Lead Scoring

CRM Funnel Attribution connects every pipeline stage — from Lead to Closed/Won — back to the marketing click that started the journey. But B2B sales cycles run 3 to 12 months. The revenue data arrives quarters after the click. Until then, your reports can only show lead volume and cost per lead.

And CPL hides real ROI. A Google campaign at $40 CPL looks twice as efficient as LinkedIn at $80. But if the Google leads close at 2% with $15K deals and the LinkedIn leads close at 8% with $45K deals, LinkedIn produces 4x more revenue. Your reports just can't show it yet — because the revenue hasn't arrived.

Budget decisions based on incomplete data systematically overinvest in cheap, low-value channels and underinvest in expensive, high-value ones.

The measurement problem doesn't stay in your reports — it flows directly into ad platform algorithms. When you run Meta, Google, or LinkedIn on Maximize Conversions or Target CPA, you're telling the algorithm: find me more leads at the lowest cost. It does exactly that. More cheap leads, not the most valuable ones.

The algorithm has no way to distinguish a $200K enterprise opportunity from a tire-kicker with a free email domain. Every form fill is "one conversion" — equal weight, equal signal. The platform optimizes for volume because that's the only signal you're sending.

Sending real revenue values would fix this — but ad platforms need the value signal immediately, at the moment of conversion. In B2B, the actual deal value arrives 3 to 12 months later. By then, the algorithm has long moved on. The feedback loop is broken by design.

SegmentStream scores every new lead with two ML predictions: how likely it is to close, and what the deal will be worth. The two multiply into a single number — predicted lead value.

A lead with 68% close probability and a $42K expected deal has a predicted value of ~$29K. Another with 5% probability and a $12K deal is worth ~$600. Both arrived today. Both look like "one lead" in your CRM. The score tells you which one to care about.

This is not rule-based scoring. Traditional systems assign static points — +10 for company size, +5 for webinar attendance. These weights are guesses. The ML models learn from your actual closed deals: which patterns predicted revenue, and which looked good on paper but went nowhere.

1. Connect your CRM data

The models train on leads with known outcomes — Closed/Won, Closed/Lost, deal value. This labeled data comes from CRM Funnel Attribution. The same pipeline that tracks Opportunity and Closed/Won conversions provides the ground truth the models learn from.

2. Enrich and clean

Before scoring, SegmentStream enriches every lead with firmographics — company size, industry, seniority — derived from the domain and email. Spam submissions, bot entries, and invalid emails are filtered with a custom-fit LLM. Better input data, more accurate scores.

3. Score instantly

Every new lead gets a predicted value within hours of arriving. The models analyze marketing source, firmographics, behavioral signals, and funnel velocity to produce a probability and a deal value estimate.

4. Finally, measure ROAS and not just CPA

Before scoring, your reports could only show lead count and cost per lead. Every campaign looked like a list of form fills with no way to tell which ones would actually become revenue.

With predicted values, each lead carries a dollar amount attributed to the campaign that produced it. Now you can compare cross-channel campaign performance on predicted pipeline and ROAS — not just volume and CPA. You instantly see which campaigns generate high-quality leads likely to close, and which ones produce form fills that go nowhere.

5. Feed predicted values back into ad platforms

The same predicted values are sent to Google, Meta, and LinkedIn via CAPI as conversion value signals. The algorithm now knows that a $2,100 lead is worth bidding more for than a $20 lead — and optimizes accordingly.

Evaluate campaigns on pipeline value, not lead count

Stop waiting months for revenue data. Predicted lead values let you compare campaign ROI within days of launch. A campaign that produced 5 leads worth $28K each is immediately visible as a better investment than one that produced 50 leads worth $200 each.

Unlock value-based bidding

Without lead values, you are limited to Maximize Conversions and Target CPA — strategies that treat every lead equally. The algorithm finds more form fills at the lowest cost, regardless of downstream value.

With predicted values, you switch to Maximize Conversion Value and Target ROAS. The algorithm bids higher for audiences that produce $2,100 leads and lower for those that produce $20 leads. Same budget, directed toward the highest-value prospects.

Invest in campaigns that drive revenue

Channels that look expensive on a cost-per-lead basis often turn out to be the most efficient on a cost-per-pipeline basis. Lead scoring makes this visible before you need to wait for deals to close — so you stop cutting profitable campaigns and start doubling down on them.