Latest Update:
AI in lead management: key lessons from our webinar
In our recent webinar on AI and lead management, Homeflow co-founder Jeremy Harford Tapp explored how the industry actually handles enquiries today, where AI already outperforms the human average, and how people and technology work best together. Here are the key ideas.
Estate agents have always known what good lead handling requires: fast, reliable responses, the discipline to ask the right questions every time, and the rigour to extract every commercial opportunity from every conversation. The gap between knowing this and doing it, especially at any scale, is not new. Response times still average over 30 minutes, more than half of all enquiries receive no meaningful follow-up, and even when agents do respond, the consistency rarely survives the pressures of a busy day. These aren't failures of intent, necessarily, they're failures of capacity, and they're precisely where AI starts to become genuinely useful.
One of the most important frameworks for thinking about AI in lead management is the value pyramid. At the top: a landlord with a large portfolio, a vendor with a high-value home in your patch, a buyer who turns out to have a house to sell. At the bottom: a letting enquiry on a lower-yield property with 40 other applicants. Both deserve a response, but the economics of how you resource that response are completely different.
The mistake many agents make is treating the whole pyramid as if it requires the same human attention, and the consequence is that everything gets a mediocre version of your focus rather than the right things getting your full engagement. The role of AI is not to replace good human judgement at the top of that pyramid, but to ensure that the bottom is handled consistently and well, freeing your people to apply that judgement where it actually matters.
Having observed thousands of lead interactions handled by both humans and AI systems, the evidence is clear: well-configured AI does a better job than the average human at the bottom of the value pyramid. Not better than the best human, and not better than a branch manager with deep local knowledge and a long-standing client relationship — but better than what those enquiries actually receive in practice, which is often a slow, generic or incomplete response, or nothing at all.
A well-trained AI responds instantly, never forgets to ask whether the enquirer has a property to sell, measures everything it does, and populates your CRM with structured data rather than leaving the follow-up to someone's memory. For the volume of enquiries that currently falls through the gaps, the difference is significant.
Alan Turing's imitation game, first proposed in 1950, posed a deceptively simple question: can a human, communicating only through text, reliably distinguish between another human and a machine? Turing's insight was that if the answer is no — if the machine's responses are indistinguishable from a person's — then the practical distinction between the two starts to dissolve. It's a test not of what AI is, but of what it can do.
It's a useful lens for estate agency, because the question that matters for lead handling isn't a philosophical one about the nature of intelligence. It's a practical one: is the quality of this interaction good enough to serve the customer well, and can they tell the difference? On text-based chat, the honest answer is that we are already at that threshold in most common scenarios.
Good AI systems can handle nuanced property questions, identify the commercial opportunity within a conversation, adapt their approach based on what the customer tells them, and hand off to a human at exactly the right moment, and most customers, engaging with a well-configured system, won't know or particularly care that they started the conversation with a machine.
The quality of an AI lead handling system depends almost entirely on how it's set up: a poorly configured system is worse than a human, while a well-configured one is consistently better than the human average.
Guardrails work in layers. The innermost layer is the AI's confidence zone: questions it can answer accurately from the information it's been given, covering property details, branch context, pricing and availability. The middle layer covers territory it shouldn't handle directly, such as offer negotiations or commercially sensitive advice, where the right behaviour is to engage the customer warmly and connect them to a human.
The outer layer manages anything outside the scope of a property conversation, redirecting firmly and, if that fails, disengaging. Getting those layers right is the work. The technology is capable, the question is whether the configuration is good enough to deploy it safely and effectively.
The most important thing to understand is that AI in lead management isn't a replacement for human relationships. When AI handles initial qualification, captures the right data, identifies the commercial opportunity and routes the enquiry with full context already loaded, the human who picks it up can skip the administrative conversation and move straight to the relationship one. That's a genuinely different way of working, and it's better for the customer too.
This model only works, though, if the handoff is clean. The system needs to know when to escalate, the humans need to be ready to take over, and the data needs to flow into your CRM in a structured and usable form. Technology, process and people have to work together, none of the three does the job alone.
The practical starting point is the bottom of the pyramid: the enquiries that currently go unhandled, receive a generic automated reply, or wait over 30 minutes for anything at all. That's where AI can have an immediate impact with the lowest risk and the clearest commercial case. As confidence grows, you work your way up. The technology will support that progression, what it won't do is compensate for poor configuration or a team that hasn't been brought along with it.
