Performance Max for bridging: when to trust the black box
Where PMax over-delivered on lead volume - and the three asset-group guardrails we never run without.
Performance Max went from 'experimental side budget' to 'meaningful share of bridging search spend' through 2024. By January 2025 it was sitting at 30-40% of the spend mix on our better-performing accounts. It works. It also fails in spectacular, expensive ways when you let it run without guardrails. This is what we learned about both.
Where PMax over-delivered
On brand and brand-adjacent traffic, PMax was unbeatable - it would harvest cheap clicks from search, display and YouTube that no manual structure could touch at the same cost. For an established broker with even modest brand search volume, it was almost free money.
It was also surprisingly good on long-tail commercial intent terms that we would never have manually targeted because the volume was too low. Aggregated across hundreds of such terms, PMax found real pipeline.
Where it failed
Anywhere the conversion signal was noisy. A bridging account optimising for 'lead form fill' will quickly drown in tyre-kicker leads because PMax will find the easiest possible audiences to make that event fire.
We had one account in mid-2024 where PMax found a way to generate hundreds of cheap leads from a Display placement that was, on inspection, almost entirely accidental clicks. Cost-per-lead looked great. Cost-per-qualified-call was infinity.
The three guardrails
First, optimise on a downstream event - calendar booking confirmed, or broker-marked qualified - never on a lead form fill. PMax will exploit any easy signal you give it.
Second, brand exclusions. Without a brand list uploaded as a negative, PMax will eat your branded search traffic and claim credit for it, making the campaign look better than it is.
Third, asset group discipline. One asset group per real audience theme, with bespoke creative and landing pages. Mixing themes inside an asset group lets PMax misallocate budget and lose the signal you are trying to give it.
PMax in 2024-25 was the right tool for the right job and the wrong tool for the wrong one. The brokers who built guardrails got something close to free distribution on brand-adjacent terms. The brokers who treated it as 'set up and let it run' mostly funded Google. The difference was about ninety minutes of setup work.
Capital Edge publishes one note a month on UK bridging finance, paid acquisition, and AI-led outbound. Written for brokers, by the team running the playbook.