Lead quality on the dashboard
Tiles for hot leads (with warm and cold counts), hot leads waiting, how many active chats are labelled and how many the team corrected, and spam and job enquiries set aside.
The AI reads each chat and labels the lead hot, warm or cold, with what they want and why. Hot leads go to the top of the inbox, and if nobody answers one within the time you set, your team is told.
Labelled two minutes after her last message
One request, start to finish
It reads the whole chat Including voice-note transcripts and the ad she came from.
One label, with a reason Hot, warm or cold, plus intent, interest and urgency.
Hot leads rise to the top With Priority on, the list is ordered for working through leads.
It suggests the next stage Apply it and the deal moves as if you dragged it.
01: THE LABEL
On a busy day, the front desk answers whoever wrote last. The customer who asked the price and said she can come on Thursday waits behind ten people who said hello. Lead classification sorts that out. Once you turn it on, the AI reads each chat after the customer writes, on every channel, including chats a person is already handling.
It reads the recent messages, any voice-note transcripts, the ad the customer came from and any open deal. A burst of quick messages is labelled once, a minute or two after the chat goes quiet, so three messages in a row don't produce three labels.
Ready to act now. They asked the price and how or when to book, asked for the address or a slot, or said they will come, book or pay.
At least two real buying signals short of hot, such as naming the service, asking about packages or availability, giving a timeframe or budget, or replying again after you answered. Asking questions alone is not warm.
Everything else: a greeting, a bare price question, an ad's pre-filled message with nothing after it, general questions, and existing customers with nothing new to buy.
Intent (booking, pricing, information, existing customer, complaint, spam, job enquiry or other), interest (the treatment, product or service), urgency (now, soon or later) and a one-line reason.
Labelling can run on its own AI model, with a backup, separate from the model your assistant uses. Only models that can return a structured answer are offered for it. Numbers on your AI exclusion list, such as your own staff, are never labelled.
When the AI is unsure, it picks the lower label. A borderline lead shows as warm or cold, never inflated.
02: THE LEAD PANEL
The lead panel sits beside the chat in the inbox. Your team can edit any label. A value a person sets stays put: the AI never changes it until someone picks Let AI decide again. Reclassify runs the label on demand, even when automatic labelling is off.
When the contact has an open deal and the chat shows things have moved on, the panel suggests the next stage, such as Move to Qualified. Apply moves the deal exactly as the board does. Dismiss ignores it.
With tagging on (the default), contacts get Lead: Hot, Warm or Cold and Intent tags, swapped as the label changes. You can filter on them in Contacts and saved lists.
On by default. A chat read as spam or a job application is closed if nobody has replied yet. The panel offers Reopen, and a later label that disagrees reopens it.
03: THE ALERT
A hot lead cools quickly. Pick how long a hot lead may wait without a reply: 5, 10, 15 or 30 minutes, or off. The default is 10. When that time passes with no answer, the chat is marked Needs attention with the wait written on it, such as Hot lead waiting 10 min, and your team gets a Hot leads notification.
If the notification stays unread in the app for fifteen minutes, it is sent to you on WhatsApp. Each person chooses which notifications come this way.
A reply from a person or from the assistant before the time runs out means no alert.
A person's reply, closing the chat, or Mark as handled lowers it.
The assistant sees the label on every turn and is told to move a hot lead to the next concrete step, or hand it to a person.
See your own leads labelled
We reply in under an hour.
04: WHERE YOU SEE THEM
Labels show wherever your team works. You don't need to open a report to find the people worth calling.
Tiles for hot leads (with warm and cold counts), hot leads waiting, how many active chats are labelled and how many the team corrected, and spam and job enquiries set aside.
A Hot or Warm badge on each row, with the AI's reason on hover. Cold leads show no badge, which keeps the list calm. A Hot filter appears with its count once a lead is marked hot, and Priority sorts the list with hot leads near the top.
Intents ranked with bars, each opening the inbox filtered to it, with urgency underneath. Leads over time shows hot, warm and cold chats per day, counted as they were labelled that day.
A worklist with the longest wait first, the interest and the wait time. Each row opens the chat.
With labelling off and nothing labelled, the section shows one line on how to turn it on, with the button for admins, instead of a row of zeros.
05: PER AD
A good lead count can hide a bad ad. In Ad Funnels, each stage of one ad's funnel, each drop-off and the ad's leads overall show how many were hot, warm, cold, spam or job enquiries, or not yet classified. A line of 83 leads reads as 83 leads, 9 of them worth a call.
Put that next to what the ad's leads became in your pipeline and its revenue, and you can tell an ad that brings chatter from one that brings buyers.
The honest part
Labelling is off until an admin switches it on. Each label is a small AI call that counts against your monthly AI budget.
Leads are hot, warm or cold, each with a written reason your team can check.
When unsure, it picks the lower label.
Until someone picks Let AI decide again.
An existing label updates the next time that chat is labelled.
Leads over time starts from when labelling arrived, and earlier ranges say so instead of showing zeros.
Numbers on the AI exclusion list are skipped, even on Reclassify.
Works with
This page
Lead classification and alerts
Questions about lead labels
Yes. It runs after a customer writes on any channel, including chats a person is already handling.
No. A bare price question is cold. A lead is hot when they also ask how or when to book, ask for a slot or the address, or say they will come, book or pay.
Change it in the lead panel. The AI won't change it back until someone picks Let AI decide again, and the dashboard counts how many labels the team corrected.
You choose 5, 10, 15 or 30 minutes, or off. The default is 10. If the notification stays unread in the app, it reaches your phone on WhatsApp after fifteen minutes.
Each label is one small AI call, and it counts against your monthly AI budget like every other AI feature.
It only sets a chat aside as spam or a job enquiry if nobody has replied yet. The panel offers Reopen, and if a later label disagrees, the chat reopens by itself. You can also switch setting aside off.
Book a demo
A 30-minute call. We turn labelling on for a test number and show you how your kind of enquiries are sorted.
We reply in under an hour.