Schema markup barely moves your AI citations

Let's say it plainly, even if it tempers an easy sales pitch: adding Schema markup to your pages won't make your brand show up in AI answers. Schema is still useful, just not for that. And conflating the two wastes your time.

What the numbers say

Ahrefs measured the effect of Schema markup on citations in generative engines. The result is unambiguous:

Adding Schema markup had no impact on AI citations: AI Overviews −4.6%, AI Mode +2.4%, ChatGPT +2.2% — variations indistinguishable from zero (Ahrefs study).

Three numbers hovering around zero, one of them negative. That's not a weak signal — it's the absence of one. Schema doesn't buy citations.

Why Schema isn't enough

JSON-LD describes your page in a vocabulary machines understand. It's an ID card, not an argument. It says who publishes and what type of content it is. It doesn't say why your page deserves to be cited over another.

Generative engines don't reward vocabulary compliance. They reward:

Substance. An answer that is genuinely complete, accurate, current. That's what gets extracted and rephrased.
Clear named entities. A brand, a place, people, dates — disambiguated and consistent everywhere. For a model to cite you, it first has to be sure who you are.
Real authority. Mentions, links, a consistent presence across the web. The signal that others trust you.

Schema can help with the second point — it makes your entities legible. But it creates neither the substance nor the authority. It labels what already exists.

Schema is still useful — for something else

Don't throw out your JSON-LD over this. For classic Google, it still does its job:

Rich results. Review stars, expandable FAQs, breadcrumbs, product cards. These enriched displays depend directly on markup, and they win clicks.
Page understanding. Schema helps Google interpret the nature of your content without ambiguity. That's technical hygiene, not an option.
Entity disambiguation. A clean Organization block — name, founding, founders, profiles — lowers the risk of being confused with a namesake.

All of this matters. It just doesn't translate mechanically into AI citations. Two different goals, two different levers.

Where to put the effort

If your goal is to be cited by ChatGPT, AI Overviews, or Perplexity, Schema markup isn't the priority lever. The order we recommend:

First, substance. Cover the topic better than anyone else. Direct answers, sourced numbers, concrete examples. Non-negotiable.
Then, entities. Be unambiguously identifiable — name, place, dates, people — consistently across your whole site. That's our GEO Guide in one sentence, and Schema contributes here, in its rightful place.
Finally, authority. Get mentioned and cited elsewhere. The rest follows.

Add Schema because it serves Google and clarifies your entities. Not because it buys AI citations: the data says it doesn't.

That's exactly the distinction Swissalytics measures. Our audit separates what weighs on classic SEO from what weighs on AI visibility — so you stop optimizing the right lever for the wrong goal.