Performance Max for Stone Shops: Feeding It Right
Performance Max is the campaign type stone shops either swear by or swear at, and the difference is almost never the algorithm: it is what the shop feeds it. PMax hunts across Search, Maps, YouTube, Display and Gmail using your assets, your audience signals and your conversion data as its only map: feed it stock photos and raw form-fills and it dutifully finds people who fill forms; feed it real project imagery and scored, qualified conversions and it goes looking for kitchen buyers. The Pennsylvania stone fabricator engagement ran exactly this play: Performance Max launched WITH showroom and project photos, inside a premium strategy that took monthly ad revenue from $12,000 to $52,000. Here is the feeding guide.
What PMax actually is (for a fabricator)
- A multi-channel hunter with your leash: it finds countertop demand you'd never keyword your way to (the YouTube remodel-watcher, the Maps browser), but it optimizes toward whatever you call a conversion: which makes your conversion definition the steering wheel.
- A complement, not a replacement: keep your core Search campaigns (the quality-intent keywords and 200-plus negative discipline of the Pennsylvania rebuild) running alongside: PMax extends reach; Search protects the high-intent core you already understand.
- An asset amplifier: it assembles YOUR photos, videos and copy into thousands of combinations. Weak assets don't get hidden; they get distributed.
Feed one: assets that prove the premium
- Real project photography only: finished kitchens in real light, waterfall and edge details, the showroom: the exact asset class the Pennsylvania campaigns ran. One stock image in the asset group tells the machine (and every viewer) the wrong story.
- Video is disproportionately rewarded: PMax leans heavily on YouTube inventory, and shops with fabrication reels (the saw, the polish, the reveal: your existing reels engine footage) get placements photo-only accounts never see. Even two or three 15-second cuts change the campaign's reach.
- Copy that filters while it attracts: craftsmanship-forward headlines ('Custom fabrication, template to install') do the same self-selection work as premium ad copy everywhere: price-shopper language in a PMax asset group recruits price shoppers at machine speed.
- Refresh on a cadence: asset fatigue applies here too: feed each quarter's best new projects in, retire the weakest, and let the asset report tell you which images earn delivery.
Feed two: audience signals that describe your buyer
PMax takes audience SIGNALS (suggestions, not fences): give it the Pennsylvania profile: higher-income homeowners engaged with renovation and luxury-home content, your past-customer list, your website's quote-page visitors: and the machine starts its search in the right neighbourhood. Add your service-area geography tightly (the Faro rule: Metro Vancouver homeowners, not the province), and remember signals steer the start, conversions steer the journey: which is why feed three matters most.
Feed three: conversions worth optimizing toward (the big one)
- Never optimize to raw form-fills: the machine will find you infinite $300-vanity inquiries. Define conversions as the actions that precede real jobs: quote requests with project details, booked template consultations, qualified calls past a duration threshold.
- Feed quality back: lead scoring (the Pennsylvania instrument: quality measured 3.1/10 to 8.2/10) becomes PMax fuel when scores flow back as conversion values or offline conversions: closed-job leads worth more than tire-kicks teaches the algorithm your actual economics. This single loop is the difference between PMax finding leads and PMax finding customers.
- Protect the brand lane: add your brand terms as exclusions where offered so PMax earns new demand instead of billing you for people already searching your name.
Running it: the operating rules
- Give it budget patience and data volume: PMax learns from conversions: starve it and it thrashes. Launch it after the conversion plumbing works, not before.
- Judge it on scored quality and cost per QUALIFIED lead, side by side with Search: not on volume. Expect it to find different demand: watch what the search-terms and placement insights show it hunting, and tighten signals when it drifts toward bargain territory.
- Keep the negative discipline alive at account level: the DIY/cheapest/prefab wall from the Pennsylvania and Faro playbooks still applies wherever the platform honors it: and where it doesn't, your copy and conversion definitions have to do that filtering instead.
- Review monthly like a channel: asset performance, audience insights, quality trend: the same one-page discipline as every other lane.
The verdict
Performance Max rewards exactly the assets a disciplined fabricator marketing system already produces: real photography, reel footage, scored conversions, premium copy: and punishes the shops that bolt it onto an empty pantry. Build the feeding system first and PMax becomes what it was in the Pennsylvania engagement: one more channel inside a premium strategy whose scoreboard read close rate 4.8% to 18% and average job $3,200 to $5,800. The full case study shows the context, and Nova builds the whole feeding system: assets, signals, scored conversions: for stone shops, audit first, free.