Performance Max was long a black box: put budget in, hope Google finds the right combinations. From mid-2026 that shifts — native asset experiments allow structured tests inside one campaign.
For lead generation in real estate, finance and B2B services it matters because creatives and landing pages often decide conversion rate — not max CPC alone.
What is newly testable
Google expanded PMax with experiment types including:
- Asset groups head-to-head
- Single assets (headlines, images, video) added and measured deliberately
- Seasonal vs evergreen creatives in controlled runs
That does not replace clean offline tracking — but it cuts blind flying on copy and images.
Setup I recommend
- One campaign per offer/vertical — not everything in one pot
- Asset groups by intent (brand, generic, light retargeting)
- At least 14 days runtime before changes — PMax needs learning phase
- Conversion action = qualified enquiry, not form click alone
Tracking as prerequisite
Without Consent Mode v2 and clean ad_storage (see 15 June 2026 cutover) experiments distort. Fix measurement first, then test creatives.
Bing & LinkedIn as control
PMax tests show Google-internal winners. I often add Bing Ads or LinkedIn with the same messages — to see if a creative works cross-platform or only wins inside Google’s mix.
Conclusion
PMax in 2026 becomes more controllable — disciplined experiments save budget and teach faster which message brings enquiries.
How I can help
Channel planning, asset structure, GA4/BigQuery reporting and ongoing optimisation — I use PMax where leads are measurable, not where clicks look good.