Product and Service Schema for B2B in AI Results

For B2B service providers, schema.org/Service is almost always the right choice, not Product. Product markup is built for purchasable items with a price, stock and a checkout button, which rarely fits a service you deliver to order. With Service schema, however, you make your offering machine-readable, so AI systems recognize your services unambiguously instead of having to guess from running text. In this article you will read when to choose which type, which fields really matter, and where the gain for your AI visibility does and does not lie.

What is the difference between Product and Service schema?

Product schema describes a purchasable item, Service schema describes a service you perform. That sounds subtle, but the consequences for B2B are significant.

schema.org/Product is designed around a tangible or digital product someone can buy: a name, a price, stock, delivery time and reviews. Google mainly ties this type to e-commerce, where a visitor can pay for the item directly. If you show no price and no buy action, Product markup loses half of its meaning.

schema.org/Service describes a service that an organization delivers, for example advice, maintenance or implementation. The type leans on other fields: who offers the service, what kind of service it is and where you deliver it. For a consultant, agency or installer, that matches reality far better than a product page with a price tag. As a rule of thumb: if you sell a thing, use Product. If you carry something out for a client, use Service.

Which type do you choose as a B2B service provider?

Choose Service if you deliver a service, and keep Product only for concrete, purchasable items you may sell alongside your services. Most B2B companies sit entirely in the Service camp.

A few practical cases make it concrete:

• Pure service delivery such as consultancy, marketing or legal advice: Service. There is no item to pay for, only an assignment you carry out.
• A productized service with a fixed scope and a fixed rate: still Service, optionally with an offers field that holds the price indication. After all, you deliver a service, not a stock item.
• Software or a license you sell as a defined product: here Product (or a more specific type) can be right, because there is a clearly purchasable item with a price.
• Physical products alongside your services, for example a wholesaler that also installs: use Product for the items and Service for the work around them.

If you are in doubt, ask yourself whether a visitor “buys” something or “has something done”. That answer points you to the right type. Avoid the temptation to mark everything as Product because you read somewhere that Product schema earns rich results. For B2B services that usually does not hold up, as you will read further on.

Which fields make your offering truly machine-readable?

The value lies in a handful of fields that unambiguously state who you are, what you do and for whom. For Service those are mainly provider, serviceType, areaServed and, optionally, offers.

• provider: the organization or company that delivers the service. This ties the service to your brand and is the basis of a recognizable entity.
• serviceType: the kind of service, for example “GEO optimization” or “industrial maintenance”. This places your offering in an understandable category.
• areaServed: the area where you deliver, from a city to a country. For B2B that works regionally or nationally, this prevents confusion about your reach.
• offers: an optional price indication or condition, handy for a productized service with a fixed rate.

A minimal example in JSON-LD, the recommended way to add schema, looks like this:

{
  "@context": "https://schema.org",
  "@type": "Service",
  "name": "GEO optimization for B2B",
  "serviceType": "Generative Engine Optimization",
  "provider": {
    "@type": "Organization",
    "name": "Customer Impact"
  },
  "areaServed": "BE"
}

The golden rule: only mark up what is also visible on the page. A field that does not match your text works against you, because both search engines and AI systems check your markup against the content they see.

Do AI systems read your Product and Service schema?

The honest answer is nuanced: the extent to which AI systems use your markup directly differs per platform and is partly unconfirmed. We therefore do not adopt hard percentages that are firmly substantiated nowhere.

Google has consistently said that structured data helps to understand content, and Google AI Overviews builds on that same index. The effort you put in for classic search engines therefore works there too. With pure language models such as ChatGPT and Perplexity it is different: they work mainly with the visible, rendered text of a page and not necessarily with the invisible markup beneath it. Schema there is at best an extra signal, not a decisive factor.

What schema does do is make your core facts unambiguous. A system that sees in fixed fields who delivers the service and where does not have to distill that from a paragraph. That lowers the chance that your offering is summarized wrongly or attributed to another party. It connects to grounding, the process by which AI determines what is true, and to entity consistency, which determines whether a model keeps your brand separate from a similarly named player. Schema is thus part of GEO, becoming visible in AI answers, and not a standalone trick.

Does Product or Service schema earn a rich result in B2B?

Usually not, and that is important to know before you invest a lot of time in it. Google’s rich results are stricter than what schema.org allows, and Service misses out.

Product rich results, with price, reviews and availability in the search results, are meant for pages where a visitor can buy an item. A typical B2B service page without a checkout button does not qualify. The Service type has no rich result of its own in Google Search, and things like reviews on a Service do not pass the eligibility test, even though every validator approves the markup. So expect no stars or price labels under your service page.

Does that mean schema is pointless for B2B? No. The value merely shifts from a visible rich result to machine readability. You make your offering clear and consistent for systems that process your facts, even when there is no eye-catching search result in return. At Customer Impact we are deliberately honest about this: we do not promise rich results that simply do not exist for your type of page.

How does schema fit into a broader GEO approach for B2B?

Schema is one layer, and almost never the first one to invest in. The order that works in practice: first make sure your content is accurate and clearly structured, then build a recognizable brand that is also mentioned beyond your own site, and then use Product or Service schema to make the whole a bit sharper and more machine-readable.

If you do it the other way around and start with the markup, you polish a page whose content is not yet strong enough to be cited. That is wasted effort. The biggest lever remains content that is crystal clear even without schema, which is exactly the core of content architecture for AI extraction. How to approach these choices specifically for your sector, you can read in our guide on GEO for B2B.

The short summary

For B2B service providers, Service schema almost always fits better than Product, because you deliver a service and not a purchasable item with a price and stock. The fields that really matter are provider, serviceType and areaServed, and they must match the visible text on your page. Rarely expect a rich result, because Product experiences are for e-commerce and Service has no rich result of its own in Google. The gain lies in machine readability: you make your offering unambiguous for AI systems, which raises the chance of a correct mention. See schema as an amplifier of strong content, not as a switch, and steer on the leads that extra AI visibility produces.

Want to know which schema and which steps make the difference on your site for your visibility in AI? We are happy to take a look with honest advice via our generative engine optimization (GEO), even if the biggest gain turns out to lie elsewhere. Plan your free intake.