The B2B bidding problem
Smart Bidding is only as smart
as the conversion signal
you give it
Google’s Smart Bidding strategies — Target CPA, Target ROAS, Maximise Conversions — use machine learning to allocate bids across keywords, audiences, and times of day to maximise performance toward a defined conversion goal. They work exceptionally well when the conversion signal accurately reflects commercial value. In B2B, the default conversion signal — a form fill — rarely does.
A B2B form fill includes demo requests from well-qualified enterprise prospects, content downloads from students, pricing inquiries from competitors, and job applications misfiled to the wrong form. Optimising a Smart Bidding strategy toward ‘all form fills’ teaches the algorithm to generate maximum form fills at minimum cost — which produces high volumes of exactly the low-quality conversions you don’t want. An account optimising toward all form fills will consistently outspend an account optimising toward pipeline-qualified leads on irrelevant clicks, because that’s exactly what it’s been trained to do.
The solution is not to avoid Smart Bidding — it genuinely outperforms manual bidding when properly configured. The solution is to give it the right signal: filtered conversion actions that distinguish demo bookings from job applications, offline conversion imports that feed closed-won revenue data back to the algorithm, and pipeline-qualified lead values that allow Target ROAS bidding to optimise toward revenue rather than volume.
01You’re using Maximise Conversions or Target CPA bidding with a conversion action that includes all form fills without filtering for pipeline-relevant submissions
02Your cost per lead has been decreasing over time but pipeline quality has also been declining — the algorithm has learned to generate cheap conversions that don’t convert to pipeline
03You have no offline conversion imports configured — meaning Google’s algorithm has never seen what happens to your leads after they fill in a form, and is optimising toward the nearest observable signal regardless of its commercial value
04You switched to Smart Bidding before accumulating 50+ conversions per month, and performance deteriorated — the algorithm had insufficient data to learn and was making random bid decisions
05You’re using the same bidding strategy across all campaign types — brand, competitor, and generic — despite the fact that each has fundamentally different conversion economics and should be managed independently