From WhatsApp Chaos to a Calendar Full of Hot Leads

The Challenge

If you sell anything online, you know the pattern. An ad or a post does its job, and your WhatsApp fills up. "Is this still available?" "What would it cost?" "Can I send you some photos?" Some of these people are ready to buy today. Some are just curious. Some want something you do not even offer.

They all get the same first conversation, and someone on your team has to have it. Over and over, usually in fragments, and always while the one enquiry that really matters waits in the same queue as everyone else.

That was the situation at an online marketplace that auctions enthusiast cars. Sellers found it through social media ads and got in touch on WhatsApp. The sales team spent a large part of every week on first conversations: which car is it, can you send photos, what do you want for it, when can we call you. Necessary work, but none of it needed their expertise.

The Approach

We built an AI agent that takes over that first conversation. By the time a colleague picks up the phone, the agent already knows the car, has photos or a link to the listing, knows the seller's minimum price, and has agreed a time the seller picked.

Every enquiry gets a quick, personal reply

The agent picks up every enquiry quickly and writes like a helpful colleague rather than a form: one question at a time, in the customer's own language, and it never asks for something the customer has already told it.

Every lead is sorted into the right pipeline

Not every car belongs in the auction, and treating every enquiry the same wastes everyone's time. So the agent sorts each seller into one of four routes:

• A car the marketplace should take on. Classics, sports cars and collector cars get the full intake and a booked call with the team.
• A clearly ordinary car. Instead of a polite no, the agent offers to put the car in front of the company's network of dealers, collects the photos and details they need, and hands it over so dealers can bid on it.
• A price below the minimum. The agent stops there, tells the seller a colleague will take a look, and flags the lead for review.
• Anything in doubt. The agent does not guess. It completes the intake without passing judgment and leaves the decision to a person.

Whatever the outcome, the team gets an alert with the reason, and the lead is handed to a colleague.

The dealer route is worth pausing on. In one market, the agent at first politely turned ordinary cars away, until it turned away a perfectly sellable car that dealers would gladly have bought. A lead you paid an ad to find is not worthless just because it does not fit your main offer. Now those leads go somewhere useful.

Just as important is what the agent never does. It never tells a seller whether their car is a fit, and it never names a price. It sorts. People decide.

The CRM keeps itself up to date

The agent fills in the lead record as the conversation happens. It notes the car the moment the seller names it, books the call into the part of the day the seller chose, and hands the lead over as soon as its part is done. Someone who fills in the ad form and then also sends a WhatsApp message ends up as one lead, not two. The team only records the decisions that need a person, such as whether a car is qualified or listed, with a single click.

The dashboard opens on what matters today: upcoming calls, soonest first. Below that, who is waiting on an answer from us, and which leads have been open the longest. The day starts with a call list, not an inbox.

People step in exactly when they are needed

The agent knows its limits. A complaint, a problem with an account, a question about a car that is already up for auction, or simply a request to speak to a person: the conversation goes straight to a colleague, with one calm message to the customer and an alert to the team.

Taking over costs the team nothing. They still use the company's WhatsApp number from their own phones, and when a colleague replies in a conversation, the agent steps back from it for good.

The ads learn what a good lead looks like

A form submission is not a good lead. A great car is. So as each lead moves forward, whether it responded, qualified, got listed, sold or was rejected, that outcome goes back to the ad platform. Campaigns can then be judged, and eventually optimised, on the leads that turned into business rather than on how many people filled in a form.

The Outcome

The team no longer spends its days in chats. It spends them on calls. Around 15 hours a week that used to disappear into first conversations now go into conversations with sellers who are ready to talk. The agent has booked more than 100 of those calls, and every month more than 30 cars that do not fit the auction are passed on to dealers.

Sorting turned out to be where most of the value was. The team only picks up leads that are ready for a person, and every outcome flows back into the ads, so campaigns can aim for the leads that turn into business. Model costs are typically a few cents per conversation.

Why the Team Can Trust It

Handing customer conversations to AI only works if you can trust what it says. A few principles made that possible:

• It only states facts the business wrote down. Prices, process, links and timelines come from one plain-language document the team maintains, and the agent is tested to make sure its replies stick to it.
• It checks whether it should answer at all before it writes a word, and if anything goes wrong, the conversation goes to a person.
• It never makes promises about price, acceptance or timing.
• Changes are tested first against dozens of realistic conversations, including customers who try to talk it out of its instructions.

What This Looks Like in Your Business

Swap the car for your own product and the pattern holds. The value is not in the answers themselves. It is in sorting, recording and handing over, so your team only spends time where a person makes the difference.

In an online store, the same approach would answer questions about sizing, stock and delivery from your own information, point ready buyers to the right product, collect the details of a wholesale or business enquiry for whoever handles those, and pass returns and complaints straight to a person.

In a service business, such as an installer, a clinic or an agency, it would qualify each enquiry, gather what you need to prepare a quote, and book the intake call into your calendar.

In B2B sales, it would separate the curious from the serious and fill your team's week with conversations worth having.

Tech Stack

What's Next

Next up is letting the ad campaigns optimise directly on the leads that turn into business, and bringing the same approach to new markets.