Performance Max 2.0: the new era of AI-driven Google Ads campaigns

Performance Max 2.0 is the new generation of Google Ads AI-driven campaigns, released progressively from 2025 and now standard in 2026. It fixes many of the problems v1 brought (lack of transparency, poor controllability, Search cannibalisation) and introduces features finally making it usable for serious projects.

In this guide we explain what really changes, how to configure it to not burn budget and when it makes sense vs classic Search. It's the same framework we apply in projects managed by +Click Ads. For the basic Google Ads guide, see Google Ads for small business.

#Performance Max 2.0: what really changes vs v1

To understand PMax 2.0 you need to know what didn't work in v1.

PMax v1 limits (2021-2024)

• Total black box: you didn't know where the budget went (Search? Display? YouTube?), nor which channel generated conversions.
• Search cannibalisation: PMax "stole" conversions from branded Search campaigns inflating its own numbers.
• Wild Display: 60-70% of budget ended up on low-quality Display (partner sites, game apps).
• Few controls: no keyword exclude, no placement exclude, rigid asset groups.
• Hard to scale: without visibility you didn't know what to optimise.

The key novelties of PMax 2.0

• Channel-level reporting: see how much budget goes to Search, Display, YouTube, Maps, Discover and what conversions each generates.
• Exclude keywords: you can exclude specific keywords (e.g. competitor brands, irrelevant searches).
• Account-level negative keywords: centralised management of negative keywords for the whole account.
• More granular asset groups: up to 50 asset groups per campaign (was 5 in v1) to target better.
• Brand exclusions: protects branded Search campaign cannibalisation.
• Customer acquisition focus: new option to optimise toward new customers (excluding existing ones).
• Signal upload V2: richer first-party feeds (predicted lifetime value, value segments, custom lookalikes).

#Channel-level reporting: the most awaited feature

The most requested feature by advertisers for 3 years. Finally you see where the budget goes and what ROI each channel generates.

What you see in the new reporting

• Impressions per channel: Search, Display, YouTube, Maps, Discover separated.
• Clicks and CTR per channel: understand where the ad performs.
• Conversions per channel: last-click + data-driven attribution separated for each channel.
• Cost per channel: budget spent on Search vs Display vs YouTube.
• CPA and ROAS per channel: the data enabling real optimisations.

Actions to take from channel reporting

Real examples of data-based decisions: "65% of budget goes to Display but generates only 20% of conversions and 3x higher CPA. I reduce the Display share." "YouTube generates awareness but few conversions: I keep it but measure view-through conversions to attribute value." "Search generates low CPA and stable conversions: I protected branded keywords with standard Search campaigns, PMax no longer cannibalises them."

#Asset groups and AI-driven creatives

Asset groups are the creative heart of PMax. PMax 2.0 made them more granular and powerful.

How to structure asset groups

In v1 you put all assets (titles, descriptions, images, videos) in a single asset group and Google combined them. In v2 you have 50 asset groups, each with a specific theme/audience/product. Operational rule: one asset group for each business "segment" (product category, main audience, geo target).

Optimal composition per asset group

• 5-15 titles (30 characters max each): variations of the main message.
• 5 long titles (90 characters max).
• 5 descriptions (90 characters max).
• 5-20 images (multiple resolutions: 1200x1200, 1200x628, 1200x1500).
• 1-5 videos (6-60 seconds duration, 9:16 format for Reels/YouTube Shorts, 16:9 for standard YouTube).
• Multiple brand logo (1:1 and 4:1).
• Specific call to action for the segment ("Learn more", "Buy now", "Book").
• Final URL and UTM tracking.

AI-generated assets: use with caution

PMax 2.0 proposes automatically generated assets: titles derived from the site, AI images from prompt. They're a useful starting point but rarely the best. Use AI for the first draft, then replace with curated assets reflecting the brand. For AI video production, see the dedicated guide on Sora 2 and Veo 3 for brand marketing.

#Signal feeding: how to "guide" the AI without constraining it

Signals are the main way to tell Google "these are your typical customers". The more good signals you feed, the better the optimisation.

The 5 signal types in PMax 2.0

1. Customer list: emails/phone numbers of real customers, uploaded as Customer Match. PMax looks for similar users.
2. First-party conversion data: past conversion events with value (order values, products purchased).
3. Affinity audience: Google predefined audiences based on interests (e.g. "fashion enthusiasts", "frequent travellers").
4. In-market audience: users actively searching for your category (e.g. "in market for new cars").
5. Custom audience: audiences created on visited URLs, used apps, searches made. The most powerful for specific businesses.

Signal quality: what makes them powerful

Signals aren't equal. A customer list of 1,000 recent buyers (last 90 days) is 10x more powerful than a customer list of 10,000 contacts 2 years old. PMax 2.0 gives more weight to quality than quantity. The rule: fresh, segmented, valuable signals. Not "all contacts ever".

#When to use PMax 2.0 (and when to prefer Search)

PMax 2.0 isn't universal. Three scenarios where it makes sense, two where classic Search stays better.

When PMax 2.0 makes sense

1. You have 30+ stable conversions/month from a working Search campaign. PMax extends reach with AI on a solid base.
2. You want to scale cross-channel: capture Display, YouTube, Discover too without managing each channel individually.
3. Ecommerce with structured catalogue: PMax with Merchant Center product feed is very effective for ecommerce scaling.
4. You have rich and updated customer list: quality signal feeding makes the difference.
5. You want cross-platform dynamic remarketing: PMax is one of the simplest ways to manage it.

When classic Search stays better

1. Budget under €1,500/month: PMax has insufficient data to optimise, burns budget on low-quality Display.
2. Brand keywords: always manage them with dedicated Search (low CPC, high CTR), PMax would "steal" them with higher CPC.
3. Very vertical B2B sectors: where keywords are very specific and few, Search gives more control and predictable ROAS.
4. New product/service launch: start with Search to validate keywords and message, then add PMax when data exists.

#Budget, target ROAS, bidding strategy

Operational parameters to set up PMax 2.0 without burning budget.

Realistic minimum budget

Below €1,500/month in PMax 2.0 results are very variable. The threshold for predictable performance is €2,500-3,500/month. For serious scaling: €5,000-15,000/month. Above €15,000/month complexity increases and structured professional management is needed.

Bidding strategy

• Maximize Conversions: to start and gather data. Typical initial setting for 14-21 days.
• Maximize Conversion Value: for ecommerce with variable order values.
• Target CPA: when you have stable CPA above 30 conversions/month.
• Target ROAS: for scaling with specific goal (e.g. 4x ROAS). Requires reliable conversion value data.
• Maximize Conversions with target CPA: useful hybrid in scaling phase.

Budget allocation between campaigns

A serious Google Ads account in 2026 typically has: 40-50% budget on Search for high-intent commercial keywords, 30-40% on PMax 2.0 for cross-channel scaling, 10-15% on protected branded Search, 5-10% on test (Shopping, standalone YouTube, Demand Gen). Proportions adjust by sector. For total marketing budget allocation, see marketing budget and ROI allocation.

#The 7 most costly traps in PMax 2.0

1. Launching PMax without dedicated Search for brand keywords: PMax cannibalises branded at much higher CPC.
2. Uploading low-quality assets: PMax uses what it has, bad assets = bad impressions = wasted budget.
3. No product feed (for ecommerce): PMax value for ecommerce comes from Merchant Center feed. Without, half the potential is missing.
4. Mediocre signals: old customer list, conversion data without value, generic audiences. PMax turns out mediocre.
5. Target CPA too low: you set CPA €15 when realistic is €25, PMax doesn't spend the budget and campaigns stagnate.
6. Not excluding irrelevant keywords: without exclude list, PMax intercepts irrelevant searches burning budget.
7. Treating it as "set and forget": needs weekly monitoring, monthly asset optimisation, quarterly signal refresh.

#Measurement and correct attribution

Measuring PMax 2.0 well is critical because it works cross-channel and attributions can mislead.

Recommended attribution model

For PMax use data-driven attribution (DDA) as default. If your conversion volume is low (under 300/month at account level), DDA doesn't work and it's better to use the 30/40/30 position-based model. Last-click systematically underestimates PMax value which works mostly in middle funnel.

Enhanced conversions: mandatory in 2026

In the cookieless 2026 world, Google Enhanced Conversions (analogous to Meta CAPI) is non-negotiable. It sends first-party data (hashed email, name, phone) server-side to improve attribution. Without Enhanced Conversions you lose 30-40% of attributed conversions. Described in the cookieless tracking 2026 guide.

#Real case: PMax 2.0 vs classic Search

To make this concrete, a recent comparison example on an automotive client.

• Setup: automotive client with 25 car models in catalogue, €4,500/month Google Ads budget, expected conversions 50-80/month.
• Phase 1 (first 60 days): classic Search only on brand + model + zone keywords. Result: 62 conversions at €72 CPA.
• Phase 2 (days 61-120): added PMax 2.0 with catalogue feed, customer list signal of buyers last 12 months, asset groups per car category. Budget split 60% Search / 40% PMax. Result: 98 total conversions (Search 58 + PMax 40), average CPA €46.
• Variation: +58% conversions, -36% CPA. PMax 2.0 added volume with CPA below expectations thanks to channel reporting that allowed quickly cutting non-performing Display areas.

Similar results are replicable for cases like F&F Autoservice where we already have 199 leads at €1.07/lead with Meta Ads, integrable with a Google Ads PMax layer to intercept active search. Deeper dive on overall ROI measurement in the how to measure digital marketing ROI guide.

PMax 2.0 is what PMax should have been from the start. Companies that abandoned PMax after v1 should retry in 2026 with this new toolset. The practical difference in controllability is enormous.

— Niccolò Giuseppetti, founder +Click

#Roadmap to introduce PMax 2.0 in an account

1. Week 1: account audit, conversion tracking verification, Enhanced Conversions configured.
2. Week 2: PMax 2.0 setup with initial asset groups (3-5), customer list uploaded, brand exclusions active.
3. Weeks 3-4: learning phase with Maximize Conversions, budget 30% of total, daily channel reporting monitoring.
4. Week 5: first asset optimisations (replace worst performers), addition of 5-10 new more specific asset groups.
5. Weeks 6-8: switch to Target CPA if CPA stable, progressive budget scaling toward 40-50%.
6. Month 3+: continuous optimisation, asset A/B test, monthly signal update, weekly channel reporting review.

KPIs to monitor in PMax 2.0

1. Average campaign CPA (target: in line with Search campaigns of the same account).
2. ROAS per channel (Search/Display/YouTube/Maps/Discover separated).
3. Budget share per channel (target: Display under 40-50% of total).
4. New customer rate (with customer acquisition mode active).
5. Conversion value per product category (for ecommerce).
6. Brand impression share (to ensure branded stays on Search).

#FAQ Performance Max 2.0