TikTok Ads Targeting Options (2026): Every Dimension and How They Combine

TikTok's targeting options are all set in one place — the ad group — and they combine according to one rule that decides whether your audience ends up workable or impossibly narrow. Most targeting mistakes on TikTok are not "picked the wrong interest." They are stacking four dimensions that each sounded reasonable, and ending up with an audience the auction cannot serve.

This guide lists every targeting dimension TikTok Ads Manager offers, explains the combination logic, and flags the options that are restricted or that behave differently from their Meta equivalents.

Where targeting lives, and how selections combine

Targeting is configured at the ad group level during campaign creation. The campaign holds the objective and budget structure; the ad group holds delivery, which is where audience, placement, schedule, bid and optimization all sit together. If you want two audiences tested against each other, they have to be two ad groups.

The combination logic is documented plainly and is worth memorizing:

• Within one dimension, selections are OR. Picking three interests reaches people who match any of them.
• Across dimensions, selections are AND. Picking "Female" and "25-44" reaches only people who match both.

Every dimension you add is therefore a multiplication, not an addition. Four AND-ed filters on top of each other is how a country-sized audience becomes a village. TikTok also notes that available targeting locations vary by market, and that some options are unavailable when advertising to people under 18.

Demographic targeting

The base layer, and the one most advertisers over-tighten:

• Location — country and, depending on market, region, state, city or metro level.
• Age — bracketed rather than year-by-year.
• Gender.
• Language.
• Spending power / household income — available in the US.

Two habits are worth breaking. First, age and gender brackets are inferred, not declared, so a tight bracket removes real buyers on the boundary. Second, the system will not expand demographics for you even when it expands everything else, so a demographic mistake is a permanent ceiling on that ad group rather than something delivery can route around.

Audience targeting: custom and lookalike

Custom audiences are built from your own data, and TikTok supports several source types, including customer file, engagement, app activity, website traffic, lead generation, business account audience, shop activity, challenge and premium audiences.

One number matters more than the rest: an audience needs 1,000 matched users before it can be used in an ad group. A list of 5,000 emails is not a 5,000-person audience — the matched count is what counts, and match rates depend on data quality. If an audience will not attach to an ad group, size is the first thing to check.

Lookalike audiences are built from a custom audience as the seed. The seed's quality is the variable that matters — a seed of recent purchasers behaves differently from a seed of all site visitors, at the same seed size.

Both can be included or excluded. Exclusion is the underused half: excluding existing customers or recent converters from a prospecting ad group is usually a bigger efficiency change than any interest selection. For app advertisers, the same logic applies to excluding people who already installed, which we cover in the TikTok app install campaign guide.

Interests versus behaviors — they are not the same thing

This pair gets conflated constantly, and TikTok draws the line clearly.

Interest targeting delivers to people based on sustained higher interaction with a category of content — gaming, makeup, outdoor equipment, sports news, pet care. It reflects a longer-run pattern and it takes both organic and paid content into account. Think of it as "what this person is generally into."

Behavior targeting reaches people based on recent in-app actions on organic content, and the lookback windows are short and specific:

• Video interactions — video completion, likes, comments and shares, over roughly the past 7 or 15 days depending on the action.
• Creator interactions — people who followed and viewed the home pages of a given type of creator in the past 90 days.

Because behavior windows are short, behavior-targeted ad groups are smaller and go stale faster, and they need more hands-on management than interest targeting. The practical split: interests for scale, behaviors for recency and for creator-adjacent angles that pair well with Spark Ads.

Device targeting

Device options are genuinely useful for app and commerce advertisers, and mostly noise for everyone else:

• Operating system and OS version.
• Device model.
• Device price — the price band of the handset.
• Carrier and internet service provider.
• Connection type — Wi-Fi or mobile network.

OS version is the one that repays attention: if your landing experience or SDK has a floor, set it here rather than paying for traffic that cannot complete. Device price is a blunt proxy for purchasing power and is restricted in some contexts — see special ad categories below.

Smart targeting, and what it will not expand

TikTok's expansion features — Smart Interest and Behavior and Smart Audience — launch your ad with the exact targeting you set, then widen it when the system predicts delivery will be difficult or observes performance declining.

Two guardrails are documented and worth quoting in your own team's playbook: expansion does not widen demographic variables such as age or gender, and it does not deliver into audiences you have excluded. So exclusions stay honest under expansion, while a narrow interest stack may quietly stop being the thing your results came from. When you review results from an ad group with expansion enabled, remember the delivered audience may be broader than the one on the settings screen.

Special ad categories: what gets switched off

Advertisers offering housing, employment or credit opportunities to users in the United States or Canada must declare a special ad category, and the restrictions are substantial:

• Age: 18+ only.
• Gender: not allowed.
• Location: ZIP code targeting not allowed.
• Lookalike audiences: not allowed, with restrictions on custom audience exclusions.
• Interest keywords and hashtag targeting: prohibited.
• Automatic targeting and targeting expansion: not allowed.
• Device price: restricted.

If you run these verticals, build the plan around what remains rather than discovering the gaps in the middle of setup.

A sane order for building an ad group

1. Location, language, and the OS floor if you have one. Hard constraints first.
2. Exclusions. Existing customers, recent converters, installed users.
3. One audience idea per ad group. Broad, interest-based, behavior-based or custom — not all four at once.
4. Check the estimated audience size before you leave the screen. If it collapsed, remove one dimension rather than loosening all of them slightly.
5. Leave demographics wide unless you have evidence, because expansion will never fix them for you.
6. Test across ad groups, not inside one.

If an ad group with plausible settings refuses to spend, targeting width is one of the standard suspects — the status-first sequence for diagnosing that is in TikTok ads not spending, and the structural settings that lock at launch are in the TikTok Ads Manager guide.

Where measurement fits

Targeting decides who sees the ad. What you learn from it depends on whether the events coming back are clean, which is a separate build: pixel and server-side events, deduplication, and a link layer that keeps your parameters intact from click to landing page. Setting that up is covered in the TikTok Pixel setup guide.

DeepClick works on the post-click half of this — link routing, destination control and attribution for Meta and TikTok advertisers, on your own domains. Details are on the Meta and TikTok advertiser solutions page.

FAQ

What targeting options does TikTok Ads Manager have?

Demographics (location, age, gender, language, and spending power in the US), audiences (custom and lookalike, include or exclude), interests, behaviors (video and creator interactions), device options (OS, model, price, carrier, ISP, connection type), and smart expansion features. All of them are set at the ad group level.

What is the difference between interest and behavior targeting on TikTok?

Interest targeting reflects longer-run affinity for a content category and accounts for both organic and paid content. Behavior targeting reflects recent actions on organic content — video interactions in roughly the last 7 to 15 days, and creator interactions in the last 90 days. Interests give you scale; behaviors give you recency.

How large does a TikTok custom audience need to be?

It needs 1,000 matched users before it can be used in an ad group. The matched count, not the size of the file you uploaded, is what the requirement applies to.

Does TikTok expand my targeting automatically?

With smart expansion features enabled, TikTok can widen targeting when delivery is predicted to be hard or performance declines. It does not expand demographic variables such as age or gender, and it does not deliver into audiences you excluded.

The short version

Targeting on TikTok is one screen with six dimensions, combined OR inside a dimension and AND across them. Start with hard constraints and exclusions, use one audience idea per ad group, keep demographics wide unless you have evidence, and remember that smart expansion can widen everything except demographics and your exclusions. If you run housing, employment or credit offers in the US or Canada, plan around a materially shorter list of available options.