Targeting on TikTok: Audiences, Signals, and Sensible Constraints

What “Targeting” Means on TikTok (and What Actually Does the Heavy Lifting)

On TikTok, targeting is best thought of as constraints and starting signals, not a guarantee of who will see your ads. TikTok’s delivery system learns from early performance signals (impressions, watch time, clicks, add-to-carts, purchases, etc.) and then expands toward people most likely to take your desired action. Your job is to give it enough room to learn while still protecting efficiency (budget, relevance, and exclusions).

How TikTok uses signals

• Ad group settings: location, age, language, device, interests/behaviors, and any audience selections (custom/lookalike).
• Real-time engagement: how people interact with your ad (view duration, replays, shares, profile visits).
• Conversion feedback: events attributed to your ads (e.g., purchase, lead submit). More consistent conversion feedback generally improves optimization.
• Context: content adjacency and user behavior patterns (what users watch and engage with).

Because the system learns from volume and patterns, overly narrow targeting often reduces delivery (limited reach, higher CPMs, unstable learning, and repetitive impressions to the same small group).

Main Targeting Methods (When to Use Each)

1) Broad (Minimal Targeting)

What it is: You set only essential constraints (usually location, sometimes age/language) and let the algorithm find buyers.

Best for: Prospecting, scaling, and when you have strong creative variety and a clear offer.

Why it works: Maximum inventory + maximum learning. TikTok can explore and quickly shift spend toward pockets of performance.

Continue in our app.

You can listen to the audiobook with the screen off, receive a free certificate for this course, and also have access to 5,000 other free online courses.

Or continue reading below...

Download the app

Sensible constraints for broad:

• Use only what you must: country/region, legal age restrictions, language if necessary.
• Avoid stacking interests + narrow age + narrow device unless you have a compliance or product reason.

2) Interest/Behavior Targeting

What it is: You select interest categories and/or behavior signals (e.g., engaged with certain content types).

Best for: New accounts needing a directional starting point, niche products with clear affinity groups, or when broad is spending but not finding efficiency yet.

Common mistake: Selecting too many interests at once and assuming it becomes “more accurate.” In practice, stacking can shrink the audience and reduce delivery.

How to use it sensibly:

• Start with 1–3 tightly related interests per ad group.
• Create separate ad groups for distinct angles (e.g., “running” vs. “weight loss” vs. “meal prep”) instead of combining them.
• Keep other constraints light (don’t narrow age/device unless required).

3) Custom Audiences

What it is: You target people you already have a relationship with (or who have interacted with your assets). These are the backbone of retargeting.

Custom audience types you’ll use most

• Website visitors: all visitors or specific pages (product page, pricing page, cart).
• Engagers: people who watched your videos, visited your profile, clicked, or engaged with your TikTok presence (depending on available options).
• Customer list: uploaded emails/phone numbers (hashed) to reach existing customers or leads.

When to use:

• Retargeting warm traffic (visitors, add-to-cart, initiated checkout).
• Exclusions (e.g., exclude purchasers from prospecting).
• Reactivation (past customers after a cooldown window).

4) Lookalike Audiences

What it is: TikTok finds people similar to a “source” audience (purchasers, leads, high-value customers, etc.).

Best for: Prospecting with a performance-informed starting point, especially once you have a meaningful source audience.

Key variable: Lookalike size (often expressed as narrow vs. broad). Narrow is more similar but smaller; broad is larger but less similar.

Practical guidance:

• Use purchasers (or highest-quality event) as the source when possible.
• Test at least two sizes (e.g., narrow and broad) in separate ad groups.
• Keep other targeting light; don’t over-stack interests on top of lookalikes unless you have a specific reason.

Why Overly Narrow Targeting Hurts Delivery (and What It Looks Like)

When you narrow too much, you reduce the system’s ability to explore. Symptoms include:

• Low reach even with budget available.
• High CPM (you’re bidding for a small slice of inventory).
• Rising frequency quickly (same people see the ad repeatedly).
• Volatile results (performance swings day to day due to small sample sizes).

Rule of thumb: If you’re not legally required to restrict it, try to avoid combining multiple narrow constraints at once (e.g., small geo + tight age + multiple interests + device + lookalike).

Framework: Prospecting vs. Retargeting Ad Groups

Prospecting (Find new customers)

Goal: Reach new people likely to convert, with enough scale for learning.

Recommended ad group “building blocks”:

• Broad (baseline)
• Lookalike (purchasers or leads)
• Interest/behavior (1–3 interests per ad group, used selectively)

Prospecting exclusions (audience hygiene):

• Exclude purchasers (customer list and/or website purchase event audience).
• If you have enough volume, exclude recent add-to-cart / initiate checkout from prospecting so those users are handled by retargeting.

Retargeting (Convert warm traffic)

Goal: Re-engage people who showed intent but didn’t convert, with tighter messaging and shorter windows.

Recommended retargeting layers:

• Hot: Initiate checkout / add-to-cart (short recency)
• Warm: Product page / key page visitors
• Engaged: Video viewers or profile engagers (longer recency, lower intent)

Retargeting exclusions: Always exclude purchasers to avoid wasted spend and poor user experience.

Step-by-Step: Build a Clean Prospecting + Retargeting Structure

Step 1: Define your “must-have” constraints

• Geo (countries/regions you ship to or serve)
• Age restrictions (if required)
• Language (only if your creative is language-dependent)

Step 2: Create prospecting ad groups (start simple)

Build 2–4 prospecting ad groups that each represent a distinct targeting philosophy:

• Ad Group A: Broad (only must-have constraints)
• Ad Group B: Lookalike (Purchasers) (test one size)
• Ad Group C: Lookalike (Purchasers) (test a second size)
• Ad Group D: Interest (1–3 interests tied to your strongest angle)

Add exclusions to all prospecting ad groups:

• Exclude Purchasers (customer list + website purchase audience if available)
• Optionally exclude last 7–14 days ATC/IC if retargeting is active and large enough

Step 3: Create retargeting ad groups by intent and recency

Use separate ad groups so you can control budget and avoid mixing very different intent levels.

Apply exclusions to all retargeting ad groups:

• Exclude Purchasers
• Exclude “Hot” from “Warm,” and exclude “Hot + Warm” from “Engaged” (so users don’t overlap across retargeting layers)

Step 4: Set a basic audience hygiene plan

Audience hygiene prevents wasted spend and keeps reporting interpretable.

• Always exclude purchasers from prospecting and most retargeting (unless you run a dedicated upsell/cross-sell campaign).
• Use reasonable recency windows:

• Hot intent: 3–7 days
• Warm intent: 7–14 days
• Engaged: 14–30 days

• Refresh or rotate retargeting audiences if frequency climbs quickly (tighten windows, split by intent, or reduce budget).
• Keep exclusions consistent across ad groups so prospecting doesn’t “steal” warm users.

Examples: Targeting Setups for Common Niches

Ecommerce: Skincare (DTC)

Prospecting

• Broad (US, 18+), exclude purchasers (180 days) and ATC/IC (7 days)
• Lookalike: Purchasers (test narrow + broad)
• Interest: “Skincare,” “Beauty & Personal Care” (separate ad group from “Makeup” if you test it)

Retargeting

• RT Hot: ATC/IC 3–7 days, exclude purchasers
• RT Warm: Product page 7–14 days, exclude ATC/IC and purchasers
• RT Engaged: 50%+ video viewers 14–30 days, exclude all above and purchasers

Local Service: Dental / Invisalign Leads

Prospecting

• Broad with tight geo (radius or city/region), age 21–55 (only if your offer truly skews), exclude existing leads/customer list
• Interest test: “Cosmetic dentistry,” “Beauty,” “Weddings” (separate ad groups; don’t stack all)

Retargeting

• RT Hot: landing page visitors 3–7 days, exclude booked appointments
• RT Engaged: video viewers 14 days, exclude visitors 7 days and booked appointments

Digital Product: Fitness Program

Prospecting

• Broad (country, language), exclude purchasers
• Lookalike: Purchasers or high-intent leads (test two sizes)
• Interest: “Weight training” OR “Home workouts” (separate ad groups by program angle)

Retargeting

• RT Hot: checkout page visitors 3–7 days, exclude purchasers
• RT Warm: sales page visitors 7–14 days, exclude checkout visitors and purchasers
• RT Engaged: 75% video viewers 14–30 days, exclude all above and purchasers

B2B SaaS: Appointment Booking Tool

Prospecting

• Broad (country), keep constraints light; exclude existing trials/customers (customer list)
• Interest/behavior tests by persona: “Small business,” “Entrepreneurship,” “Marketing” (separate ad groups)
• Lookalike: qualified leads (if you have enough volume)

Retargeting

• RT Hot: pricing page visitors 3–7 days, exclude trials
• RT Warm: site visitors 7–14 days, exclude pricing visitors and trials

Reading Delivery Metrics: Reach, Frequency, and Saturation

Reach

What it tells you: how many unique people saw your ads. In prospecting, you generally want reach to expand steadily as the system finds new pockets of performance.

Red flags: reach stalls early while budget remains available (often a sign of overly narrow targeting or limited inventory).

Frequency

What it tells you: average number of times each person saw your ads.

• Prospecting: frequency should usually stay relatively low; if it climbs quickly, your audience is too small or your budget is too high for that audience size.
• Retargeting: higher frequency is normal, but it can become wasteful if it rises without conversions.

Practical actions when frequency is high:

• Expand audience (broaden targeting, widen geo if possible, loosen stacked interests).
• Shorten retargeting windows (e.g., 14 days → 7 days) to focus on fresher intent.
• Add new creatives (frequency often rises when the same few ads dominate delivery).
• Reduce budget on small retargeting pools.

Audience saturation (how to spot it)

Saturation happens when you’ve shown your ads to most of the available audience segment and incremental impressions are less productive.

Common saturation pattern:

• Frequency rises
• CPM rises (sometimes)
• CTR and conversion rate fall
• Spend concentrates on a shrinking set of users

What to do:

• For prospecting: remove unnecessary constraints, test broader lookalikes, or add a broad ad group if you don’t have one.
• For retargeting: tighten recency, split by intent, and ensure purchasers are excluded.

Quick Targeting Checklist (Use Before You Launch)

• Did you create at least one broad prospecting ad group?
• Are interests limited to 1–3 per ad group (not stacked across unrelated categories)?
• Are purchasers excluded everywhere they should be?
• Are retargeting ad groups split by intent + recency with overlap exclusions?
• Do your audience sizes support your budgets (avoid tiny pools with aggressive spend)?
• Do you have a plan to respond to rising frequency (expand, tighten windows, rotate creatives, adjust budget)?