Look-alike models are used to increase scale by finding new people likely to be interested in your business because they resemble existing customers.

Use your CRM data to create lookalike audiences on Taboola! You can upload either a customer list of hashed email addresses, mobile device IDs, or 5 digit US zip codes and Taboola’s predictive engine will find similar users that are more likely to convert based on the assumption that these users will be “like” your current customers in your database.

Visit our Advertiser Data Use Policy here.

How Lookalikes Help You

• Proven Increase in Conversions - Targeting users who resemble your clients is proven to increase conversions and lower CPA
• Grow Reach - using lookalikes allows you to expand your reach and find new customers
• Harness Your Proprietary Data - Lookalikes takes advantage of the data you have on your customers and allows you to build off your past conversion success

Want to get started with Lookalikes? Learn how below.

Create a customer file audience in the 'Audience' page in backstage.

1. In the left menu in Backstage, click on ‘Audiences’ > ‘New Audience’ > 'From Customer File' to upload a customer file
   
   
2. Upload a CSV file with emails (hashed or non-hashed) mobile device IDs (IDFA or GAID), or 5 digit US Zip Codes
   
   
3. Choose emails, hashed emails, mobile device ID, or 5 digit US zip codes
4. Name the audience
5. Click "OK"
6. Files appear in the audiences table with "Processing" status
   
   
7. After 2-3 days, the lookalike audience status will change to "Ready"
   
   
8. When it's "ready", edit any campaign or create a new one to target the audience. In the 'Lookalike' tab of the 'Audience' section, you will see all your lookalike audiences.
9. You can choose the similarity of the audience (Top 5% - Top 25%) that you want to target using slider functionality:
   a. Top 5% are the top 5% of users that mostly resemble your seed file. This means higher performance, but lower scale.
   b. Top 25% are the top 25% of users that mostly resemble your seed file. This means the threshold for similarity is lower, but the scale of users is much bigger.
   c. We recommend starting with top 20% as a default. If you’re satisfied with the performance of the lookalike audience, continue to expand the audience to reach higher scale. Otherwise you can narrow your audience to achieve higher performance at the expense of scale.
   
   
   
   
10. Once a campaign is targeting an audience, the status of the corresponding audience will change to "Active"