Performance Max Campaigns: The Complete Guide for 2026

Performance Max campaigns have been around long enough now that the "is PMax worth it?" debate is mostly settled. The answer is yes — but only if you structure them properly. The accounts we audit at Workflow AI Advisors that are struggling with PMax almost always have the same set of problems: bloated asset groups, weak audience signals, no search theme layering, and zero segmentation logic. The accounts performing well have done the opposite of all of that.

This guide is written for paid media practitioners who are already running Google Ads and want a serious, practical breakdown of how to handle Performance Max in 2026 — including the newer controls Google has released, what smart bidding actually needs to work, and where PMax consistently cannibalises or complements your other campaign types.

What's Actually Changed in Performance Max for 2026

Google has steadily handed back more control to advertisers, which signals they're listening to the feedback from agencies and in-house teams. The significant updates you need to understand heading into 2026 include:

• Search themes at the campaign level — now supporting up to 25 themes per asset group, these function as intent signals to guide where Google serves your ads. They're not keywords, but they're the closest thing to keyword control you have inside PMax.
• Brand exclusions at account level — a long-overdue improvement that lets you stop PMax from eating your branded search volume and inflating ROAS artificially.
• Placement-level reporting — you can now see where your Display and YouTube inventory is showing at the URL/channel level, which means you can actually make exclusion decisions based on real data.
• Asset group-level reporting improvements — conversion data is now more granular at the asset group level, making it easier to diagnose which segments of your PMax campaign are actually driving results.
• Customer lifecycle goals — the ability to set separate bidding priorities for new customer acquisition versus returning customers, which is genuinely useful for ecommerce and subscription businesses.

If you haven't revisited your PMax setup since 2024, you're almost certainly leaving performance on the table by not using these controls.

The Asset Group Architecture That Actually Works

The single biggest structural mistake in Performance Max campaigns is creating one asset group and throwing everything into it. Google will find something to serve, but it won't be optimised — it'll be scattered. Here's how we structure asset groups across client accounts:

Segment by Product Category or Service Line First

Each distinct product category or service line should have its own asset group. If you're running an ecommerce account selling homeware and clothing, those are two asset groups minimum. Why? Because the creative assets, the landing pages, and the audience signals are fundamentally different. Lumping them together means the machine is trying to optimise for too many conflicting signals simultaneously.

Then Layer by Audience Intent, Not Just Demographics

Inside each product-focused asset group, consider creating separate asset groups for different stages of intent. A prospecting asset group with broader creative and a TOFU landing page behaves differently from a retargeting-focused asset group with tighter creative and a direct conversion page. You won't always have the spend to support this level of segmentation, but at budgets above £3,000–£5,000/month per segment, it's worth testing.

Landing Page Alignment Is Non-Negotiable

Every asset group must point to a landing page that directly reflects the assets in that group. Google's Quality Score logic applies at the landing page level even inside PMax. If your assets are talking about summer dresses and your landing page is a generic homepage, you're wasting your budget and confusing the algorithm.

Audience Signals: The Most Underused Lever in PMax

Audience signals are not targeting. Google will clarify this every time you ask, and they're technically correct — PMax will go beyond your signals if it thinks it can find conversions. But signals are the fastest way to point the algorithm in the right direction during the learning phase, and they continue to influence serving patterns after that phase is complete.

The signals we apply consistently across client campaigns at our paid media service:

• First-party customer lists — your existing customers, segmented where possible (high-value customers uploaded separately from general customers)
• Website visitor lists — all visitors, plus specific product/category page visitors as separate lists
• Customer Match from CRM data — especially valuable for businesses with longer sales cycles where email is still a primary channel
• In-market segments — relevant to your product category, used to set a baseline intent signal for the algorithm
• Custom intent audiences built from search terms — create custom segments using URLs of competitor sites and category-level keywords

The accounts that see the fastest exits from the learning phase are almost always the ones with strong first-party data uploaded before launch. If you're launching a PMax campaign with no audience signals and no customer lists, you're essentially asking Google to start from zero — and that burns budget inefficiently for the first 2–4 weeks.

Smart Bidding Strategy: What to Use and When

Performance Max campaigns run exclusively on smart bidding. Your choices are Maximise Conversions, Maximise Conversion Value, tCPA, and tROAS. Here's the practical guidance:

Start with Maximise Conversions (if new)

If a campaign has fewer than 30 conversions in the past 30 days, do not apply a tCPA or tROAS target. The algorithm doesn't have enough data to hit a constrained target — it will either underspend or spend erratically. Run on Maximise Conversions until you have consistent conversion volume, then layer in a target.

tROAS is Your End State for Ecommerce

For ecommerce, tROAS is the right long-term bidding strategy — but only when you have reliable conversion value data flowing through. That means properly implemented purchase value tracking, not just purchase event tracking. We've seen accounts running tROAS on campaigns that are only passing a static conversion value of £1 for every transaction. The algorithm is optimising for the wrong thing entirely.

Be Conservative with Your Targets Initially

When you transition from unconstrained to target-based bidding, set your initial tROAS or tCPA at roughly 20–30% more lenient than your actual performance to date. If you're hitting a 4x ROAS on Maximise Conversion Value, start tROAS at 320% rather than 400%. Give the algorithm room to operate before tightening the constraint.

Search Themes: Using Them Without Over-Relying on Them

Search themes, introduced as a proper feature in late 2023 and significantly expanded since, give you influence over what search queries trigger your PMax ads. In 2026, with the 25 search themes per asset group limit, you have meaningful room to work with.

Think of search themes as signals, not as keyword lists. They're best used to:

• Steer the algorithm toward high-intent commercial queries in your category
• Prevent PMax from drifting into tangentially related but irrelevant query territory
• Accelerate learning for new campaigns where the algorithm has no historical data to lean on

What they won't do: guarantee impression share on specific queries, match exact keyword strings, or function as negatives. For proper negative keyword management at the campaign level, you still need to work with your Google rep or submit exclusion requests — though account-level negatives can be applied directly.

PMax and Your Existing Search Campaigns: Managing Cannibalisation

This is one of the most common concerns we hear, and it's legitimate. Performance Max will compete with your branded search campaigns and your high-performing exact match campaigns if you don't manage the overlap. Here's the framework:

• Apply brand exclusions at the account level — this prevents PMax from capturing branded queries and artificially inflating your ROAS figures
• Give your standard search campaigns priority — Google's own documentation states that exact match keywords in standard campaigns take precedence over PMax for the same query. But this only applies when the ad quality and relevance is comparable
• Run your top-performing search campaigns in parallel, not instead of PMax — the best-performing accounts we manage at Workflow AI Advisors typically run both, with clearly delineated roles
• Monitor the search terms report across all campaigns — look for query overlap between PMax and standard search, and adjust accordingly

Creative Assets: Volume, Quality, and Rotation Logic

Google's asset strength indicator is a useful guide but shouldn't be treated as gospel. "Excellent" asset strength doesn't guarantee strong performance, and we've seen "Poor" rated asset groups outperform "Excellent" ones on ROAS. That said, the principles behind asset strength are sound:

• Upload the maximum number of assets across all formats — 15 images, 5 videos, 5 headlines, 5 descriptions
• Include video assets. If you don't provide them, Google will auto-generate videos from your images and copy, and those auto-generated videos are consistently weak
• Write headlines that work standalone, not as a sequence — Google mixes and matches them without your input
• Use a mix of product-specific and benefit-led copy. Don't write five variations of the same message
• Refresh underperforming assets every 60–90 days rather than letting them run stale

For clients where creative production is a bottleneck, this is an area where AI-assisted content workflows can genuinely accelerate output without compromising quality — generating initial asset variations that a human creative then refines and approves.

Reporting and Optimisation Cadence

PMax's black-box reporting reputation is somewhat deserved, but it's improved. Here's what you should be reviewing and at what cadence:

Weekly: Spend pacing, conversion volume, CPA/ROAS versus target, asset group-level performance, search themes report

Bi-weekly: Placement exclusions (check for irrelevant Display and YouTube placements), audience signal performance indicators, landing page conversion rates

Monthly: Asset performance review, search theme refresh, bidding strategy adjustment, cross-campaign cannibalisation check, overall channel ROAS with and without brand

One metric that doesn't get enough attention: new customer rate. Especially for ecommerce businesses, PMax has a tendency to over-index on returning customers because they convert more easily and signal better ROAS to the algorithm. If your new customer acquisition rate is declining while ROAS looks healthy, PMax may be serving largely to people who would have converted through direct or branded search anyway.

The Numbers Behind Well-Run PMax Accounts

Across accounts we manage at Workflow AI Advisors, properly structured Performance Max campaigns consistently deliver measurable improvements versus poorly structured predecessors. We've seen CPA reductions of 31% on average when migrating from legacy Smart Shopping or disorganised PMax setups to properly segmented, signal-rich campaigns. ROAS improvements of 4x or higher are achievable in ecommerce verticals when asset groups are aligned to category pages and first-party data is fully utilised.

These numbers don't come from PMax being magic. They come from the structural work that happens before and during the campaign — the audience list setup, the asset production, the landing page alignment, and the ongoing optimisation discipline.

If you want a deeper look at how this fits into a broader paid media strategy, see how we approach paid media management for growth-focused businesses.

Common PMax Mistakes to Eliminate in 2026

• Not uploading brand exclusions — your ROAS looks better than it is
• Single asset group for your entire product catalogue
• No video assets — letting Google auto-generate is a waste of inventory
• Applying tROAS or tCPA targets before hitting 30+ conversions per month
• No first-party audience signals — cold starts burn budget
• Ignoring placement reporting — Display and YouTube can drain budget on irrelevant inventory
• Treating PMax ROAS in isolation without accounting for branded query cannibalisation
• Not separating high-value customer lists from general customer lists in your signals