Performance Max Best Practices for 2026: A Practical Playbook
Real-world Performance Max best practices that actually move ROAS — asset groups, audience signals, conversion goals, and the data structure Google needs to learn fast.

Performance Max sounds simple in Google's marketing materials. Launch a campaign, hand over creative, let the AI optimise. In reality, most accounts we audit are wasting 30 to 50 percent of their Performance Max budget on poorly structured campaigns.
The good news: a handful of Performance Max best practices separate the campaigns that scale from the ones that flatline. This playbook covers them in the order we apply them on client accounts.
If you are still trying to figure out whether to use Performance Max at all, the short answer is yes — for e-commerce, lead gen with a clear conversion event, and local store visits. Skip it if your conversion volume is below 30 per month.
What Performance Max actually is
Performance Max is Google's fully automated campaign type. One campaign serves ads across Search, Shopping, Display, YouTube, Gmail, Discover, and Maps. You provide creative assets, audience signals, and a conversion goal. Google's AI decides what shows where.
The trade-off is control. You cannot see search term reports the way you can in Search campaigns. You cannot bid by placement. You give up granular control in exchange for the algorithm's reach.
That trade-off only pays off if the algorithm has clean signals to learn from. Get the inputs right and Performance Max scales. Feed it junk and it burns budget.
The three inputs Google's AI actually uses
Most Performance Max best practices boil down to optimising these three inputs.
Conversion data quality
Google's AI bids using your conversion data. If your tracking is wrong, the algorithm chases the wrong signal. Garbage in, garbage out.
Before launching Performance Max, verify your conversions in GA4 and Google Ads. Send the same conversion events from both. Use Enhanced Conversions to feed first-party data — hashed email, phone, address. This is non-negotiable.
For e-commerce, send the full purchase event with revenue. For lead gen, send qualified leads, not raw form fills, ideally with an offline conversion import for closed deals.
Asset quality and variety
Performance Max needs creative variety. The algorithm tests combinations of headlines, descriptions, images, and videos to find what converts.
Plan for at least 5 headlines, 5 descriptions, 4 long headlines, 15 images, and 5 videos per asset group. If you skip the video, Google generates one automatically — and the auto-generated version usually underperforms.
We cover the visual side of this in detail in our conversion rate optimization guide. The same hierarchy principles apply to Performance Max creative.
Audience signals
Audience signals tell Google who you think will convert. The AI uses them as a starting point and expands from there.
Build custom segments based on keywords competitors target. Add your customer list. Layer in interests and life events. The more specific the signal, the faster the algorithm learns. Generic "in-market for digital marketing" signals waste 4 to 8 weeks of learning.
Twelve Performance Max best practices that move ROAS
These are the specific tactics that make the difference between mediocre and excellent Performance Max performance.
1. Separate asset groups by margin or product category
A single asset group means Google optimises for the average. Separate by margin tier or category and the algorithm can push budget toward the winners.
For an e-commerce client, splitting one asset group into three by gross margin lifted ROAS from 2.1x to 5.8x in 12 weeks. The high-margin products got more spend, the low-margin ones less.
This was the structure behind our Marseille cosmetics e-commerce case study.
2. Add search themes to guide intent
Search themes act like keywords for Performance Max. Add 5 to 25 per asset group with phrases your buyers actually type — not industry jargon.
For a B2B SaaS client, switching from broad themes like "fleet software" to specific ones like "GPS tracking for 50 vans" cut cost per MQL by 67 percent. Specificity wins.
3. Exclude brand traffic from Performance Max
Brand searches convert at high rates and skew Performance Max numbers upward. If you let Performance Max compete for your branded queries, it cannibalises Search campaigns and inflates ROAS reporting.
Add your brand terms as account-level negative keywords. Run brand traffic through a dedicated Search campaign where you can see CPC and search term data.
4. Feed shopping data through a clean Merchant Center
For Shopping ads inside Performance Max, your product feed quality is the entire game. Optimise titles with the keyword first, brand last. Use real product images, not lifestyle shots. Keep stock and price feeds updated every 24 hours minimum.
A feed with 95+ percent product approvals beats a feed with 70 percent every time. Fix attribute errors before launch.
5. Use customer match lists as audience signals
Upload your customer email list. The AI uses it to find lookalikes who behave like your best buyers.
Segment by lifetime value if possible — your top 20 percent customers send a stronger signal than the full database. Refresh monthly.
6. Set realistic target ROAS or CPA
Start with Maximise Conversion Value (no target) for the first 2 to 3 weeks. Let the algorithm learn what your good days look like.
Then layer in a target ROAS at 80 percent of your baseline. Tighten it monthly. Setting an aggressive target on day one starves the campaign of impressions and stalls learning.
7. Build different asset groups for prospecting vs retargeting
Performance Max does both prospecting and retargeting from one campaign. By default it leans heavily on retargeting because it converts faster.
Split into two asset groups: one with prospecting signals (customer match lookalikes, in-market segments) and one for retargeting (site visitors, cart abandoners). This forces budget into new customer acquisition.
8. Use the asset group preview tool weekly
In the Performance Max interface, the asset group section shows which combinations are running. Pause asset groups labelled "Low" if they have spent more than €500 with no conversions.
Review weekly for the first 8 weeks. After that, monthly is enough.
9. Run Performance Max alongside Search, not instead of it
A common mistake is shifting all paid budget into Performance Max. You lose visibility, search term control, and the ability to bid hard on high-intent terms.
Keep brand and high-intent commercial keywords in Search campaigns. Use Performance Max for breadth, Search for depth. Our Google Ads service page explains how the two work together.
10. Watch the page-feed for lead-gen campaigns
If you do not sell products through a feed, upload a page feed listing your landing pages with descriptive labels. Performance Max uses this to send the right traffic to the right page.
Without a page feed, the algorithm may send every click to your homepage. Conversion rate suffers.
11. Add account-wide negative keywords
You cannot add negatives at the campaign level in Performance Max from the UI. You can at the account level. Use this to block irrelevant searches that drain budget.
You can also request your Google rep to enable campaign-level negatives — it is available for most managed accounts now.
12. Refresh creative every 6 to 8 weeks
Asset fatigue applies here too. Swap in fresh images, headlines, and at least one new video every 6 to 8 weeks. Keep the structure stable but rotate the inputs.
We see CTR drops of 10 to 25 percent on stale Performance Max campaigns. A planned refresh keeps performance steady.
The Performance Max metrics that actually matter
Forget impression share and click numbers. These are the metrics that tell you whether Performance Max is working.
- Conversion volume vs target: are you hitting the volume needed for the algorithm to learn?
- ROAS or CPA: is the campaign profitable at your tolerance?
- New customer acquisition cost: filter for new customers only — the algorithm hides this metric by default
- Asset group performance: which groups are pulling weight, which are dead weight
- Search themes contribution: which themes are driving conversions
If you cannot answer the new-customer cost question, you cannot scale Performance Max profitably.
Common Performance Max mistakes to avoid
Some habits look harmless but quietly burn budget.
- Launching with weak conversion tracking. Fix tracking first, launch second.
- Single asset group for the entire catalogue. Always segment by margin or category.
- Using stock photos in image assets. Real product or service photography wins.
- Setting tROAS on day one. Let the algorithm learn before adding constraints.
- Forgetting brand exclusions. Performance Max will eat your brand traffic if you let it.
- Skipping video assets. Auto-generated videos drag CTR down.
- Mixing prospecting and retargeting signals in one asset group.
A 60-day Performance Max launch plan
If you are launching Performance Max from scratch, follow this sequence.
Week 1 — Audit conversion tracking, set up Enhanced Conversions, clean Merchant Center feed if applicable.
Week 2 — Build 3 to 5 asset groups segmented by margin or category. Upload customer match list. Write 5 headlines, 5 descriptions, gather 15 images, produce 1 to 2 videos.
Week 3 — Launch with Maximise Conversion Value, no target. Set daily budget to at least 5x your target CPA so the algorithm has room to learn.
Weeks 4 to 6 — Review weekly. Pause underperforming asset groups after €500 with no conversions. Adjust search themes based on what is driving sales.
Weeks 7 to 8 — Add a tROAS target at 80 percent of baseline. Refresh creative on the strongest asset group.
By day 60, expect ROAS to stabilise within 10 percent of your steady-state. Anything wildly above or below means tracking or signals need fixing.
Frequently asked questions
When should I use Performance Max vs Search campaigns?
Use Performance Max when you have at least 30 conversions per month, clean tracking, and creative variety. Use Search for branded terms, high-intent commercial keywords, and accounts with low conversion volume.
How long until Performance Max learns?
The official "learning period" is 6 weeks but real stability comes at 8 to 12 weeks with clean signals. Less if you start with strong customer match data and conversions.
Can I see search terms in Performance Max?
Partially. The Insights tab shows search categories but not raw queries. To get full visibility, run a parallel Search campaign on your most important keywords and read the search terms there.
What budget do I need for Performance Max to work?
A rough rule: at least 50x your target CPA per month. So if your target CPA is €50, plan for €2,500 minimum monthly spend. Below that, the algorithm does not collect enough signal.
Should I run multiple Performance Max campaigns?
Yes, if your product mix is varied. Split by region, language, or product category. Avoid more than 5 Performance Max campaigns per account — at that point you fragment learning.
Does Performance Max work for B2B?
Yes, but with caveats. Use lead-gen conversions with offline import for closed deals. Add high-quality first-party data through customer match. Expect a longer learning period than e-commerce — closer to 3 months than 6 weeks.
Get an audit of your Performance Max account
If you are running Performance Max and not sure whether it is working, we audit accounts free of charge. We pull asset group performance, signal quality, and projected ROAS improvement in 48 hours.
Book a free 30-minute audit. We screen-share, look at your account live, and give you a prioritised action plan.
Or explore our Google Ads service for the full system we deploy on accounts spending €5K to €50K monthly.
Want these strategies applied to your business?
30 minutes of free audit with concrete recommendations tailored to your business.
Read next
Google Shopping Campaigns Optimization: The 2026 Playbook for E-commerce
A practical guide to optimising Google Shopping campaigns — feed quality, product titles, bidding, campaign structure, and the silent leaks that hurt e-commerce ROAS.
Google Ads Attribution Models Explained: Which One Should You Use?
A plain-English guide to Google Ads attribution models — last-click, first-click, linear, time-decay, position-based, and data-driven. When each one helps and when it lies.
The Google Ads Audit Checklist: 47 Points We Check on Every Account
A comprehensive Google Ads audit checklist covering tracking, structure, keywords, bidding, creative, landing pages, and budget. The exact 47-point list we run on client accounts.