Your own data is the most valuable targeting input you have. Customers who have already bought, people who visited your product pages and viewers who watched your ads to the end all tell TikTok something real about who your business serves. Custom audiences let you reach those people again. Lookalike audiences let you find more people like them.
This guide covers every main type of custom audience on TikTok, how lookalikes work, how to size and refresh them and how to use them alongside broad targeting.
What Custom Audiences Are
A custom audience is a group of people defined by their relationship with your business rather than by demographics or interests. You create them in the Audiences section of Ads Manager, and you can include or exclude them in any ad group. TikTok matches your data to its users, so audience sizes are always smaller than your raw lists.
Types of Custom Audience
Customer file
Upload a list of emails, phone numbers or mobile advertising IDs. TikTok hashes the data and matches it to users. Use customer files to exclude existing customers from acquisition campaigns, to retarget past buyers with new products and as a seed for lookalikes.
Only upload data you have permission to use for advertising, and follow the privacy laws in your markets. Most advertisers hash or let the platform hash data before upload. Match rates vary, and lists of a few thousand people may match only a fraction.
Website traffic
Built from the TikTok pixel and Events API: all visitors, visitors to specific pages, or people who completed events such as add to cart or purchase, within a time window. Our guide to the TikTok pixel covers setup.
App activity
People who installed your app or performed in-app events, through your mobile measurement partner. See our guide to app campaigns.
Engagement
People who interacted with your ads: viewed for certain lengths, clicked, liked, commented, shared or followed. Video view audiences are especially useful because they build cheaply from any campaign.
Lead generation
People who opened or submitted your instant forms. Useful for retargeting people who started but did not finish a form.
Business account and LIVE
People who engaged with your TikTok business account, such as followers, profile visitors and people who interacted with your organic videos or LIVE streams.
Shop activity
For TikTok Shop sellers, people who viewed products, added to cart or bought in your shop.
Choosing Time Windows
Most custom audiences let you choose how far back to look. Shorter windows mean warmer, smaller audiences. Longer windows mean bigger audiences with weaker intent. A typical pattern:
- 1 to 7 days: cart abandoners and recent product viewers, the hottest audience.
- 8 to 30 days: site visitors and engaged video viewers who have cooled a little.
- 31 to 180 days: past visitors and customers for re-engagement or new product announcements.
Match the window to your buying cycle. An impulse product needs short windows. A high-ticket purchase may take weeks of consideration.
How Lookalike Audiences Work
A lookalike audience starts from a source custom audience and finds TikTok users with similar characteristics and behaviour. You choose the location and a size setting, usually described as narrow, balanced or broad. Narrow lookalikes are most similar to the source but smaller. Broad lookalikes reach more people with less similarity.
The quality of a lookalike depends on the quality of the source. A lookalike of your best customers is more valuable than a lookalike of everyone who visited your homepage.
Choosing Good Seed Audiences
- Purchasers are the strongest seed for e-commerce.
- High-value customers, such as repeat buyers or the top spenders, can produce even better lookalikes if the list is large enough.
- Qualified leads or closed deals beat all form submissions for lead-based businesses.
- Engaged video viewers work when you lack purchase volume, though they find watchers more than buyers.
Seeds need a minimum number of matched users, and larger seeds produce more reliable lookalikes. Check current minimums in Ads Manager.
Lookalikes vs Broad Targeting
As TikTok's delivery system has improved, broad targeting often performs as well as lookalikes for conversion campaigns, because the algorithm already learns from your conversion data. Lookalikes still help when:
- Your account is new and the pixel has little data.
- You are optimising for a higher-funnel event and want to point the system toward buyers.
- You run awareness campaigns without a conversion signal.
- You enter a new market where you have customer data but little ad history.
Test a lookalike ad group against a broad one with the same creative. Our guide to targeting explains how to test fairly.
Using Custom Audiences for Exclusions
Custom audiences are as useful for exclusion as inclusion. Exclude recent purchasers from prospecting, exclude retargeting pools from top-of-funnel campaigns and exclude converted leads from lead campaigns. Clean exclusions stop you paying twice for the same person and keep reporting honest.
Retargeting With Custom Audiences
The most common use of custom audiences is retargeting: showing new ads to people who engaged but did not convert. Our dedicated guide to retargeting ads covers windows, creative and frequency.
Keeping Audiences Fresh
- Pixel, app and engagement audiences update automatically.
- Customer file audiences do not. Refresh uploads regularly, monthly at minimum, or connect your CRM or e-commerce platform if an integration is available.
- Rebuild lookalikes periodically from updated seeds, especially after strong sales periods.
- Delete or archive audiences you no longer use, so the library stays manageable.
Privacy and Compliance
Custom audiences rely on personal data, so treat them carefully. Only use data collected with appropriate consent and notices. Respect opt-outs. Do not build audiences around sensitive characteristics such as health conditions, and follow TikTok's rules for restricted categories. Our guide to advertising policies covers the platform rules, and your legal adviser can cover the law in your markets.
A Worked Example: Building an Audience Library
An online pet supplies retailer set up a small library of audiences that covered most of its needs:
- Purchasers in the last 180 days, used as an exclusion for prospecting and as a lookalike seed.
- Repeat purchasers, used as a second lookalike seed and for new product announcements.
- Add to cart in the last 7 days without purchase, used for short-window retargeting.
- Product page viewers in the last 30 days, used for broader retargeting.
- Viewers who watched most of any ad in the last 60 days, used to retarget engaged prospects.
- A monthly customer file upload, used to exclude existing customers from acquisition campaigns.
With these few audiences, named clearly and refreshed on a schedule, the team could build any campaign it needed without creating new audiences each time.
Naming and Organising Audiences
Audience libraries become messy quickly. Name each audience with its source, event and window, such as "Web_AddToCart_7d" or "CRM_Purchasers_Monthly". Record when customer files were last refreshed, and delete audiences built for one-off campaigns once they end. A clean library makes exclusions reliable and prevents campaigns from using stale or wrong audiences.
Audiences Across Markets
When you advertise in several countries, build lookalikes separately for each market rather than applying one market's lookalike elsewhere. Customer behaviour differs by country, and a lookalike built from local customers finds better local prospects. If a new market has too few customers to seed a lookalike, use broad targeting with strong local creative until enough customer data builds up.
Common Mistakes
- Uploading tiny customer lists that barely match.
- Building lookalikes from low-quality seeds such as all site visitors.
- Never refreshing customer file uploads.
- Forgetting to exclude purchasers from prospecting.
- Assuming lookalikes will always beat broad without testing.
Frequently Asked Questions
How many people do I need for a TikTok lookalike audience?
TikTok sets a minimum number of matched users for a source audience, and the exact minimum can change, so check Ads Manager. In practice, larger seeds of high-quality users, such as a few thousand purchasers or more, produce more reliable lookalikes.
Why is my customer file match rate low?
Not every customer uses TikTok, and some register with different emails or phone numbers. Formatting also matters. Use clean, consistently formatted data, include phone numbers with country codes and upload larger lists.
Do lookalike audiences still work on TikTok?
Yes, but their advantage over broad targeting has narrowed for conversion campaigns with good data. They remain useful for new accounts, new markets and campaigns without a strong conversion signal.
How long do TikTok custom audiences last?
Pixel, app and engagement audiences have rolling windows that you set, up to the maximum the platform allows. Customer file audiences stay as uploaded until you update or delete them.
Can I share custom audiences between ad accounts?
Within a Business Center, audiences can usually be shared between ad accounts you control. This is useful for brands running several markets or agencies managing several accounts for the same brand.
Should I exclude existing customers from TikTok ads?
For acquisition campaigns, usually yes, so you do not pay to reach people who already buy from you. For repeat purchase, cross-sell or new product campaigns, target them instead.
How often should I refresh a TikTok customer file audience?
At least monthly, or more often if you acquire customers quickly. Where an integration with your store or CRM is available, use it so the audience updates automatically.