Automation should make your salespeople better, not invisible
Automating lead capture pays off when it puts better conversations in front of your sales team. Why replacing people misses the point, and what to automate.
There’s a version of website automation that treats salespeople as a cost to be engineered away. The chatbot answers everything, the forms qualify everything, the emails nurture everything, and somewhere at the end of the funnel a human signs the contract, if the board still insists on one.
We think that version gets the economics backwards. For most of the businesses reading this (dealerships, estate agencies, B2B services, agencies), the salesperson is the conversion. The website’s job is to put the right people in front of them, at the right moment, already understood. Automation that does that is enormously useful. Automation that tries to replace the conversation usually just makes it worse.
What customers are actually buying
Think about what someone buying a used car, a flat or a six-month consulting engagement wants from the process. They want answers to factual questions quickly. They want to feel confident they’re not being taken for a ride. And at some point they want to talk to a person who knows what they’re doing and can make a judgement call: “this one’s had one owner and a full service history, but if you’re doing long motorway runs I’d look at the diesel instead.”
The first of those is a great job for automation. The second is mostly about honesty and consistency, which automation can help with. The third is a person. Nobody wants to negotiate a trade-in with a widget.
So the useful question isn’t “how much of the sale can we automate?” It’s “what’s getting in the way of our people having good conversations, and can software remove it?”
The jobs worth taking off people’s plates
In our experience the list is fairly consistent across industries:
- Answering the obvious questions at 22:00. Opening hours, whether the car is still available, whether the flat allows pets. A visitor shouldn’t have to wait until morning for facts you’ve already published.
- Collecting the basics. Name, contact details, what they’re interested in, rough timing. Every salesperson hates doing this on a call, and every customer hates repeating it.
- Booking the meeting. Back-and-forth emails to find a slot are pure waste. Let people pick a time.
- Getting the lead to the right person. The right office, the right specialist, the person who covers that postcode. Manual triage is slow and tends to favour whoever checks the inbox most.
- Writing it down. Logging the conversation into the CRM so the next person to pick it up has the history.
Notice what’s on that list: tasks that sit between the customer and the salesperson. Strip them out and the salesperson’s first contact with a lead is a real conversation, not an intake interview.
Where it goes wrong
The failure mode we see most often is automation that’s measured on deflection. If the chatbot’s success metric is “conversations resolved without a human”, it will get very good at keeping humans out of conversations, including the ones that should have reached a human ten minutes ago.
You’ve probably been on the receiving end of this. You have a slightly unusual question, the bot doesn’t understand it, offers you three help articles, and the “talk to a person” option is buried or missing. You leave. The dashboard records a resolved conversation.
The second failure is invisible handover. The lead arrives in someone’s inbox with a name and a phone number and nothing else. The salesperson rings, and the customer has to explain everything again. From the customer’s side, the automation was a hoop to jump through before the real process started.
The third is tone. A business whose salespeople are warm and knowledgeable puts a stiff, over-cautious assistant on the website and wonders why the leads feel colder. The website is a salesperson too, as we argued in your website is a salesperson; if it doesn’t sound like your team, it’s representing a different company.
What “better” looks like
Here’s a scenario. A buyer looks at a family estate car on a dealer’s site on Sunday evening. The site’s assistant answers two questions about the boot size and the service history from the listing data, then offers a test drive. The buyer picks Tuesday at 17:30.
On Monday morning the salesperson opens the lead and sees: which car, what was asked, the answers given, the test drive slot, and the fact that the buyer mentioned a part-exchange. They ring the customer, confirm the slot, ask about the part-exchange and suggest a second car that might suit them better.
That salesperson isn’t invisible. They’re better prepared than they’d have been from a phone call cold. The automation did the clerical work and handed over a conversation already in progress. That’s the whole idea.
Compare that with the version where the assistant tries to close the deal itself, quoting finance figures and pushing an online reservation. Some buyers will want that, and fine, let them. Most will want to sit in the car first and talk to someone.
Questions to ask of any automation
If you’re evaluating a tool, or reviewing one you already run, a few questions cut through the brochure:
- When does a human take over, and how fast? There should be a clear, visible route to a person, and rules for when the system hands over without being asked.
- What does the salesperson see? Ask to see an actual lead as it arrives. If it’s a name and an email address, the automation is collecting data for itself, not for your team.
- What is it measured on? Appointments booked and leads accepted by sales are good measures. Conversations “resolved” without anyone is a warning sign.
- Can your team change it? If adjusting a question or a routing rule needs a support ticket, it’ll drift away from how your team actually sells.
We go into more of these in questions to ask before buying a chat tool.
Where we stand
We build this kind of software, so take our view with that in mind. Serviceform does conversational forms, AI agents, live chat, booking and lead routing, and the design principle behind it is the one in this piece: let software handle the intake and the scheduling, and get the lead to a person with its context intact. Our real estate product, Uprealer, is built the same way, with routing to the right agent or office and human handover as core features rather than add-ons.
That’s not a claim that every conversation needs a human. Plenty of visitors want a quick answer and nothing more, and that’s fine. The claim is narrower: when a conversation is worth a salesperson’s time, the automation’s job is to get it there faster and better briefed.
The point of it all
Your salespeople are probably your best conversion asset, and also your most expensive and least scalable one. That makes it tempting to automate around them. Resist it. Automate the parts of the job they shouldn’t be doing anyway, measure the tools on how many good conversations they create, and judge success by whether your team walks into each call knowing more than they used to.