The 12 Lead Scoring Signals That Actually Predict Sales in India

Quick answer

Twelve signals consistently predict close-rate for Indian SMB inbound. The top five — source trust, response speed, intent words, customer language, and historical pattern — do 80% of the work. The remaining seven (message length, time-of-day, prior touch, source velocity, geo match, channel cost, repeat signal) tune the long tail. This post lists all twelve with rule weights you can copy.

The full list, ranked by predictive power

1. Source trust — weight 30%

The channel the lead came in through. IndiaMART verified = trustworthy. Scraped contact list = junk. Set per-source weights based on your close-rate history.

Default weights:

• IndiaMART verified: +30
• WhatsApp from paid Meta ad: +25
• Organic Instagram DM: +15
• JustDial enquiry: +10
• Scraped list: −5

2. Response speed — weight 20%

Time from your first message to their first reply. Indian buyers who reply within 60 seconds close at 3× the rate of those who reply in 30+ minutes.

• Replied < 60 sec: +25
• Replied < 5 min: +15
• Replied < 30 min: +5
• Replied > 30 min: 0
• No reply within 24 hours: −10

3. Intent words — weight 15%

Specific words signal purchase readiness:

• "Price", "quote", "cost", "rate": +20
• "Available", "in stock", "delivery", "today": +20
• "Free", "discount", "trial only": −5

These work across English, Hindi, and Hinglish ("kitne ka hai", "abhi chahiye", "available hai" all signal high intent).

4. Customer language — weight 8%

Match buyer's language to a rep who can respond fluently. Hinglish/Hindi lead routed to an English-only rep converts ~3 percentage points lower.

• Language detected matches available rep: +5
• Mismatch: 0
• Mismatch with no available rep: −10

5. Historical pattern — weight 12%

Once you have 200+ closed leads, train a small classifier on your own data. Which combinations of source + message length + time-of-day actually close for your business?

This signal grows in weight over time. Cap at 15% to avoid overfitting to one bad month.

6. Message length — weight 4%

Real questions are usually 25+ characters. "Info" or "?" alone is low signal.

• Message > 50 chars: +10
• Message 25–50 chars: +5
• Message < 25 chars: 0
• Single-character message: −5

7. Time-of-day — weight 3%

Indian buyers messaging at 9–11 PM are often researching for the next morning's decision. Higher intent than 2 PM browsers.

• 9 PM – 11 PM: +5
• 7 AM – 10 AM: +5
• 11 PM – 7 AM: −5 (often automated spam)

8. Prior touch history — weight 4%

Has this number messaged you before? If yes, weight up — repeat interest signals serious intent.

• Repeat contact within 90 days: +10
• First-time contact: 0
• Marked as "lost" last time: 0 (re-engagement opportunity)

9. Source velocity — weight 2%

How many leads is this source sending per hour? If a source is suddenly spiking, it might be running a low-quality campaign (downweight) or running a high-quality one (you'll see in close-rate). Track and adjust.

10. Geographic match — weight 3%

If you only deliver in 5 cities, a Bhopal lead for a Bengaluru-only product is Cold regardless of intent words. Filter early.

• In service area: 0 (neutral; expected)
• Out of service area: −15
• Adjacent area (delivery possible): −5

11. Channel cost — weight 2%

Higher-cost channels (paid Meta CPC > ₹50) attract more deliberate buyers. Lower-cost channels (organic Instagram) attract more browsers.

• Paid ad source, high CPM: +5
• Organic source: 0

12. Product-mention specificity — weight 2%

A buyer who names the specific product/SKU is further along than one asking generically. Run light entity extraction on the message text.

• Named specific SKU/model: +10
• Named product category: +5
• Generic "your products": 0

Putting it all together

Pariq's scoring engine runs all 12 in parallel for every incoming lead. The total score lands in one of three buckets:

• Hot (50+ points) — call within 30 min
• Warm (20–49) — follow up within 24 hr
• Cold (< 20) — automated nurture only

The bucket thresholds shift dynamically based on your team's capacity — if you have one rep available, the threshold for Hot rises so they only see truly urgent leads.

What you can do this week

1. Pick the top 5 signals (source, speed, intent, language, history).
2. Set static weights using the defaults above.
3. After 30 days of data, see which signal is driving the most score variance. If it's not predictive of close-rate, reweight it.
4. Add signals 6–12 only after you have stable Hot-bucket precision > 35%.

Or skip the spreadsheets — Pariq runs all twelve out of the box →