Blog · Lead classification
How to automatically classify your store's leads: who buys today and who is just browsing
To classify your store's leads, sort them into three groups by what they said and did: hot (asks price, stock, shipping or payment), warm (asks generally or compares) and cold (just looking or gone quiet). Answer hot first, follow up warm inside WhatsApp's 24-hour window, and leave cold for campaigns.
Key takeaways: Not every chat is worth the same in the same minute; sort them by buying signals and spend your attention in that order.
Why classify chats instead of answering in order of arrival?
Because arrival order treats the person asking "do you have it in size M and do you ship to Austin?" exactly like the person who writes "info". When an ad campaign or the holiday season sends in more conversations than a small team can read, the chats left waiting are often the ones that were about to turn into a sale. While you carefully answer someone who is only curious, the decided customer gets tired of waiting and messages another store. Classifying is not cold; it is deciding with some judgment where your limited minutes go.
A classification also tells you what to do after the first reply. Without one, follow-up is improvised: you write to the people you remember and forget the rest. With one, every group has a next step and a deadline, and that can be repeated every day, even when you are not around. If you also want to decide which conversations a person handles and which an agent handles, this guide on AI sales agent or human seller will help.
What signals tell a hot, warm and cold lead apart?
What separates the groups is what the person asks and does, not how friendly they sound. A hot lead asks for the details you need right before paying: exact price, availability in their size or color, shipping cost and time to their city, payment methods. A warm lead shows interest but is missing something: they ask in general terms, compare two products, ask for photos or say "I'll think about it". A cold lead writes with no clear intent, asks the price once and disappears, or has gone days without answering.
| Group | Typical signals | What to do |
|---|---|---|
| Hot | Asks about stock, size, shipping to their city, payment methods, or says "I want it" | Reply first, with real data, and send the payment link or close the order |
| Warm | Asks for recommendations, compares, asks for photos, says "I'll think about it" or objects to the price | Solve the specific doubt and follow up within 24 hours |
| Cold | Writes "info", is just looking, stops answering after several messages | Spend no manual attention; include in future campaigns |
| Needs a person | Complaint, special case, large order, or a question the catalog can't answer | Hand it to a human with the full thread |
Notice that where the chat came from also counts. Someone who arrives from an ad for a specific product, or from a QR code in your shop, already brings clearer intent than someone who writes with no context. If you record where each chat comes from, as we explain in the WhatsApp link and QR with a prefilled message, that information adds to your criteria without asking the customer for anything extra.
How do you classify leads by hand in one afternoon?
You can start today with what you already have, using the labels in your messaging app or even a spreadsheet. The goal of this first exercise is not perfection but finding out which phrases repeat in your store and which ones announce a purchase. After one afternoon you will have your own criteria, far better than a generic list, and you will be able to automate them knowing what you are doing.
- Open your last 30 to 50 chats and split them into people who bought and people who didn't.
- In the ones that bought, find the first questions they asked and how long they took to decide.
- Write down the phrases of each group: buying ones ("do you ship to…?") and curiosity ones ("how much is it?" and nothing else).
- Create three labels, Hot, Warm and Cold, plus a fourth for cases that need a person.
- Set a reply time for each group, for example minutes for hot and the same day for warm.
- Each morning, start with the hot labels and finish by checking which warm leads were left without a reply.
The most useful thing about this exercise is that it shows your exceptions. A gift store may have hot leads asking about delivery dates, while a clothing store has them asking about sizes. Adjust the signals to what happens in your business, and review them every month against new chats.
What should you do with each group once it is classified?
Each group calls for a different action, and that is where classification turns into sales. With the hot lead, the rule is not to make them wait: answer with real price, stock and shipping time, and make paying easy, for example with a WhatsApp payment link. A common mistake is sending a long welcome paragraph before answering what they asked, and for someone who already wants to buy, that gets in the way.
With the warm lead, the goal is to remove the doubt that is holding them back. If the doubt was price, see what to reply when a customer says "too expensive"; if they stopped answering, use the method in following up with a customer who went quiet, with a useful message rather than pressure. Follow-up works best within 24 hours of their last message, because after that WhatsApp requires an approved template to write to them, and it stops as soon as the customer replies.
With the cold lead, the most profitable move is not to invest a person's time. Keep them in your contact base and talk to them again when you have something relevant: a launch, a restock or a coupon. For that you need their permission, and here is how to get opt-in. As for people who bought once and cooled off, a win-back workflow is usually worth more than chasing a stranger.
Which part of classification can an AI agent do?
It can do the repetitive part: read every conversation, assign it to a group and act accordingly, at any hour. Super Agent classifies each lead and follows up before it goes cold, inside WhatsApp's 24-hour window, and stops as soon as the customer replies. It answers on WhatsApp, Instagram, Messenger, TikTok, SMS, phone and your website chat with a single voice, so the lead who writes on Instagram and the one who writes on WhatsApp go through the same criteria.
Because it reads your Shopify or WooCommerce catalog, it answers with live prices and real stock, which are exactly the questions a hot lead asks. It only reads your store and never changes it. When a conversation needs a person, it pauses, lets the customer know and passes the full thread to your team, so nobody has to repeat questions. You can see how that step works in how to hand a chat from the bot to a person.
On results, one store that already uses it sells 22% more since it started, and another gets 55 times what it invested back in orders when it messages its own customers on WhatsApp. These are specific cases, not a promise for your store: they depend on your catalog, your traffic and what you do with each group. The Free plan costs US$0/month, needs no card and includes 250 WhatsApp campaign messages a month. We do the initial setup with you, free, in a 30-minute call, and you can create your free account here.
How do you know your classification is working?
You can see it in three things: how long you take to answer hot leads, how many warm leads end up buying, and how many leads go cold without a reply. You don't need a complex dashboard to start; each week, count how many chats fell into each group and how many ended in an order. If many "hot" leads don't buy, your signals are too broad; if many "cold" ones do buy, you are missing a signal you haven't noticed.
Also review the mistakes, because they are the best source of adjustments. If a customer was misclassified, look at what they said and add that phrase to your criteria. For a full measurement method, see the 6 numbers that matter for an AI sales agent. And if peak season is close, combine this classification with your Black Friday WhatsApp plan so the surge in chats doesn't catch you without a system.
Frequently asked questions
How many lead categories should I use?
Three are enough to start: hot, warm and cold, plus a separate label for cases that need a person. More categories usually complicate the work without changing what you do with each chat.
Can a customer move from one group to another?
Yes, and that is normal. A warm lead who asks about shipping becomes hot, and a hot lead who stops answering can drop back to warm. The classification updates with each message.
Do I need an online store to classify leads?
No. You can classify by what people ask in the chat. If you sell only by message, see how to take orders on WhatsApp without an online store.
When should I follow up with a warm lead?
Within 24 hours of their last message, while the conversation is still open on WhatsApp. After that you need an approved template, so it is better not to leave it for later.
Are leads from ads classified differently?
They usually arrive with more intent because they already saw a product. If you know which ad they saw you can answer more precisely; see how to answer leads from your WhatsApp ads.
Classifying doesn't have to be sophisticated: three groups, a few signals per group and one clear action for each already put you ahead of answering in order of arrival. Start by hand with your recent chats, adjust the criteria with what you see and, when volume overwhelms you, hand off the repetitive part so you can keep the conversations where a person makes the difference.
Sort your leads without reading every chat
Create your free account, no card needed. Put in your store's website and we'll write your first agent, which classifies leads and follows up for you.
Start for freeFree plan, no credit card. See plans and pricing.
Keep reading
- All the guides on the blog
- How to follow up on WhatsApp with a customer who asked a question and went quiet
- AI sales agent or human seller: which store conversations to give each one
- How to answer leads from your WhatsApp ads at any hour, with an agent that knows which ad they saw
- How to hand a WhatsApp chat from the bot to a person, without the bot talking over them
- Missed-reply revenue calculator
- Customer segmenter from an orders CSV
- Online store self-audit
- Free tools for online stores