keyword, they alert you when the keyword appears. Simple. Cheap. And mostly useless.

A post that says "I love Reddit so much, what CRM do you all use?" matches the keyword "CRM". So does "I hate how our CRM crashes every Monday". One is a buyer; one is a complaint. Keyword matching can't tell the difference.

What AI scoring actually measures

Brew Me A Coffee scores each candidate post on three dimensions:

Relevance (0–100): Does this post actually relate to the category of product you're building? A post about API costs is relevant to an API product; a post about design tools probably isn't.

Intent (0–100): Is this person looking to buy, switch, or evaluate? Signals include: asking for recommendations, expressing frustration with a current tool, mentioning a budget, describing evaluation criteria.

Urgency (0–100): Are they deciding now? "We're evaluating options this week" scores higher than "someday I might look at this".

The three scores combine into a final 0–100 score. Posts above a threshold get a Hot, Warm, or Cold tier. Every lead also comes with a plain-English explanation — so you can decide in 10 seconds whether it's worth a reply.

The prefilter

Running AI on every post would be slow and expensive. Before any AI runs, a fast keyword prefilter narrows tens of thousands of posts down to a shortlist. Only the shortlist hits the AI. This keeps costs near zero while maintaining quality.

The result: you see the handful of posts that actually matter, not a firehose of false positives.