Skip to main content
brighter websites logoBrighter Websites Logo White

Product vs Service Schema: I Argued With AI, Turns Out We Were Both Half Right

Author

Summary

Service businesses should use both Product and Service schema when an offering is priced and bookable. Product defines it as a sellable offer, while Service preserves what it actually is. This structure helps search engines and AI systems understand your business properly, even if rich snippets don’t appear.

Table Of Contents

Google Search Console threw a schema error.

Nothing major, Review was attached to the wrong parent. Easy fix. 

But I’ve got this habit, I ask myself Is this actually the best way to do it?

So I dropped it into Gemini and asked for a rewrite.

Read the advice… and just sat there for a second.

Yeah… I don’t agree with this.

Schema Isn’t a Rulebook, Even Though It Looks Like One

Most people treat schema like a checklist:

  • Pick a type
  • Fill the properties
  • Pass validation
  • Done

And technically… that works.

But it’s not what schema actually is.

Schema is a relationship system. It’s describing:

  • what something is
  • how it connects to other things
  • and how machines should interpret that context

That’s why properties like about, subjectOf, provider, itemOffered exist. They let you connect entities in ways that aren’t rigid parent-child structures.

Once you see that, the whole thing opens up.

You stop asking:  “What type do I use?”

And start asking:  “What am I actually trying to represent here?”

And maybe even stop trying to cram every possible type into every parent.  (Or at least I used to.)

AI Gave Me the “Right” Answer But it Felt Completely Wrong

So I’m debugging this schema issue.

Trying to integrate reviews properly into a service page. Keep it clean. Keep it compliant. Make sure the relationships actually make sense.

And Gemini suggests: “Use Product schema for your service so reviews can show.”

Immediate reaction: …what?

A massage is a service. Not a product. That’s basic.

And this is Gemini.
Google’s Gemini.

How does it get that wrong?

Plus, I don’t care about rich snippet reviews! I was trying to model the schema properly. Not the same problem.

I had a whole internal logic stack backing that up:

  • Services use offers (plural)
  • Products use Offer (singular)
  • Offers are tied to tangible things
  • Therefore → Product must mean physical

< examples here>

It felt consistent. It felt right. 

The Line I’d Never Actually Read Properly

So I switched to Claude.

“Gemini said this… I don’t agree.
A massage is NOT a product… Am I wrong?”

And the response I got back?

Gemini was actually pretty close, it gave good advice but just in the wrong context. 

I re-read what Gemini had actually said. And what Claude explained further. 

I still didn’t believe it. I visited https://schema.org/Product 

Right there, in black and white (and red) Any offered product or service. 

Wait… what? A haircut?

No really. Read that again.

Any offered product or service. For example: a pair of shoes; a concert ticket; the rental of a car; a haircut; or an episode of a TV show streamed online.

Product includes services.

And the canonical example?

A haircut. That’s the moment.

I never stopped to really read and understand that.  I was running on inference and intuition 

I was treating product as physical.

Schema.org is treating Product as a defined, sellable offer.

Different model entirely.

Why the Advice Was Directionally Right but Conceptually Wrong

Gemini wasn’t exactly wrong. It wasnt exactly right either. 

It skipped the layer that actually mattered.

It assumed: “You’re asking about reviews → you want stars” So it gave: Rich snippet strategy

But I was asking: “How do I model this correctly?”

Two different intents.

This matters more than it seems. Because:

  • Schema modelling = describing reality – relationships
  • Rich snippets = what Google might choose to show

They overlap. But they are not the same system.

The Real Constraint: Google, Not Schema

People think the option or rather limitation is: Service vs Product

It’s not.

The real constraint is Google.

More specifically:

  • Review snippet eligibility
  • Self-serving review suppression
  • Selective display rules
  • Search console Rich Snippet “Errors”

So:

  • Schema says what’s valid
  • Google decides what’s visible

And now we’ve got a third layer:

  • AI systems decide what’s understandable

That’s the Next Level shift.

 How to Model Product vs Service Schema for Service Businesses

Once you strip all the noise away, it becomes simple.

A massage session is:

  • A Service → what it is
  • A Product → how it’s sold

Its not  a workaround or a hack. Just accurate modelling.

Schema is where you decide: how machines understand what you sell.

If you start with understanding the parent definition and then:

  • relationships
  • constraints
  • and system behaviour

You’re not just implementing a schema. You’re thinking like a schema architect.

Instead of choosing one… You can use both.

  • Service → to define the nature of the offering
  • Product → to define it as a purchasable, structured offer
  • Offer → pricing, availability
  • Reviews → attached appropriately, not forced

It comes down to this: Is this a specific, priced, bookable offer?

  • Yes → Product + Service
  • No → Service only

And everything connects into a clean graph.

Subscribe to our blog or get in touch.

Get in touch with with Vanessa. Your Ballarat website strategist, designer and developer.
Occasional emails from Vanessa at Brighter Websites about AI, SEO, Website Marketing for regional service businesses.  

What Others think about this article

    Leave a Reply

    Your email address will not be published. Required fields are marked *

    See what we have done

    More Articles

    Discover more expert insights to the latest Web Design and SEO trends, these curated blog posts are designed to broaden your horizons and spark your imagination.

    Web Design
    AI-generated design system layout showing typography scales, colour tokens, and spacing rules for a website build

    AI Design Systems for Websites: How to Use AI to Build Better Website Rules Before You Build the Site

    A website design system is a collection of reusable coded rules defining how a site looks and behaves through typography, spacing, color tokens, and components. Unlike branding which establishes emotional identity, design systems translate that identity into buildable infrastructure. AI can generate structured design systems from logos and colors in minutes, but human judgment is essential for refining accessibility, eliminating redundancies, and ensuring production-ready foundations before building pages.

    Social Media Marketing
    Digital data trust highway leading towards bright horizon, Representing the path towards future visibility through a trust building proof library

    Social Media Management Automation From Website Content: How My One-Click Workflow Works

    Your Website Already Has the Content for Social Media Automation Service businesses already have plenty to talk about on socials. You’ve completed jobs. Solved difficult problems. Taken before-and-after photos. Answered the same customer questions dozens of times. You may even have years of useful knowledge sitting in project folders, email threads and phone galleries. The […]

    AI Citation Strategy
    Diagram showing local business visibility across Google Maps, AI tools, social platforms and review sites in a zero-click search world

    Search Everywhere Optimisation: How Local Businesses Win in a Zero-Click Search World

    Search Everywhere Optimisation helps businesses become visible across all platforms customers use to research purchases, from Google Maps to AI tools like ChatGPT. With 68% of searches now ending without clicks, success depends on clear business information, consistent details across platforms, credible proof and conversion paths that work outside your website.

    Artificial Intelligence (AI) and Generative AI
    Breakdance contact form HTML editor showing toolname and tooldescription attributes added for WebMCP AI agent compatibility

    How I Made Contact Forms Ready for AI Agents

    Chrome’s new Agentic Browsing specification allows AI agents to discover and interact with website tools like contact forms. This article documents how I retrofitted Breakdance forms for WebMCP compliance, including workarounds for limitations, JavaScript patches, and testing with Gemini. Key findings reveal how agents interact differently with forms than humans, particularly regarding checkbox fields and field descriptions.

    Uh oh, the form hit a snag.

    Looks like something didn’t load right.

    Give it another go, or flick us a message at support@brighterwebsites.com.au  if it keeps failing. We’ll fix it faster than you can say “cache clear.”

    You can phone Vanessa too - she doesn't mind a chat 0412401933