Real Estate Chatbot: Qualifying Property Buyers

Why property leads decay faster than most

A property enquiry has an unusually short life. The buyer is looking at four listings from three agencies on the same evening, and the viewing they book first shapes every comparison afterwards. A lead answered the next morning is competing against a viewing that has already happened.

The second factor is that most enquiries arrive out of hours. People browse property listings after work and at weekends, which is precisely when the agency phone is not answered. On many agency sites the majority of enquiries arrive when nobody is available.

Both factors point the same way: the value is in responding substantively at the moment of interest, and in arriving at the agent's morning queue already sorted by how ready the buyer is.

The four signals a property conversation gives you

1. 1Budget, expressed as a rangeBuyers state this readily, unlike in most B2B contexts, because they know it determines what they will be shown. Ask it as a range rather than a figure, and record it as a range — a buyer who says 'up to 200' will look at 215.
2. 2Rooms and sizeThe most reliable filter and the easiest to ask. It also carries family context that shapes the viewing: two bedrooms plus a study is a different conversation from two bedrooms.
3. 3Location, at the right granularityDistrict rather than address, plus what matters nearby — a school, a metro station, a workplace. A buyer who names a school catchment has told you more about their timeline than any direct question would have.
4. 4Finance, which is the real qualifierCash, approved mortgage, mortgage not yet applied for, or selling an existing property first. These four answers describe four completely different timelines, and they are the single strongest predictor of whether a viewing turns into a transaction.

The listing question that predicts intent

Beyond the four, the conversation itself carries signals that do not need to be asked. They come free with the enquiry and are routinely discarded.

• Which listing the conversation started from, and whether the buyer named several. A buyer comparing three specific properties is further along than one asking what is available.
• Whether they asked about the building rather than the flat — maintenance charges, the neighbours, the year of construction. These are second-stage questions.
• Whether they asked about documents, registration or the purchase process, which usually indicates a decided buyer rather than a browsing one.
• Whether they asked for a viewing time before asking anything else. This is the strongest single signal in property and it should route immediately.
• Repeat visits. Somebody returning to the same listing twice in a week is worth a call even if they gave no contact details the first time.

What the agent needs to receive

A property lead that arrives as a name and a phone number wastes the conversation that produced it. The agent should be able to open the call knowing what to offer.

• The four signals, as structured fields rather than free text, so the queue can be sorted and filtered.
• The listing or listings discussed, linked rather than described.
• A one-sentence summary in the buyer's own terms — 'family moving from a rented two-room flat, needs three rooms near a specific school, mortgage not yet applied for'.
• Preferred viewing times, if the conversation reached that point, which it should whenever the buyer showed viewing intent.
• The channel and time of the enquiry, because a Saturday-evening enquiry answered Monday needs a different opening line than one answered in ten minutes.

Where the bot must stop

• Price negotiation of any kind. The listed price is the bot's ceiling and its floor; anything else is an agent's conversation.
• Anything about legal status, ownership, encumbrances or documentation specifics. These are legally consequential and must come from a person who can be accountable.
• Mortgage advice. The bot may collect whether finance is arranged; it must not advise on products, rates or eligibility.
• Statements about future value, rental yield or the area's prospects. These are predictions, and a prediction made in your name is a claim.
• Confirming that a specific property is still available, unless that comes from a live lookup rather than an indexed page. A viewing booked on a sold property is the most damaging error in this sector.

Metrics that matter in property

1. 1Viewing booking rateConversations that produced a scheduled viewing, divided by qualified enquiries. In property this matters more than lead count, because a viewing is the real unit of progress.
2. 2Show rate on booked viewingsProperty no-show rates are high and largely determined at booking: near-term slots, a named agent and one reminder make most of the difference.
3. 3Time to first contact for out-of-hours enquiriesMeasured separately from daytime. This is where property agencies lose most of their pipeline and where the gap is most easily closed.
4. 4Finance mix in the pipelineWhat share of your qualified leads are cash, approved, unapproved or chain-dependent. A pipeline that is mostly unapproved is a pipeline with a long tail nobody has planned for.

How Vexvon handles property enquiries

The bot answers around the clock from your own material — your site crawled to as many as 4,500 pages, PDFs up to 25 MB, and a product catalogue with variants and images, which is how a listing portfolio is held. Entries are typed into thirteen categories including branch, pricing and campaign, so district-specific and development-specific material can be kept separate rather than blurred.

Availability is the case that should not be answered from an index, and it does not have to be: your own API can be registered as a tool the agent calls mid-conversation, so whether a property is still on the market comes from the system that holds that fact. When the lookup fails, the defined behaviour is to say so rather than to guess.

Extraction runs on the conversation rather than a form: phone numbers are detected through a five-stage process, free-form fields are captured alongside, and a one-sentence summary is written to the record. A buyer-intent filter decides whether an extracted contact is genuinely a lead before one is created, and the agency's own numbers are excluded.

The lead reaches the sales group as a Telegram notification with a Start button; the agent who presses it is assigned, and the record moves through a nine-stage pipeline carrying a priority marker, a close reason and a contact-attempt count, with reminders checked every minute. Because all channels share one customer record, a buyer who wrote on Instagram and later called is one record rather than two.

Frequently asked questions

1. 1What should a real estate chatbot qualify on?Budget as a range, rooms and size, location at district level with what matters nearby, and finance status. Finance is the one most often skipped and the one that most changes the priority of the lead.
2. 2Should the bot quote prices?Only the listed price. Any negotiation, discount or flexibility is an agent's conversation, and a bot that improvises around price creates an expectation the agency has to honour.
3. 3Can the bot confirm availability?Only from a live lookup against the system that holds it. Booking a viewing on a property that has sold is the most damaging error in this sector, and an indexed listing page goes stale within days.
4. 4What should reach the agent?The four signals as structured fields, the listings discussed, a one-sentence summary in the buyer's own words, preferred viewing times and the enquiry's channel and time.
5. 5Which metric matters most?Viewing booking rate rather than lead count, paired with show rate. A viewing is the real unit of progress in property, and no-shows are largely determined at the moment of booking.
6. 6When must a person take over?On price negotiation, anything about legal status or documentation, mortgage advice, and any statement about future value or yield.

Start with last weekend's enquiries

Take the enquiries that arrived between Friday evening and Sunday night, and check two things: how many received a substantive answer before Monday, and how many contained the four signals somewhere in the text without anyone recording them. In most agencies both numbers are a surprise, and together they are the entire business case.