Lead Scoring for D2C: How to Stop Treating Every Contact the Same

The myth: a lead is a lead

Here's a pattern that quietly drains D2C margins. A brand collects contacts from every direction — WhatsApp opt-ins at checkout, abandoned carts, a skin-type quiz, a newsletter box in the footer — dumps them all into one list, and sends everyone the same 15% welcome code and the same festive blast. The person who was thirty seconds from paying full price gets handed a discount they didn't need. The person who signed up six months ago and never returned gets the same nudge as a hot cart. Both messages underperform, and the brand blames the creative.

The fix isn't a better message. It's admitting that these contacts are not the same, and treating them differently. That's all lead scoring is: a way to rank contacts by how likely they are to buy right now, so your best offers and your human attention go where they'll actually pay off.

What lead scoring actually is (and isn't) for D2C

Lead scoring came from B2B, where sales reps score prospects on company size, job title and budget to decide who to call. Ignore that framing — you're not selling enterprise software. In D2C, you score on behaviour: what someone browsed, added, abandoned, bought and engaged with. The output is a number, and the number decides what happens next.

What it isn't: a replacement for RFM segmentation of your existing customers, and not a one-time setup you forget. Think of scoring as the front door — qualifying new and re-engaging contacts by intent — while RFM handles how you treat people you've already sold to. The two work together. Scoring decides who gets your attention today; RFM decides how you keep them.

It's also not a machine-learning project. Some platforms sell "predictive" lead scoring that promises to find intent you can't see. That can help at scale, but you don't need it to start, and a black-box score you can't explain is hard to trust or fix. A simple points model you built yourself has one big advantage: when a band underperforms, you can see exactly which behaviour is miscalibrated and change one number. Start transparent, earn confidence in the approach, then reach for heavier tooling only when the manual model has clearly hit its ceiling.

Two axes: who they are and what they did

A good D2C score blends two kinds of signal.

Fit signals — who they are

Light touch for most brands. Did they land on a high-intent product versus a blog post? Are they in a city you ship to reliably, or a COD-heavy pin code with high return rates? Did the quiz say they're in your core category? Fit signals set a baseline but rarely move the needle alone.

Behaviour signals — what they did

This is where D2C intent lives. Every action a contact takes is evidence, and actions closer to purchase are stronger evidence. An abandoned checkout is a louder signal than an abandoned cart, which is louder than a product view, which is louder than a footer signup. Two more rules make behaviour scoring work: recency (an action today counts more than one from March) and repetition (three visits this week beats one). Intent in D2C decays quickly, so a score that ignores time will keep chasing people who've moved on.

A simple scoring model you can build this week

Start crude. Assign points to five behaviours, then band the totals.

Add up recent behaviour, apply a simple decay (halve points older than, say, 14 days), and you get a live score per contact. Then band them:

Don't agonise over the exact numbers. The point bands matter less than the fact that you're now separating a hot cart from a cold signup — which you weren't before.

Watch a real contact move through it

Picture Riya. She takes your skin quiz on Monday (+5), reads a blog post, and leaves — score 5, cold. On Thursday she comes back, views four products (+15), and adds a serum to her cart but doesn't check out (+25). Her score is now 45 — hot. On Friday morning your hot flow fires: a WhatsApp reminder about the serum in her cart with free shipping, no discount. She buys.

Now rewind two weeks. If Riya does nothing after that quiz, decay halves her 5 points, and she drifts toward zero — correctly, because she showed no real intent. The same contact was cold, then hot, then a customer, and your system spoke to her differently at each stage instead of sending one generic welcome code on day one. That movement is the entire value of scoring: it's a live read on intent, not a label you stamp once.

Turning scores into actions: which flow fires when

A score that doesn't change an action is a vanity metric. Map each band to a specific play:

Hot (40+). Move fast and light on discount. A checkout abandoner usually needs a reminder and maybe free shipping, not 20% off — they were already close. Fire a WhatsApp nudge within an hour, and for high-value carts, consider a quick human message or call. This is exactly where a well-built CRM setup earns its keep, routing hot leads to the right flow automatically.

Warm (15–39). These need a reason and a little patience. A two or three message sequence over a few days — social proof, a bestseller nudge, a modest first-order offer — does more than a single blast.

Cold (under 15). Nurture, don't discount. Value content, category education, the occasional bigger offer to test if they're revivable. If they stay cold across a couple of cycles, stop spending message credits on them and protect your WhatsApp quality rating.

Notice how discount depth follows the score. The deepest codes go to the coldest leads, because that's where you need to overcome the most resistance. Handing a deep code to a hot lead just gives away margin you'd have earned anyway.

Where D2C brands get lead scoring wrong

Three traps come up again and again. The first is scoring everything — twenty signals, elaborate weights, a model nobody can explain. Five signals you understand beat twenty you don't. The second is forgetting decay, so a person who browsed once in summer still shows as "warm" at Diwali and gets treated like live intent. The third, and most expensive, is building the score and never wiring it to different messages — the whole point is that the hot cart and the cold signup now hear different things.

One more, specific to India: don't let a high behaviour score override a bad COD signal. A contact with a hot cart but a pin code that returns half its COD orders isn't a clean win. Nudge them toward prepaid with a small incentive rather than pushing COD you'll pay to ship twice.

A fourth trap is quieter: scoring only new contacts and ignoring re-engagement. A past customer who suddenly views three products and abandons a cart is one of your highest-intent leads, yet many setups leave existing customers out of the scoring model entirely and let RFM handle them on a slow monthly cycle. Feed repeat-buyer behaviour into the same live score, and a returning customer showing fresh intent gets caught in the moment rather than a month later. The reactivation window in D2C is short — often days, not weeks — so speed is the whole game.

Start small: the one-week rollout

Pick the five behaviours above, set the three bands, and tag contacts in whatever WhatsApp or email tool you already use. Build three flows — hot, warm, cold — and route new contacts by band. Run it for two weeks, then look at which band actually converts and adjust the points. You'll usually find one behaviour is a much stronger buy signal than you assumed, and you'll lean into it.

That's the whole discipline: rank by intent, match the message to the rank, and let your best offers chase the people most likely to buy. Every contact stops getting the same email — and your margins stop paying for the ones who were going to buy anyway.

If you want a single number to watch after a month, compare the conversion rate of your hot band against your cold band. In a working model, hot should convert several times better than cold — that gap is proof your score is actually separating intent rather than shuffling names. If the two bands convert about the same, your points are wrong, not your idea: something you're calling "hot" isn't, and it's time to re-weight. Get that gap wide and healthy, and every downstream decision — which offer, which channel, whether a human steps in — gets easier, because you finally know who you're talking to.