Customer Impact

SEO & GEO

How to Create a Wikidata Entity So AI Recognises Your Company

Copy for AI

To create a Wikidata entity is to add your company to Wikidata’s open knowledge graph.

AI models and search engines consult that graph to understand who you are. It gives a language model a machine-readable identity card: your official name, sector, location and relationships.

The verdict: a Wikidata item is worth it for an established B2B company with independent sources. Without those sources, it gets deleted and you’ve wasted your time.

Below: why it matters for AI recognition, whether you qualify, and how to build an item that holds up.

What is Wikidata and why do AI models look at it?

Wikidata is a free, structured database of facts about entities that machines can read directly.

The numbers behind it

The project has been running since 2012.

Its main page counted 123,886,647 data entities when we checked on 4 November 2026. Those are people, organisations, places and concepts.

Every item gets a unique number, called a QID. It also gets a set of properties, such as:

  • “instance of”
  • “headquarters location”
  • “industry”

The structured data is published under a CC0 licence. Anyone, including AI companies, may reuse it freely.

Why AI likes it

It removes guesswork. Ordinary web text is ambiguous, while Wikidata delivers explicit, typed facts.

Research into grounding language models in Wikidata shows models become more reliable when they can anchor themselves in such a graph. They also become less prone to hallucination.

It feeds more than one system

Wikidata also feeds the Google Knowledge Graph. That graph powers the knowledge panels you see on the right of the search results.

So it isn’t an isolated wiki. It’s a hub that ripples through several systems that shape how AI positions your brand.

For the bigger picture of how AI search works, start with our guide to generative engine optimization.

It turns your company into a recognisable node in a network of facts.

That’s exactly what AI needs in order to talk about you with confidence.

From loose text to one anchor

Instead of loose fragments about “a marketing agency in Belgium”, the model gets an unambiguous anchor:

  • this company
  • with this name and this QID
  • in this field of activity
  • with these external identifiers

Connections do the work

Your item refers to other items, such as your sector, your place of establishment and your founding year.

It also refers to external identifiers, for example your official company registration number or your LinkedIn page.

That creates a web of confirmations. It helps the model tell you apart from namesakes.

It’s the same principle as grounding, the process by which AI determines what is true. The more reliable, consistent anchors there are, the more certain the answer.

A signal, not a switch

A Wikidata entry only strengthens your entity if the rest of your digital footprint tells the same story.

You can read more about that in our piece on entity consistency for AI.

Does my company actually qualify for Wikidata?

Probably, if serious public sources describe you. Wikidata applies lower thresholds than Wikipedia, but not unlimited ones.

The three notability criteria

The Wikidata notability rules say an item qualifies if it meets at least one of three criteria:

  1. It has a valid sitelink to a page on Wikipedia or a sister project.
  2. It refers to “a clearly identifiable conceptual or material entity” that serious, public sources describe.
  3. It fulfils a structural need, such as making statements on other items more useful.

For most companies, the second criterion is the realistic route.

What counts as a source

You need independent references that back up the existence and identity of your business.

Good examples:

  • an official company registration
  • trade press coverage
  • industry registers
  • other sources you didn’t write yourself

Your own website alone isn’t enough as the only backing.

Be sober about it

A newly founded sole trader with no external traces probably doesn’t belong on Wikidata yet.

An established B2B company with a registration number, trade mentions and a demonstrable track record usually does.

Rule of thumb: the project is a general knowledge graph, not a business directory. Items that look purely promotional get removed.

How do you create a Wikidata entity step by step?

Through the “Create a new Item” function, but the preparation weighs more heavily than the click.

Work in this order.

Step 1: Check whether you already exist

Search Wikidata for your company name and its variants.

A duplicate item is counterproductive. The community will merge or delete it.

Step 2: Create an account

A registered, recognisable account increases trust. It also lets you track your contributions.

Step 3: Gather your sources in advance

Have your independent references ready before you start.

You want to be able to back up every important statement.

Step 4: Create the item with your official name

Use exactly the name as it appears in your registration. Add a short, factual description and any aliases.

For example, a label like “Example Consulting BV” with the description “Belgian B2B consultancy firm”. Factual, short, no adjectives.

Step 5: Fill in the core statements

Add the properties covered below. Give each one a source reference where possible.

Step 6: Maintain the item

If you move, change your name or change your offering, update it.

Watch out: an outdated item sends the wrong signals to every system that reads it.

Keep the tone strictly factual

Wikidata is no place for marketing language or value judgements.

The more neutral and better backed, the safer the item.

Here are the six steps in one picture:

METHOD A Wikidata item in six steps 1 Check duplicate does it exist? 2 Account register first 3 Sources gather upfront 4 Create item official name 5 Statements with sources 6 Maintain keep it current Preparation weighs more heavily than the click
A Wikidata entity is built in six steps, from duplicate check to maintenance.

Which properties does your Wikidata item need?

The ones that make your entity unambiguous and connect it to other items.

A minimally strong company profile usually contains:

PropertyWhat it doesExample value
Instance of (P31)Tells the model what kind of thing you are”enterprise” or a more specific company type
Official name and aliasesMatches your registration and websiteYour exact legal name
Country (P17) and headquarters location (P159)Geographic anchoringBelgium, your city
Industry (P452)Places you in the right semantic fieldYour sector item
Inception (P571)Founding date, if a source supports itThe year of registration
Official website (P856) and external identifiersLinks to other databasesCompany number, verified social profiles

How do you add a source to a statement?

Every statement can carry its own reference. Three reference properties cover most cases:

  • reference URL (P854): the web page that confirms the fact
  • stated in (P248): the publication or register it comes from
  • retrieved (P813): the date you checked it

For example, say a Ghent IT firm adds “inception: 2015”. The reference would point to its entry in the official company register, with the date it was retrieved.

That’s three extra fields per statement. It’s also what keeps the item alive when someone reviews it.

Watch out: don’t use your own homepage as the reference for every statement. Mix in at least one independent source.

External identifiers are underrated

They link your Wikidata item to other authoritative databases.

Those cross-connections make your entity hard to confuse or train away.

It’s the same logic behind why brand mentions beat isolated backlinks for AI authority. Multiple, consistent traces weigh more than one isolated signal.

Mirror it on your own site

Your website should confirm the same facts in structured data.

The sameAs property in your Organization schema can point to your Wikidata item. Our guide to sameAs schema shows how.

Here’s what a validator shows on our own homepage:

Schema Markup Validator result for the customerimpact.be English homepage listing the detected structured data types that should match your Wikidata entity The Schema Markup Validator on our own English homepage, October 2026. Check that each detected type matches what the page visibly says.

Your schema and Wikidata item should match word for word.

Which mistakes get a Wikidata item removed or merged?

Five mistakes cause most of the trouble:

  1. A duplicate: you skipped the search, so the community merges your new item.
  2. Promotional wording: “leading” or “innovative” in the description reads as advertising.
  3. No independent references: every statement points back to your own site.
  4. The wrong “instance of”: “business” for a product, or a person for a brand.
  5. Stale data: an old address or name that no longer matches your registration.

The first three are the classic reasons for deletion. The last two don’t delete the item, but they confuse every model that reads it.

Rule of thumb: write it as if a neutral librarian must approve it.

What does a Wikidata entity deliver and where are the limits?

A better chance that AI recognises your brand correctly, but no guarantee of visibility.

It works best as part of a coherent entity SEO strategy, not as a standalone trick.

Expect a slow build-up

AI models and knowledge graphs work with update and training cycles.

So the effect builds up over time rather than overnight.

It needs confirmation

Without confirmation from your website, your profiles and your mentions, a Wikidata entry stays an island.

The gain lies in the interplay. A Google knowledge panel usually appears only when several of those sources agree.

Who it pays off for

Above all for established B2B companies aiming at AI visibility that already have a demonstrable digital presence.

If you don’t have that foundation yet, you’ll get more from investing first in consistent NAP data and structured data on your own site.

Our view: we’d rather steer on visibility that brings quality leads than on a ticked-off checklist. Sometimes a Wikidata item simply isn’t the priority yet.

Frequently asked questions about creating a Wikidata entity

Is a Wikidata entity the same as a Wikipedia page?

No. Wikipedia is editorial text with stricter thresholds, Wikidata is structured data with lower, but real, thresholds. You can be on Wikidata without a Wikipedia article, and both feed knowledge graphs in their own way. Our guide on creating a Wikipedia page for your business covers the stricter route.

Does creating a Wikidata item cost money?

No, Wikidata is free and open. The real cost is your time: gathering sources, filling them in correctly and above all maintaining the item.

Does a Wikidata entity immediately increase my AI visibility?

Not right away. It’s an anchoring signal that works through over time, and only if the rest of your digital footprint confirms the same story. One item on its own rarely moves the needle.

What if my company is not notable enough?

Then your time is better spent on consistent data and structured data on your own channels. Build up independent traces first, and consider Wikidata once your entity is more mature.

Should you create a Wikidata entity?

Yes, if independent sources already describe your company; otherwise build those traces first.

It gives AI a reliable, machine-readable anchor, as long as it fits a consistent footprint.

Want to know whether it’s worth it for your company? Look at our generative engine optimization or contact us for a check of your entity signals.

Free website scan

Enter your website and get an automatic scan within minutes, with concrete technical and SEO improvements. No sales pitch.

Where should we send your report?

We only use your details for your scan. No spam, unsubscribe anytime.