Companies Like X ​

Enter the name or URL of any company and get 10 lookalike companies.

Try
Ready

Enter a company and hit search to see its 10 closest lookalikes.

How it works ​

The Lookalike model is two parts: retrieval and ranking.

Retrieval starts with a nearest-neighbor search in the Diffbot Knowledge Graph, using a 128-dimensional embedding model trained on an entity co-occurrence graph. Those results are then unioned with the results of a TF-IDF-weighted search over each entity's descriptor, description, and industry category.

Ranking scores every candidate with a gradient-boosted decision tree across 20 similarity signals — cosine similarity, TF-IDF, word-vector overlap, shared industry categories, matching entity subtypes, and relative entity importance and notability. The final score (0–1) is calibrated so the decision boundary sits at 0.5.

Get Lookalikes with DQL ​

Lookalike is included in DQL. To use it, add the similarTo() operator to any type:Organization DQL request.

DQL
type:Organization similarTo(name:"Stripe")
python
import requests

TOKEN = 'Your Diffbot Token'
QUERY = 'type:Organization similarTo(type:Organization name:"Hubspot")'
url = 'https://kg.diffbot.com/kg/v3/dql?type=query&token=TOKEN&query=QUERY'

response = requests.request('GET', url)
print(response.text)

Available free ​

Lookalike is available on all Diffbot plans, including the free tier.

Get started with a free token or reach out to our solutions team for a demo.