How to score leads from 99acres and MagicBricks
Portal leads are not low quality so much as undifferentiated. Scoring separates the enquiries worth a call in five minutes from the ones worth a message and nothing more.
Score portal leads on three dimensions: intent signals such as which listing was viewed and whether a phone number was revealed, fit between the stated requirement and your inventory, and responsiveness to the first contact. Route the highest scores to an immediate call and everything else to a lighter cadence.
Why portal leads feel low quality
Ask any agent about 99acres or MagicBricks leads and the answer is that most of them are worthless. That is true in the sense that a large majority will not transact, and misleading in the sense that it describes almost every source of enquiry in almost every industry. The problem is not that portal leads are bad; it is that they arrive undifferentiated.
A portal delivers a serious buyer with financing arranged and a viewing booked for Saturday in exactly the same format as someone idly browsing at midnight — the same fields, the same notification, the same place in the queue. An agent working them in arrival order spends the same effort on both, which means the serious buyer waits while somebody calls the browser.
Scoring is simply the decision to stop treating them identically. It does not make bad leads good. It changes where the first hour of the day goes, and in a business where speed to first contact is one of the strongest predictors of conversion, that is most of the available gain.
- Low conversion is normal, not a defect. The issue is that leads arrive undifferentiated.
- Arrival order is the worst possible priority. It sorts by chance rather than by likelihood.
- Scoring reallocates the first hour. Which is where most of the conversion difference lives.
A simple lead scoring model
Three dimensions, each scored out of ten. Anything more elaborate will not survive contact with a working sales team.
Intent signals
Intent is what the lead did, and it is the most predictive of the three. Revealing a phone number to see contact details is a stronger signal than saving a listing. Enquiring about a specific unit beats a general area enquiry. Viewing several listings in the same project or price band indicates a search that has narrowed. Submitting a written message rather than clicking a default enquiry button indicates effort.
Timing carries information too, though less than agents assume. Enquiries during working hours convert somewhat better than those at two in the morning, but late-night enquiries from people who work long hours are common enough that this should be a minor weighting rather than a filter.
Budget and requirement fit
Fit is whether you can actually serve this person. A stated budget within the range of your available inventory, a configuration you have, and a location you operate in are the components. A lead wanting a two-bedroom flat in a micro-market where you hold nothing is not a bad lead; it is not your lead.
Be honest about the budget gap. A buyer whose stated budget is twenty per cent below your cheapest available unit will occasionally stretch, and that occasional stretch is why agents keep working these. But it is a lower-probability path than an in-range buyer, and it should score accordingly rather than being worked as though it were the same.
Responsiveness
The third dimension only becomes available after first contact, and it is the most informative of all. Did they answer the call? Did they reply to the message, and how quickly? Did they engage with a substantive question, or give one-word answers? Did they agree to a viewing, and did they attend?
This is where most of your score should end up living after the first twenty-four hours. Pre-contact scoring is inference from thin evidence; post-contact scoring is observation of actual behaviour, and behaviour beats inference. A lead that looked mediocre on intent and fit but answered immediately and asked three specific questions is a better prospect than one that scored well and has ignored four messages.
A scoring sheet you can start with
Start here, then recalibrate against your own closed deals after a quarter. The weights that matter are the ones your own data supports.
| Signal | Points | Why |
|---|---|---|
| Revealed contact details on a specific listing | +3 | Deliberate action with intent to contact |
| Wrote a message rather than a default enquiry | +2 | Effort indicates genuine interest |
| Viewed three or more listings in one project or band | +2 | Search has narrowed to a decision set |
| Stated budget within your available range | +3 | You can actually serve this requirement |
| Configuration and location match your inventory | +2 | Fit, not just affordability |
| Answered the first call | +4 | The strongest single post-contact signal |
| Replied within an hour to a message | +3 | Active in their search right now |
| Agreed to a site visit | +5 | The only signal that reliably predicts a transaction |
| Budget more than 20% below your cheapest unit | −3 | Low probability, not zero |
| No response to three contacts across five days | −4 | Move to nurture rather than continuing to call |
Scoring 99acres vs MagicBricks leads
Both are large Indian property portals with broadly similar mechanics, and the differences between them are less about quality than about mix. The composition of enquiries — rental against sale, price band, which cities and micro-markets are strongest, how many enquiries come from browsing rather than searching — varies between them and varies by city and segment within each.
That means portal-level generalisations are usually wrong for any specific agency. An agent selling premium apartments in one city and an agent letting mid-market flats in another will have genuinely different experiences of the same two portals, and neither experience generalises.
The right approach is to score by source and measure the outcome. Track conversion to site visit and to transaction by portal, by city and by price band, over at least a quarter. If one portal converts meaningfully better in your segment, weight its leads accordingly in the score. Do not carry an opinion formed in one market into another.
- Differences are about mix, not quality. Composition varies by city, segment and price band.
- Portal-level generalisations rarely transfer. Another agent’s experience is not evidence about yours.
- Measure conversion by source yourself. At least a quarter of data before adjusting weights.
Routing and prioritizing by score
A score is only useful if it changes what happens next. Three bands are enough. High scores get called immediately, within minutes, by your strongest closer. Medium scores get called the same day and enter a structured follow-up sequence. Low scores get an automated message and a light nurture cadence, with a call only if they respond.
Route by capability as well as by priority. Your best salesperson should be receiving the highest-scoring leads, which sounds obvious and is frequently not what happens — most agencies distribute leads round-robin for fairness, which optimises for team morale rather than for conversion. If you do distribute evenly, at least ensure the highest scores are not sitting in the queue of whoever is on leave.
Re-score continuously. A lead that scored low and then replied to a message with a specific question has just told you something that outweighs everything you inferred before contact. The score should move, and the routing should move with it.
- Three bands, three treatments. Immediate call, same-day call, automated nurture.
- Match the best closer to the best leads. Round-robin optimises for fairness, not conversion.
- Re-score on every interaction. Behaviour after contact outranks inference before it.
Measuring whether your scores work
A scoring model is a hypothesis and needs testing. The test is straightforward: compare score bands against outcomes over a meaningful period and check that higher bands actually convert better. If your high band converts at the same rate as your medium band, the model is not discriminating and the weights need revisiting.
Measure to site visit as well as to transaction. Site visits happen far more often than transactions, so they give you a usable signal in weeks rather than quarters, and site visit is the strongest leading indicator of a deal in residential property.
Watch for the trap in this: if you only ever call high-scoring leads, you will never learn whether the low band would have converted, and the model will confirm itself regardless of whether it is right. Keep a small random sample of low-scoring leads in the full-contact treatment specifically to check that you are not systematically discarding good prospects.
- Check that bands actually separate. If high and medium convert alike, the weights are wrong.
- Measure to site visit for a faster signal. Transactions take too long to steer by.
- Keep a control sample from the low band. Otherwise the model confirms itself whether or not it works.