VrikshaSignal AI
A technical explainer for developer marketing and sales leaders. The model identifies, scores, and ranks people in your micro-market who are statistically most likely to buy your project before they fill a single portal form.
The premise
A portal lead tells you one thing: the buyer raised their hand. It does not tell you whether they can afford your ticket size, whether your location fits their life, or whether you are the tenth call they will receive today.
Propensity modeling inverts the question. Instead of waiting for a hand to go up, we ask which households in your micro-market look most like people who bought a similar home in this configuration and ticket band in the last 24 months. The answer is a probability, calibrated on Indian residential conversion outcomes.
The four signal pillars
Where a person lives, works, sends children to school, and already has social gravity is the strongest predictor of where they will buy next.
Home buying is usually triggered by a life event. We map whether the household is in a purchase window that matches the project.
A buyer who fits the place and life stage still needs the EMI capacity and asset position to transact at your ticket size.
The fourth pillar measures stability and timing, so the sales team can separate qualified buyers from near-term buyers.
Scoring engine
Every candidate prospect receives a propensity score from 0 to 100. The production model is a gradient-boosted ensemble trained on Indian residential outcomes such as site visits, expressions of interest, and bookings collected across pilots and partner CRMs over the last 24 months.
Scores are calibrated by sub-market and ticket band, then translated into operational buckets for the sales floor. A Rs. 1 Cr buyer in Whitefield is not the same signal pattern as a Rs. 4 Cr buyer in BKC, so every project gets its own calibration profile.
Top-decile prospects against your project. Financially ready, in a triggering life stage, and matched to the location.
Right buyer profile, but missing one near-term trigger. These prospects convert with nurture rather than a cold close.
Profile-matched prospects that are not ready yet. They stay available for future batches and re-scoring as signals evolve.
Production trust
For every prospect, the platform surfaces the signals that pushed them into Hot or Warm. That powers the telecaller conversation starter, such as school commute fit, recent promotion, and loan eligibility at your ticket size.
Residential patterns shift across cycles, interest rate regimes, and sub-market dynamics. Feature importances shift with them, so the model is monitored and recalibrated quarterly.
Signals come from a multi-source network of professional databases, public registries, employment context, and enrichment APIs. Vriksha does not scrape private records or access financial accounts, and the platform is aligned with India's DPDP Act 2023.