How much is a lead worth? Build your value model
The value model is the heart of value-based bidding: a small set of rules that turns each lead's attributes into a euro amount. Done right, it gives the ad algorithm a clear, differentiated signal. Here's how to build one that works — without a data team.
Start from expected value
Anchor on the same idea every time: expected value = probability of closing × deal size. Even rough numbers beat a flat value. If SQLs from enterprise accounts close at 25% for €8,000, that segment is worth ~€2,000 a lead; SMB inbound at 8% × €1,200 is ~€96. That 20× gap is exactly what you want the platform to see.
Pick 3–5 fields that move value
You don't need every attribute — you need the few that separate good leads from the rest. The usual suspects: company size, industry, lifecycle stage, source, job role, deal amount. Resist the urge to add ten fields; a sharp model with four strong signals beats a noisy one with twenty.
Three rule types cover almost everything
- Tiers — for numbers. Company size 1–50 → €0, 51–500 → +€200, 500+ → +€500.
- Mappings — for categories. Industry: Fintech → ×1.7, Manufacturing → ×1.0, Other → ×0.8.
- Flags — for yes/no. “Booked a demo” → +€300.
Rules combine into one value, typically a base amount plus the euro rules, times the multipliers. Add a floor and a cap so a single weird record can't send €0 or €50,000.
Aim for spread, then sanity-check
A good model produces a wide distribution — some leads at €40, some at €2,000 — not a tight cluster around the average. Once you have outcomes (closed deals), compare predicted value to realized revenue: your best-predicted leads should genuinely bring in more. If they don't, a field is mis-weighted.
Ship in shadow, then go live
Run the model in shadow mode first: values are computed and logged but nothing leaves the platform. Check the spread and the segment breakdown, then publish and switch to live. Models are versioned, so you can compare a new draft against the active one and roll back in a click.
Let your data point the way
You don't have to guess which fields matter. The Segments view mines your closed deals to show which attribute combinations drive the most revenue — then you encode those into the model. It's the fastest path from “I think enterprise is better” to “enterprise fintech is worth 3.3× and I've captured it.”