Real estate CRM: leads, site visits and follow-ups with AI

Real estate CRM development has one job: make sure no lead is lost between the portal enquiry and the booking, and that management can see, by project, how many bookings the pipeline will produce. Packaged CRMs manage contacts and deals; a builder or broker needs leads tied to projects and units, site visits scheduled and confirmed, follow-ups that happen on time, and channel partners tracked for commission. This article sets out what a property CRM for the Indian market must do, where AI removes real work, and how the build is phased and priced.

Why builders outgrow generic CRMs and spreadsheets

A residential project launches, leads pour in from property portals, digital campaigns, hoardings, walk-ins and channel partners, and the sales team works them from phones and a shared sheet. Within weeks, the same buyer has been called by three people, site visits clash, and nobody knows which campaign actually produced bookings. Generic CRMs help with contacts but do not understand a unit inventory, a payment plan or a channel partner's claim on a lead. Spreadsheets understand nothing.

The decision framework is the one we use for any custom CRM versus off-the-shelf question: the process is the differentiator when the object model does not fit, and for real estate it does not. Our sister product, TheEazy CXM, covers standard pipeline needs; the project-and-inventory layer is where a custom build earns its place.

Real estate lead management: the modules that matter

Lead capture: automatic, de-duplicated, attributed

The first rule is that no lead is typed in. Portal enquiries arrive by email or API and are parsed into the CRM; ad platform leads arrive through their lead forms; WhatsApp enquiries are captured from the business number; walk-ins are registered on a tablet at the site office. Each is de-duplicated by phone number, because the same buyer enquires on three portals. Attribution is kept per enquiry, not just per lead, so marketing can see that a buyer first came from a portal and later from a hoarding. Response time is the metric: the first call within minutes matters more than any later feature, so the assignment rule and the alert to the sales executive are built in phase one.

Site visits: the conversion that matters

For most projects the site visit is the step that predicts a booking. The CRM schedules it, sends confirmation and directions on WhatsApp, reminds the buyer and the site executive, and records the outcome: which units were shown, what the objection was, when to follow up. Visit-to-booking rate per project and per executive becomes the management number, replacing raw lead counts, which say little. The site executive works from a phone app that shows the day's visits and lets them record the outcome in a minute; the pattern is the same one we use for field teams in the dispatch platform and field apps build.

AI follow-ups and auto-logging

This is where AI removes real work. Sales executives make dozens of calls a day and log almost none of them. With call recording and consent, an AI pipeline transcribes each call, summarises it, updates the lead stage and unit interest, and drafts the next WhatsApp message for the executive to approve. The executive taps send or edits; they do not type notes. The same applies to WhatsApp threads and emails. The result is a CRM that is complete without anyone doing data entry, which is the problem CRMs usually die of. We set out the design in auto-logging: AI that keeps your CRM data complete.

Guardrails on AI messages

Follow-ups touch price, availability and possession dates, all of which carry regulatory weight in a RERA-registered project. AI drafts must use approved facts from the inventory and project master, never invent a discount or a date, and be sent by a person or under a policy that limits what can be sent automatically. Messages in Hindi, Kannada or Tamil are drafted from the buyer's own language in the conversation. A WhatsApp agent for a D2C brand shows the conversational side; the property version is more conservative about what it may say.

Channel partners and forecasting

Channel partners bring a large share of leads for many projects and dispute attribution constantly. The CRM registers partners, tags leads to them at capture with a time stamp, resolves claims by a rule the management agrees, and calculates commission on booking. Visible, rule-based attribution with a time stamp ends most disputes before they reach the sales head.

Forecasting closures by project uses stage conversion rates, visit-to-booking rates and available inventory to estimate bookings for the coming month. This is arithmetic on clean data, not machine learning, and it is only possible because auto-logging keeps stages current. Management sees the forecast per project and per tower, and marketing sees which sources convert to bookings rather than enquiries.

A worked example

A builder launching two residential projects in a large Indian city was receiving leads from four portals, two ad platforms, a website and a dozen channel partners, worked by a sales team from a shared sheet. Duplicate calls and missed follow-ups were the daily complaint. Phase one built automatic capture with de-duplication and assignment, the project and unit inventory, and site visit scheduling with a site executive app. Phase two added call and WhatsApp auto-logging with AI-drafted follow-ups approved by executives, and channel partner registration with rule-based attribution. Management moved from lead counts to visit-to-booking rates by source and executive. Follow-ups happened on schedule because the executive's task list was generated from real conversations, not from memory. Booking and payment plans stayed in the builder's existing ERP, integrated for handover.

Team and timeline

A real estate CRM build needs a product lead who understands the sales floor and site office, two engineers, one mobile engineer for the executive app, a designer for the screens the team will use all day, and on the client side a sales head who decides assignment rules and a marketing owner for source integrations. Portal and ad-platform credentials are needed on day one.

The scope and phase plan come from a Sprint Zero at $3,250 / ₹2,00,000, credited to the build. ERP and CRM development starts at $28,000 / ₹18.4L for phase one; AI auto-logging and follow-ups are added as an LLM application module from $21,000 / ₹13.6L. Phase one typically goes live in eight to ten weeks, ideally before a launch. The pricing page lists the Care Plans that cover the integrations after go-live; portal formats change without notice, so monitoring matters.

Before you start: a checklist

• List every lead source and how each delivers leads today
• Prepare the project, tower and unit inventory with the current price sheet
• Decide lead assignment rules: round robin, by project, by source, by language
• Agree the channel partner attribution rule and commission structure
• Confirm call recording consent and the retention policy for recordings
• Define the approved facts AI drafts may use: prices, availability, possession dates
• Identify where booking and payment plans live and how the hand-over works
• Set the first management metric: visit-to-booking by project and source

Questions clients ask

• Can it integrate with the property portals we list on? Yes; leads arrive by email or API depending on the portal, and both are parsed automatically.
• Will executives actually use it? If the app takes under a minute per visit and the AI does the logging, yes. Adoption fails when the CRM asks for typing.
• Does it handle RERA-related disclosures? It keeps approved facts in the project master and restricts AI drafts to them; compliance content itself is the builder's responsibility.
• Can brokers use the same system? Yes, with the inventory module replaced by listings across projects and developers.

Related reading

See AI in CRM: what a copilot should do for sales and service teams, personalised messaging for email, push and WhatsApp and our customer service agents service. For WhatsApp Business messaging rules, the primary source is the Meta developer documentation.

Put the unit inventory at the centre, capture every lead automatically, and let AI do the logging so the team can do the selling.