Back to docs
Lead scoring is how Kilo separates genuine buying signals from casual browsing. Every conversation gets a qualification score from 0 to 100, based on real signals the AI detects during the conversation. When a visitor's score crosses your threshold, your team gets notified immediately.
This means your sales reps only spend time on leads that have demonstrated real intent — not every person who says "hi" to the chatbot.
Kilo looks for five key signals during each conversation. Each one adds to the overall score:
Persona detected — The visitor mentioned their role or title, like "I'm the VP of Sales" or "I lead our engineering team." This matters because knowing who you're talking to tells you a lot about whether they can buy.
Pain point identified — The visitor described a specific problem or challenge, like "We're losing deals because we can't track pipeline health." Explicit pain is one of the strongest buying signals because people who articulate problems are actively looking for solutions.
Domain captured — The visitor shared a work email address (e.g., jane@acmecorp.com). This is valuable because it identifies the company and confirms they're a real business prospect, not a student or competitor snooping.
5+ messages exchanged — An engaged conversation with multiple back-and-forth exchanges. Short conversations (1-2 messages) are usually casual. Longer conversations indicate genuine interest and active evaluation.
Demo requested — The visitor explicitly asked for a demo, trial, or meeting. This is the clearest buying signal — they're ready to take the next step.
Each signal contributes points toward the total score. A visitor who matches all five signals will score near 100. A visitor who only says "hi" and asks one quick question will score near 0.
Your score threshold determines when Kilo notifies your team. You set it in Notifications settings in your dashboard.
When a visitor's score reaches your threshold during a conversation, three things happen:A Slack notification fires with a detailed lead brief (persona, company, pain points, qualification data, and a link to the full transcript).
An email alert goes to the account owner with the same information plus a one-click link to the conversation.
The conversation is marked as "handed off" in your dashboard, so your team knows to follow up.
The right threshold depends on your sales team's capacity and how aggressive you want to be:
60 (default) — a balanced setting that catches most genuinely qualified leads while filtering out casual browsers. This is the right starting point for most teams.
40 (aggressive) — you'll get more notifications, which means more leads reaching your team, but some will be lower quality. Use this if your team has bandwidth and wants to cast a wider net, or if you're in a high-volume market where every conversation matters.
80 (conservative) — only the most qualified leads trigger notifications. Use this if your team is small, your product has a high price point, and you only want to talk to prospects who are clearly ready to buy.
Tip: Start at 60. After your first week, review the leads that came through. If too many were unqualified, raise the threshold to 70. If you feel like you're missing good ones, lower it to 50. It takes most teams one or two adjustments to find their sweet spot.
Every conversation in your dashboard shows its current score. You can sort and filter by score to quickly find your hottest leads.
Scores update in real time as conversations progress. A visitor might start at a score of 10 (just said hello), move to 40 after describing their problem, and jump to 75 when they share their work email and ask about pricing. If they request a demo, they might hit 90+.
"Can I customize which signals affect the score?" Not currently — the five scoring components are fixed. However, the framework you choose (MEDDPICC, BANT, or SPICED) influences which qualification questions the AI asks, which in turn affects how quickly visitors reveal scoring signals.
"What if a great lead doesn't hit the threshold?" You can always browse all conversations in your dashboard, not just the ones that triggered notifications. Check in periodically to see if any interesting conversations flew under the radar, especially in your first week while you're dialing in the threshold.
"Does the score reset between visits?" Each conversation has its own score. If the same visitor returns and starts a new conversation, it starts fresh. This is because their intent and context may have changed between visits.
Lead Scoring
How Scoring Works
Lead scoring is how Kilo separates genuine buying signals from casual browsing. Every conversation gets a qualification score from 0 to 100, based on real signals the AI detects during the conversation. When a visitor's score crosses your threshold, your team gets notified immediately.
This means your sales reps only spend time on leads that have demonstrated real intent — not every person who says "hi" to the chatbot.
What goes into the score
Kilo looks for five key signals during each conversation. Each one adds to the overall score:
Each signal contributes points toward the total score. A visitor who matches all five signals will score near 100. A visitor who only says "hi" and asks one quick question will score near 0.
Setting your score threshold
Your score threshold determines when Kilo notifies your team. You set it in Notifications settings in your dashboard.
When a visitor's score reaches your threshold during a conversation, three things happen:
Choosing the right threshold
The right threshold depends on your sales team's capacity and how aggressive you want to be:
Tip: Start at 60. After your first week, review the leads that came through. If too many were unqualified, raise the threshold to 70. If you feel like you're missing good ones, lower it to 50. It takes most teams one or two adjustments to find their sweet spot.
Reading scores in your dashboard
Every conversation in your dashboard shows its current score. You can sort and filter by score to quickly find your hottest leads.
Scores update in real time as conversations progress. A visitor might start at a score of 10 (just said hello), move to 40 after describing their problem, and jump to 75 when they share their work email and ask about pricing. If they request a demo, they might hit 90+.
Common questions
"Can I customize which signals affect the score?" Not currently — the five scoring components are fixed. However, the framework you choose (MEDDPICC, BANT, or SPICED) influences which qualification questions the AI asks, which in turn affects how quickly visitors reveal scoring signals.
"What if a great lead doesn't hit the threshold?" You can always browse all conversations in your dashboard, not just the ones that triggered notifications. Check in periodically to see if any interesting conversations flew under the radar, especially in your first week while you're dialing in the threshold.
"Does the score reset between visits?" Each conversation has its own score. If the same visitor returns and starts a new conversation, it starts fresh. This is because their intent and context may have changed between visits.