Why you're asking this now
Until 2024, online product discovery had a single front door for most buyers: Google. You searched, you saw a Shopping carousel, you visited a few sites, you decided. In 18 months, the behaviour split. More and more buyers open ChatGPT or Perplexity before Google and ask the question directly: what to buy for a specific need. The answer comes back as 3–5 recommended products with links, without seeing a carousel or a classic result.
For an online store that means two parallel systems with different rules. The owners who understand the distinction now have time to prepare their catalogue for AI without pressure. Those who don't, find out in 6 months that sales have dropped without any clear explanation in Google Analytics.
Google Shopping in short: what you already have
Google Shopping is the product carousel at the top of Google results when someone actively searches. You pay per click. You appear because you have a feed in Merchant Center, Shopping campaigns in Google Ads and a budget. The buyer sees you, clicks, lands on your product page and decides. It's a 15-year-old system: predictable, measurable.
If you run Performance Max, half your budget still ends up in Shopping, only Google decides where the ad appears. The underlying mechanism stays the same: catalogue in Merchant Center, bid on products, paid click, measured sale.
AI Shopping: a new category with rules you can't buy
AI Shopping is the moment a buyer asks an AI assistant what product to take and gets a short list of recommendations with direct links. The assistants that matter right now in Romania:
- ChatGPT Shopping: built into ChatGPT answers for purchase questions, product comparisons and category recommendations.
- Perplexity: answers questions with products cited from multiple sources, including direct comparisons.
- Google AI Overviews: the AI carousel that appears above classic Google results for complex purchase questions.
- Gemini: integrated answers in Google Workspace and on Android, including product recommendations.
The fundamental difference: you can't buy the slot. You appear only if your product data is structured the way AI reads it, and only if the platform decides you're a relevant option for the question. There's no boost button and no auction.
Direct comparison
| Criterion | Google Shopping | AI Shopping |
|---|---|---|
| How the customer arrives | Active search on Google | Asking an AI assistant |
| When you appear | When you pay and have a valid feed | When AI selects you from the catalogue |
| Cost of visibility | Cost per click | Zero direct, preparation cost |
| Signals that matter | Bid, feed quality, historical CTR | Structured data, schema, brand authority |
| Speed to results | Hours | Weeks to months |
| Reporting visibility | Google Ads + GA4 | Manual test, no official report |
| Stability | Predictable | Variable, AI learns monthly |
Why you can't choose between them
Buyers don't all behave the same way. Some still start with Google. Others open ChatGPT directly. Many do both on different days, for the same decision. If you're only on Google Shopping, you lose the second type. If you're only on AI, you lose the first. Market estimates put roughly 30% of decisions in categories with research demand (electronics, supplements, cosmetics, appliances, tools, B2B) moving through AI by the end of 2027 in Romania.
For a store that lives off e-commerce, that's a strategic variable, not a tactical one: do you want to be present at both points where the decision happens, or do you bet on one and hope the other doesn't grow faster than you can prepare for.
The most common mistake
"I pay for Google Shopping, so I'm covered on AI as well." Not true. Being in the Shopping carousel means Google can show you when someone searches. It doesn't mean ChatGPT, Perplexity or Google AI Overviews will recommend you when someone asks. They are separate systems with separate signals.
Quick check, anyone can run it in 5 minutes: open ChatGPT and ask what the best products in your category are for a typical customer. Use a natural phrasing, the way a real buyer would ask the question. If your brand doesn't appear in the answer, you are invisible on that channel. Repeat on Perplexity and Google. Three tests, three places where buyers are already asking.
What needs to change for AI Shopping
Four areas, in order of impact:
- Product titles written for natural questions. In Google Shopping, the optimal title is dense with keywords. In AI Shopping, the title needs to be intelligible for a model that receives a question in natural language. The difference: an AI-friendly title contains brand, model, primary attribute and use case, phrased as a product description, not as a list of terms.
- Complete attributes in the feed. Material, size, colour, age, gender, GTIN, MPN, Google category. AI assistants parse these attributes when matching the product to the question. Missing attributes mean missing matches.
- Schema markup on the site. Product, Offer, AggregateRating, Brand. This is what an AI assistant sees when it visits your page. Without correct schema, you are structurally invisible, no matter how good the customer-facing copy is.
- Monthly check across AI platforms. Visibility in AI is not a one-off setting. Platforms change their rules month by month, the competition moves, your catalogue grows. You check monthly that you still appear and adjust.
The cost of waiting
AI systems learn from what they already surface. Every month a competitor sits in the answers and you don't, their recommendation becomes the default for your category. At some point, even doing everything perfectly, recovery takes significantly longer than prevention.
The loss doesn't show up in any report. The buyer never even reaches Google to search for you, so they don't appear in GA4, in Search Console or in Google Ads. You see sales stagnate or drop slightly without knowing why, because every other channel looks the same.
Visibility in AI Shopping isn't bought with budget. It's earned with product data written correctly and with monthly maintenance discipline.
Where to start
If you have under 50 active products, AI optimisation doesn't justify a monthly retainer. Rewrite the titles, complete the basic attributes and come back in 6 months when the catalogue grows. If you have 100+ products and direct competitors already appear in AI answers for your category, now is the time. Start with an audit that tells you exactly where the sale is being lost and which fixes go into the first month. Then execute, in-house or with a partner who works directly with your tech team.
The question worth asking yourself this week is simple: if 30% of your future buyers ask AI instead of Google, are you there when AI answers? If the answer isn't clear, you have a simple test and a decision to make.
AI Shopping isn't bought with budget. It's earned with correct data and maintenance discipline.
At DAFE Digital we handle the entire technical layer: title rewrites, attributes, schema markup, monthly checks on ChatGPT, Perplexity and Google AI. You see the result: your products appear in recommendations.
Adela Mincea
Marketing Economist · Fondatoare DAFE Digital · Formator ANC
Adela is a Marketing Economist with over 10 years of paid media experience across Europe, the US and Asia. She founded DAFE Digital for one reason: serious Romanian businesses deserve the same paid media expertise companies get in any other market. That's what DAFE Digital does.
Tags
