Google Performance Max in 2026: the account structure that scales

Performance Max is to Google Ads what Advantage+ Shopping is to Meta. It took five years to roll out, ate most of the campaign types it replaced, and by 2026 is the default for almost any Google account targeting conversions. Most accounts running PMax are running it wrong — same way most ASC accounts were running ASC wrong in 2023.

Here’s the 2026 PMax setup that actually scales, based on what we run for clients across SaaS, e-commerce, and lead gen.

What Performance Max actually is

A single campaign type that runs your ads across all Google inventory simultaneously: Search, Display, YouTube, Discover, Gmail, and Maps. Google’s algorithm decides where to spend, which creative variant to show, and which audience signal to lean on — all in real time, per impression.

The selling point: simpler than running six separate campaigns. The catch: you give up most of the controls you used to have. Negative keywords, individual placement bids, manual audience targeting — gone or severely limited. The bargain is “trust the algorithm and give it good inputs.”

In 2026, with five years of model maturity behind it, the bargain is usually worth taking. But only if you set it up right.

The 2026 account structure

For most accounts, the right PMax setup is:

1. One or two PMax campaigns max

For e-commerce: typically one campaign with the full catalog. For lead gen / SaaS: one campaign per major product or per major audience segment (B2B SMB vs enterprise, for example). Don’t fragment further.

The accounts we’ve seen with eight PMax campaigns are starving each one of conversion volume. PMax needs at least 30+ conversions per month per campaign to optimize well. Below that, the model is guessing.

2. Asset groups as the segmentation layer

Within a PMax campaign, asset groups are how you express “this set of creative + signals goes with this audience theme.” Build asset groups around:

• Audience theme (“First-time buyers,” “High-LTV repeat,” “Cart abandoners”)
• Product category (for e-com — “Outerwear,” “Footwear,” “Accessories”)
• Use case (for SaaS — “Solo founders,” “Growing teams,” “Enterprise IT”)

3-5 asset groups per campaign is the sweet spot. Each gets dedicated creative, dedicated audience signals, and dedicated final URLs. They share the campaign-level budget and bidding.

3. Audience signals (suggestions, not constraints)

This is the most misunderstood part of PMax. Audience signals tell Google “here’s what your ideal converter looks like.” Google uses them as a starting point but isn’t bound by them. Strong signals to provide:

• First-party customer match lists (your CRM-uploaded converters)
• In-market segments relevant to your category
• Custom segments built from URL behavior (people who searched for your category)
• Detailed demographics if relevant

Audience signals are inputs to the model. They are not “we will only show ads to these people.” If you treat them as targeting, you’ll be confused why the model “ignored” your audience.

4. Negative brand list

PMax by default will run your ads on competitor brand searches and your own branded searches — both usually wasteful. Apply a campaign-level negative keyword list for:

• Your own brand (you don’t need PMax bidding against your free organic and your dedicated brand-search campaign)
• Major competitor brands (unless deliberately conquesting)
• Job-related searches (“[brand] careers”)
• Customer service / login terms

You can apply negatives only via your Google account rep through 2025-26 (Google deprecated the self-serve UI temporarily during the campaign-type migration). Worth the email.

What kills Performance Max in practice

1. Bidding for the wrong conversion

PMax optimizes for whatever conversion event you tell it. Pick wrong and the model trains toward the wrong outcome. Common mistakes:

• Optimizing for “All conversions” when you have multiple conversion events of different value (newsletter signup + demo request + purchase all summed). PMax will chase whichever is cheapest, usually the newsletter signups. Net result: your CPA looks great, your pipeline doesn’t move.
• Optimizing for top-of-funnel events (PageView, Engagement) when you have real conversions to optimize for. Tighten the bid to actual revenue events.
• No value tracking for e-commerce. Send purchase value with the event so PMax can optimize ROAS rather than just conversions.

Set up enhanced conversions and ensure conversion values flow through correctly. This is the equivalent of CAPI for Meta — table stakes in 2026.

2. Creative asset starvation

PMax wants 5+ headlines, 5+ long headlines, 5+ descriptions, 1-5 logos, 5-15 images in multiple aspect ratios, 1-5 videos. Most accounts ship 2 headlines and a stock image and wonder why performance lags.

The Google “ad strength” indicator is roughly accurate. Aim for Excellent on every asset group. The accounts running Good or Average asset strength leave 20-40% performance on the table.

3. Final URL expansion

By default, PMax can send users to any page on your site, not just the URLs you specified. Sometimes this is great (PMax finds a deep-linked product page that converts well). Often it’s bad (PMax sends users to your About page).

For most accounts, restrict final URL expansion to known good URLs. Set the URL contains rules so PMax only sends to /products/*, /services/*, etc.

4. Not separating brand and PMax

If you run a dedicated brand search campaign AND PMax, PMax will eat your brand traffic at higher CPCs and report it as PMax conversions. Your brand campaign’s reported volume drops, your PMax CPA looks great, and you can’t tell what’s really driving incremental conversions.

Two fixes:

• Add your brand terms as negatives to PMax (preferred)
• Use Google’s brand exclusion list feature in PMax (newer, works at the campaign level)

What to actually measure

• Account-level CPA / ROAS vs your target. PMax’s whole selling point is account-level optimization. Judge there.
• Incremental conversions via Google’s Conversion Lift studies or your own holdout tests. PMax sometimes claims credit for conversions that would have happened anyway through other channels.
• Brand vs non-brand split in PMax’s “Insights” tab. If 60% of your PMax conversions come from brand queries, you’re paying for traffic that should be coming through your dedicated brand campaign at lower CPCs.
• Asset group performance in the Insights tab. Audience signals report tells you which signal the algorithm leaned on most.

What’s overrated

• Hyper-granular asset groups. 10+ asset groups per campaign starves each one of signal. Stay at 3-5.
• Constant creative refreshes. PMax tolerates creative for longer than Meta does — the placements are more diverse (some inventory is forgiving). 4-6 week creative lifecycles are fine.
• PMax for awareness. It’s a conversion-optimized format. Brand campaigns should run separately on Discovery (now Demand Gen) or YouTube directly.

The honest framing

Performance Max is Google saying “trust us.” For accounts that give it clean conversion data, strong creative assets, smart audience signals, and proper brand exclusions, it consistently outperforms hand-built campaigns. For accounts that just enable it and hope, it underperforms.

The work moved from campaign configuration to signal quality and creative production. Same shift as Meta. Same winners and losers. The accounts scaling on Google in 2026 aren’t the ones tweaking bids — they’re the ones running tight PMax setups with industrial creative and pristine conversion tracking.