Conversion value strategy: what each conversion is worth (and how to make Google chase it)

Without values, every conversion weighs the same and Google chases the easiest one. With values, Google chases money. This lesson is the method for setting them: in ecommerce it is almost automatic; in lead generation it is where the competitive advantage sits, because almost nobody does it properly.

The four value models

Lead generation: from the form to expected value

A lead's expected value is P(close) × average customer value. With a 20% close rate and a customer worth €480 in margin, a lead is worth €96. But not every lead is the same: one who asked for a quote with a phone number and a company name closes at 40%; one who downloaded a PDF, at 2%. Assigning different values by lead type is the first step:

And the second step: upload the real value once you know it (qualified lead → a higher value; sale → the amount), with offline conversions. Google learns from both.

The 50% rule for navigation conversions

L1 actions (show phone number, email click, contact page) are useful signals but they are not customers. If they add up to more than half the account's total conversion value, Smart Bidding learns to bring in browsers. Rule: the aggregate value of L1 actions never exceeds 50% of the total value; ideally, a good deal less. You get there by giving them small values or leaving them as secondary.

Ecommerce: revenue, margin and returns

• Revenue excluding VAT: sending it with VAT inflates ROAS by 21% and shifts your break-even target.
• Margin as the value (where the platform allows it): Google optimises towards profit; the target ROAS becomes a target POAS (module 5).
• Returns and cancellations: conversion adjustments (subtract or cancel) by order id; without them, the measured value overstates reality by 10-40% depending on the sector.
• Customer value versus the order: for repeat-purchase businesses, the first order can be worth more than its amount (new customers). With customer data, Google allows value bidding for new customers.

Moving to value-based bidding

With values in place, the strategy can move from target CPA to target ROAS (or maximise conversion value) in lead generation too. Conditions: values that mean something (not all the same), volume (50+ conversions with a value per month), and a target ROAS calculated from the expected value (if the average lead is worth €96 and you accept paying €48, the target ROAS is 2). The transition follows the plan from module 2: a bridge, a target taken from what is actually happening, four weeks.

Mistakes that make the value lie

• Values invented with no close rate behind them ("I'll put 100 on everything").
• Fixed values never reviewed when prices or close rates change.
• A dynamic value that sends 0, or the price of a single product in a multi-item order.
• Mixing currencies, or VAT included in one form and not in another.
• Uploading the sale value without subtracting returns.
• L1 actions with high values "to get more conversions".

💡 Ninja trick: the Lead Rating score is an expected value calculated per lead: behaviour on the site (A), positive actions (B), the ranking of the source campaign (C) → 0-100, uploaded to Google as the value of the raw conversion that same day. Afterwards, qualified and sale correct that value with the reality from the CRM. It is this lesson's "expected value" model, automated and recalibrated with every sale.

What you should remember

• Four models: fixed, dynamic, margin, expected value; lead generation lives at the two extremes.
• Value by lead type = P(close) × customer value; corrected later with offline data.
• L1 < 50% of total value, and preferably much less.
• Ecommerce: excluding VAT, with returns, margin if you can.
• Bid by value once your values mean something and you have volume.