Google Knowledge Graph Search API: What It Returns and How to Use It for SEO

As of 28 September 2026 the original endpoint still works, with a free quota of 100,000 read calls per day per project. But the overview page now warns that Google is moving the Knowledge Graph Search API to Cloud Enterprise Knowledge Graph, and tells new users to start there. It also says the API "is not suitable for use as a production-critical service."

This guide covers the status of both versions, a first request, what each field in the response means, and the SEO checks worth running. It's the practical companion to our guide on building a knowledge graph for AI, which covers the markup side: connecting founders, products and profiles so machines can link them.

Status in September 2026: The Original API and Its Cloud Successor

Two products now do the same lookup. As documented on 28 September 2026:

The Cloud details come from Google's editions page, its quotas page and its migration guide, which says plainly: "The resultScore field has been removed from the response." Google calls Basic "a good fit for community applications that aren't for production and high-traffic use cases" and Advanced the edition for production. No pricing page for Enterprise Knowledge Graph was live on 28 September 2026; the pricing URL returned a 404.

Which one should an SEO use?

For spot checks on a handful of names, the original API is simpler: one key, one call, and it still reports resultScore. The Cloud editions need a Google Cloud project, the Enterprise Knowledge Graph API switched on and a service account or gcloud login, per the setup guide. Choose Cloud if you're building a tool that has to keep working, since that's where Google says new features go.

Is the endpoint live right now?

Yes. A call without a key on 28 September 2026 returned HTTP 403: "Method doesn't allow unregistered callers." A made-up key returned HTTP 400, "API key not valid." Both are answers from a working service that wants a real key, so you can't use it anonymously.

How to Make Your First Request

You need three things: a Google account, a Google Cloud project, and an API key. Google's prerequisites page walks through them.

1. Create or pick a project in the Google API Console.
2. Enable the Knowledge Graph Search API for that project.
3. Under Credentials, choose Create credentials, then API key.
4. Restrict the key to this one API, as Google's authorization guide suggests, so a leaked key can't be used for anything else.
5. Send a GET request with the key as the key parameter.

Here is a search for a brand name, limited to organizations. Replace YOUR_API_KEY with your key.

curl -G "https://kgsearch.googleapis.com/v1/entities:search" \
  --data-urlencode "query=Plannora" \
  --data-urlencode "types=Organization" \
  --data-urlencode "languages=en" \
  --data-urlencode "limit=5" \
  --data-urlencode "key=YOUR_API_KEY"

Plannora is our made-up example brand; swap in your own name. To look up an entity you already know, pass its ID instead of a query:

curl -G "https://kgsearch.googleapis.com/v1/entities:search" \
  --data-urlencode "ids=/m/0dl567" \
  --data-urlencode "key=YOUR_API_KEY"

The method reference lists every parameter. These are the ones that matter for SEO work:

What the Response Fields Mean

The response is JSON-LD: a list of matches, each wrapped in an EntitySearchResult. This is Google's documented sample, shortened (the image and licence fields are cut):

{
  "@type": "EntitySearchResult",
  "result": {
    "@id": "kg:/m/0dl567",
    "name": "Taylor Swift",
    "@type": ["Thing", "Person"],
    "description": "Singer-songwriter",
    "detailedDescription": {
      "articleBody": "Taylor Alison Swift is an American singer-songwriter and actress...",
      "url": "http://en.wikipedia.org/wiki/Taylor_Swift"
    },
    "url": "http://taylorswift.com/"
  },
  "resultScore": 4850
}

Google's definitions, and what each field means for an audit:

The ID: /m/ versus /g/

The kg: prefix is shorthand. The response's own context maps it to http://g.co/kg, so kg:/m/0dl567 expands to http://g.co/kg/m/0dl567. On 28 September 2026 that link redirected to a Google search for the entity.

The part after the prefix comes in two styles. Wikidata's property for the Google Knowledge Graph ID covers IDs that start with /g/, and says IDs starting with /m/ belong under its Freebase ID property. Freebase was a shared database that ran from 2007 to 2015. Wikidata notes that new Freebase IDs are no longer issued, but the old /m/ IDs "remain in use, including by Google Knowledge Graph." Both kinds work in a lookup.

resultScore has no scale

Google defines resultScore only as "an indicator of how well the entity matched the request constraints." It publishes no range; the sample above just happens to show 4850. Use it to see which result Google ranks first for your query, and nothing more. A score of 12 for your brand today and 30 next month doesn't tell you Google trusts you more.

SEO Uses for the Knowledge Graph Search API

1. Check whether your brand is an entity

This short script asks the API for your brand as an organization and flags any result whose official URL is your domain. It runs on Python 3.9 or later, reads the key from an environment variable, and prints one line per match.

import json
import os
import urllib.parse
import urllib.request
 
API_KEY = os.environ["KG_API_KEY"]  # keep the key out of your code
BRAND = "Plannora"
DOMAIN = "plannora.io"
 
params = urllib.parse.urlencode({
    "query": BRAND,
    "types": "Organization",
    "languages": "en",
    "limit": 5,
    "key": API_KEY,
})
url = "https://kgsearch.googleapis.com/v1/entities:search?" + params
 
with urllib.request.urlopen(url, timeout=20) as resp:
    data = json.load(resp)
 
rows = data.get("itemListElement", [])
if not rows:
    print("No entity found. Google has not matched this name to an Organization.")
 
for row in rows:
    entity = row["result"]
    mid = entity.get("@id", "").removeprefix("kg:")
    site = entity.get("url", "")
    verdict = "YOUR SITE" if DOMAIN in site else (site or "no official url")
    print(mid, "|", entity.get("name"), "|", entity.get("description", "-"),
          "| score", row.get("resultScore"), "|", verdict)

Run it once without types too, so you also see namesakes of other kinds: places, people, products.

2. Find your ID and the pages behind it

When your entity shows up, save its ID in your brand fact sheet, so you can recheck it later with ids= instead of a name search. The detailedDescription.url tells you which reference page Google leans on for the long description, often a Wikipedia article.

Those reference pages are what belong in your markup. Schema.org describes sameAs as the URL of a page "that unambiguously indicates the item's identity," such as a Wikipedia page, a Wikidata entry or an official website. Google's docs don't say it reads a Knowledge Graph ID from your markup, so don't rely on that. If your company has a Wikidata item that meets Wikidata's rules, the ID can be recorded there instead, under the properties above. Our guide to building a knowledge graph for AI shows where each of these links sits in a complete brand graph.

3. Check disambiguation against namesakes

Search your bare brand name without types. If a town, a band or another company comes back first, a search engine has to work harder to know which one a question means. That's your cue to pair the name with your category everywhere you describe yourself, the core idea behind entity SEO.

4. Map the entities in your topic

Query your category terms and competitors' names. The types and descriptions show how Google labels the players in your space. Remember that the API returns entities, not rankings.

The Entity Check Decision Table

Run the script, find your result in the left column, and take the action on the right. We call it the Entity Check: it turns one API response into a to-do list.

The fixes in the right column are the site-level work. When a wrong fact has spread beyond Google, our quick guide to fixing incorrect brand facts in LLM citations covers reporting and rechecking it. Our guide to the SEO knowledge graph shows how to mark up your own pages so they describe one consistent set of entities, and the definition of a knowledge graph in SEO explains where panels come from.

Limits Worth Knowing Before You Rely on It

• It can't edit anything. The API "is a read-only API." To change what Google shows, use the knowledge panel's Feedback link, or get verified and suggest changes. Google notes that "not all knowledge panels are claimable."
• It returns no relationships. Google says the API gives "individual matching entities, rather than graphs of interconnected entities," and recommends Wikidata's data dumps if you need the graph.
• It isn't a panel predictor. An entity in the API doesn't guarantee a knowledge panel. Panels are "automatically generated" from sources across the web.
• It isn't a monitoring service. Google warns against a "critical dependence" on it. A monthly spot check is fine; a production pipeline belongs on the Cloud editions.
• It says nothing about AI answers. Google's AI features draw on its wider systems, and other assistants use their own search providers. Check what AI engines say about you separately, as covered in our brand presence guide for Perplexity.

Rankbox doesn't query the Knowledge Graph or track how engines describe you. Its part is the work behind the right-hand column of the Entity Check. Answer-Space Research maps the questions buyers ask ChatGPT, Perplexity and Google about your category, and the Citation-Ready Writer turns them into source-backed articles, such as a facts page, that reach your site through Rankbox's API. The Business plan costs $49.50 a month with a 7-day trial; see pricing. For the markup itself, the free schema generator builds Organization, Person and other JSON-LD blocks you can paste in.