Google AI Max for Shopping is live in the US and the UK, but it has not yet launched in the Czech Republic.
This article is not about what you need to deploy today. Instead, it shows you product feed changes you can make right now. Success with this new format will depend on your offer, your e-commerce store's service, and your data quality. This investment will pay off in your campaigns both now and in the future.
What AI Max for Shopping is
Google focuses on three main pillars for this new campaign type:
- Final URL Expansion. The algorithm redirects people from a specific product to another page on your site if it determines they are more likely to buy there. This could be a category, a blog, or a product bundle. This feature already exists in Performance Max campaigns.
- Automatically generated text creatives. Ad copy is generated automatically from the feed and landing page content. This goes beyond simply pulling product names from Google Merchant Center. This is a major change for Shopping ads.
- Conversion targeting without explicit audiences. The algorithm picks the audience itself. Your job is to provide a strong conversion signal and a high-quality feed.
All three components share a common denominator: success depends on the content of your feed and landing page, not on keywords or ad text as in traditional search campaigns.
The shared lever for PMax and AI Max
If your PMax campaign runs on a feed with thin data, the algorithm is left with poor choices. When a product has a generic name like Jacket M-501, L, Blue and no conversational attributes, the system is groping in the dark. It either targets blindly on the wrong queries or shifts budget to cheaper placements on YouTube and the Display Network just to get some clicks. Either way, you're burning money for nothing. Before you launch AI Max, it's smart to review the top 10 reasons for disapproval in Merchant Center and fix the common mistakes.
AI Max makes these requirements even stricter. Without a specific product name, Final URL Expansion cannot work. The system has no data to use for matching. Without structured attributes, the automatically generated ad copy will be painfully generic. If parameters like material or product_highlight are missing, the algorithm can't identify your ideal customer.
Demo: generic vs structured product name
Imagine a specific product: a leather shoulder handbag.
Weak product name in the primary feed:
Handbag KH-204 brown
Strong product name with conversational attributes:
Picard Sonja leather shoulder handbag, A4 format, 38 × 28 × 11 cm, brown
How does that play out in PMax (today) on the query "leather shoulder handbag for laptop"?
- Weak name: The algorithm knows nothing about the material, the cut, or the dimensions. The match is poor. The product either doesn't show at all, or it appears for the wrong query.
- Strong name: The system sees a match on most of the words. It has a clear reason to show the product.
How does that play out in AI Max (tomorrow) on the same query?
- Weak name: Final URL Expansion has too few signals. The automated copy generator can only produce a generic phrase like "Beautiful handbag for every occasion."
- Strong name: The system can send people straight to the shoulder handbags category. Additionally, the generated copy can include a real benefit: "Fits A4 documents."
Editing a single product takes only a few minutes. Based on experience in the US and the UK, product names restructured this way see a single- to double-digit percent lift in CTR. Google has not published official numbers yet.
3 things to do with the feed now
1. Product names for Final URL Expansion
Basic rule: The first 70 characters of the product name must contain the brand, the type, the key spec, and the main dimension.
| Weak product name | Strong product name |
|---|---|
Herman Miller Aeron | Herman Miller Aeron ergonomic office chair, black, size B |
Adidas UltraBoost 23 | Adidas UltraBoost 23 men's road running shoes, blue, size 42 |
La Roche-Posay Toleriane 50ml | La Roche-Posay Toleriane hydrating day cream for sensitive skin, 50 ml |
Without this structure, the system will not have enough information to redirect correctly.
2. Conversational and supporting attributes
In Merchant Center, Google officially labels six fields as conversational attributes: question_and_answer, document_link, related_product, item_group_title, variant_option, and popularity_rank. I cover these in detail in the article AI Overviews and conversational attributes.
To prepare for AI Max, populate the supporting fields too. The algorithm reads them in the same context:
material: Put values like "leather" or "100% cotton" in its dedicated field. Do not leave them only in the description (description).product_highlight: Use 3 to 5 key benefits per product.age_groupandgender: Use specific values (adult,kids,male,female).size: Send a clean, structured value. Do not leave it only in the product name.
3. Photo quality
Experience from US testing shows AI Max places a heavy emphasis on visual quality. It simply filters out products with low-quality photos. What's the bare minimum?
- A resolution of at least 800 × 800 px, but 1,200 × 1,200 px is better for Retina displays.
- A clean white or neutral background.
- Clear angles, no watermarks or text baked into the image.
If you see an image-quality warning in Merchant Center, that's a problem. Today it means a lower position in PMax, and tomorrow, it could mean disqualification from AI Max.
2 things to fine-tune outside the feed
The feed is the main lever, but the algorithm pulls data from other places too. All these data sources must be aligned. This is a commonly overlooked part of preparing for AI Max.
4. The landing page must speak the same language as the feed
AI pulls content for automatically generated ad copy from your H1 headings, meta descriptions, and other site content. For example: in the feed, you have the name Picard Sonja leather shoulder handbag. But the main heading (H1) on the online store reads generically Summer collection sale -40%.
The system can turn that into an awkward combination of copy. This can cause your CTR to plummet. The main heading on the site and the name in the feed should carry the same core message. Banners and seasonal promos belong on category pages, not on product pages.
Before launch, review your top 50 products. Verify that the feed name and the H1 on the site roughly match. It's a half-hour of work that can save you a significant amount of budget.
5. The safety brake: account-level negative keywords
The more freedom the algorithms get, the more important proactive search query cleanup becomes. Account-level negative keywords become essential. The process is simple:
- Regularly review the search query reports for your existing campaigns.
- Identify where the campaign is leaking budget on irrelevant queries. These are typically searches for warehouses, contact information, manuals, or free items.
- Add them to the account-level negative keyword list across the entire Google Ads account.
A list of 300 to 500 negatives acts as a reliable shield. The algorithm simply cannot work around it. Feed preparation is step one; cleaning up negatives is step two. Without the second step, AI Max will start burning budget on queries with no conversion intent. Google won't refund that money.
Before you change anything: a 15-minute DIY audit
Before making major feed edits, establish your baseline. A simple checklist will help. In 15 minutes, you can determine if you have issues to fix. You'll also see if the problem lies elsewhere, maybe in your offers, campaign structure, or attribution. Without a baseline, you won't know if your changes had any impact.
The PDF is 6 pages. It covers a GMC Diagnostics check, product-name quality, and an attribute audit. At the end, you'll find a scoring table.
If the checklist scores below 8 points, your data has significant room for improvement. If it's above 13 points, the tuning will be more fine-grained, and you can expect a lift in the single-digit percent range.
We will test 50 products free
Doing the audit yourself is fast, but you need to know exactly where the problem is. I'll show you the specific fixes for 50 of your products.
Status in the Czech Republic: when to expect general availability
Google has not announced an official launch date for AI Max in the Czech Republic yet. Based on earlier releases, we can expect a gradual rollout. The best estimate for the Czech Republic is the second half of 2026, but this is just speculation. In the meantime, I recommend you:
- Keep an eye on the Czech Google Ads Help Center and its changelog.
- Follow my blog :)
- If you have access to beta programs through a Google account manager, ask.
What agencies in the US and UK are doing now
Three main patterns are emerging in the US PPC community:
- They treat feed enrichment as a priority. Conversational attributes and image audits are not a one-off task. It's an ongoing process.
- They have a PMax-to-AI-Max transition plan ready. As soon as Google flips the switch, they execute a clear checklist. They review creatives, conversion actions, and clean up exclusions.
- E-commerce stores that delayed cleaning their data are now playing catch-up. In the first months after AI Max launches, the system learns and splits the budget very aggressively. Without a clean feed, it shifts money to only the top 10% of products. The rest of the catalog gets no visibility.
The lesson is clear. Preparing now is cheap. Playing catch-up later will be expensive. AI Max for Shopping is the first wave, followed by fully autonomous agents in Agentic Commerce that will push the entire buying process out of the browser and through APIs.
Summary
- AI Max for Shopping is not generally available here yet, but check the status regularly.
- The main lever is feed quality. This holds true for PMax today and AI Max tomorrow.
- Focus on 5 things: optimized product names for Final URL Expansion, conversational attributes, photo quality, site-to-feed alignment, and negative keywords.
- Final URL Expansion calls for caution. Without URL exclusions and a clean site structure, the campaign will send budget to pages like your Contact page.
- The best estimate for a Czech launch is the second half of 2026. Prepare now.
- The signals from the US and UK are clear: those who delayed enriching their data are now struggling to catch up, while those who started early are reaping the rewards.
If you manage PMax for an e-commerce store with more than 500 products, a feed review is a great investment. It will pay off now and again when AI Max launches. The underlying principle is the same for both.
