In short: audience targeting is the decision about who sees your ad, built from four layers: who the person is (demographics), what they are interested in, what they are actively shopping for and what they have already done with you (visited the site, left details, bought). In 2026 most platforms treat the audience you define as a signal the algorithm may go beyond, not as a fence: Google turned Lookalike segments in Demand Gen into signals from March 2026, and Meta removed interest-based exclusions entirely. That makes your own data - site visitors, customer lists and conversion events - the most valuable audience you have, and the place to invest most of the work.
Written and reviewed by Shay Cohen, CEO and founder of SFB, who manages Google Ads and Meta campaigns for businesses in Israel. Platform facts were checked against the Google Ads and Meta help centers on 22.09.2026.
What is audience targeting
Audience targeting is a campaign setting that decides which people see an ad, based on characteristics the platform knows or estimates about them, or on their previous interaction with the business. Instead of paying for exposure to everyone browsing, you pay for exposure to people who are more likely to become customers.
It is worth separating audience targeting from the two other targeting types that live in the same campaign: keyword targeting (what the person typed into search right now) and content targeting (which site, video or topic the ad appears on). In a Search campaign the keyword is still the primary targeting and the audience is an extra layer; in visual campaigns such as Display, YouTube, Demand Gen and Facebook, the audience is usually the primary targeting.
The four audience types, and what each platform calls them
Every platform uses different names, but the logic is the same. The table maps the four layers to the names you will actually see in the interface:
| Audience layer | Question it answers | In Google Ads | In Meta (Facebook and Instagram) | Example |
|---|---|---|---|---|
| Demographics | Who is the person? | Demographics, detailed demographics (parents, homeowners, education), life events | Age, gender, location, language, demographics in detailed targeting | Parents of children under 3 in a given region |
| Interests | What are they into over time? | Affinity segments, custom segments | Interests and behaviors in detailed targeting | Running enthusiasts, home cooks |
| Purchase intent | What are they shopping for now? | In-market segments, custom segment based on search terms | No direct equivalent; the algorithm infers intent from behavior | Actively looking for a used car |
| Your data | What have they already done with you? | "Your data": website and app visitors, YouTube viewers, Customer Match, Lookalike | Custom Audiences and Lookalike Audiences | Added to cart and did not buy in the last 30 days |
Note Google's renaming: what used to be "audience types" is now audience segments, and "remarketing" is now "your data". Older guides that talk about Similar audiences in Google Ads are out of date: they were replaced by Lookalike segments, which are only available in Demand Gen campaigns.
Audience targeting in Google Ads
Which segment is available in which campaign type
Not every segment is available in every campaign. This is the table according to the Google Ads Help Center:
| Campaign type | Affinity | Custom | Detailed demographics | Life events | In-market | Your data |
|---|---|---|---|---|---|---|
| Demand Gen | Yes | Yes | Yes | Yes | Yes | Yes |
| Display | Yes | Yes | Yes | Yes | Yes | Yes |
| Search | Yes | No | Yes | No | Yes | Yes |
| Video (YouTube) | Yes | Yes | Yes | Yes | Yes | Yes |
| Standard Shopping | Yes | No | Yes | No | Yes | Yes |
In Performance Max campaigns there is no audience targeting in this sense. You add audience signals to the asset group, and Google states explicitly that there is no guarantee ads will serve only to people in those signals: if the system finds other users converting well, it will reach them.
Targeting vs. Observation
When you add an audience to a campaign or ad group in Google Ads, you choose one of two settings, and this is probably the decision advertisers get wrong most often:
| Targeting | Observation | |
|---|---|---|
| What it does | Restricts delivery to people in the audience only | Does not change delivery; only measures and allows bid adjustments |
| When to use it | Display, YouTube, dedicated remarketing | Search and Shopping campaigns (Google recommends it for all Search campaigns) |
| Effect on Smart Bidding | Limits the system | First-party segments in Observation are used as Smart Bidding signals |
| Common mistake | Using it in a Search campaign, which chokes volume | Adding dozens of audiences and never reading the report |
The practical rule: in Search, add your data segments and relevant in-market segments as Observation, wait for data, and only then decide whether a segment behaves differently enough to deserve its own ad group.
Optimized targeting and Lookalike: the fence became a signal
Optimized targeting in Display and Demand Gen campaigns lets the system expand beyond the audiences you selected when it identifies people likely to convert. In addition, according to the Customer Match help page, campaigns using Smart Bidding and optimized targeting automatically include all Customer Match lists in the account as signals, unless you opt out.
The big 2026 change is in Lookalike segments in Demand Gen: from March 2026 Google is moving them to "suggestion mode". The reach slider (narrow, balanced, broad - 2.5%, 5% or 10% of the target location) no longer restricts delivery but acts as a signal, and the system may reach people outside the similarity thresholds. Advertisers who want the old behavior have to apply through an opt-out form per ad group, with UI opt-out controls expected later in the year. Google also dropped the 100-user minimum for seed lists.
What it means: the quality of the seed list now matters more than the reach setting. Google itself recommends high-intent seed lists such as recent converters.
Audience targeting on Meta (Facebook and Instagram) in 2026
On Meta the direction is the same, and even more pronounced. The default in campaigns is Advantage+ audience, where your settings (age, interests, custom audiences) are only suggestions and the algorithm looks for buyers beyond them. Three changes worth knowing:
- Sensitive options removed (January 2022): Meta removed detailed targeting options related to health, religion, sexual orientation, political affiliation and ethnicity.
- Interest exclusions removed (2024-2025): Meta removed the ability to exclude people by detailed targeting - first in new campaigns, then in existing ones. According to Meta's internal testing, median cost per conversion was 22.6% lower without those exclusions.
- What still works: excluding custom audiences (existing customers, recent purchasers, lists) remains, and it acts as a hard constraint even inside Advantage+ audience.
So on Facebook, audience work has moved from interest settings to two things: the creative, which filters the audience by itself (an ad that speaks to business owners attracts business owners), and custom audiences. The full interface walkthrough is in the guide on how to build audiences on Facebook, and using them to bring visitors back is covered in Facebook remarketing.
Your own audiences: GA4 and customer lists
When platforms treat interests as a weak signal, your data is the one competitive advantage a competitor cannot copy. Three sources:
Building audiences in Google Analytics 4
In Google Analytics 4 go to Admin, then Data display, then Audiences, then New audience. You can start from suggested audiences (for example purchasers, or users who added to cart) or build from scratch using events, traffic source, pages and demographics. Three things are required for the audience to reach Google Ads:
- The GA4 property is linked to the Google Ads account.
- Personalized advertising is enabled for the property and region.
- The audience fills mainly from the moment it is created, not retroactively, so build audiences before the campaign, not on launch day.
Audiences every account should have: added to cart and did not purchase (7 and 30 days), viewed the pricing page or form and did not convert, purchasers (for exclusion or cross-sell), and visitors who came from a specific Facebook campaign, so you can keep reaching them on YouTube and Google Display. If the underlying data is unreliable, the audiences will be wrong; the GA4 audit checklist helps confirm events are collected correctly.
Customer lists (Customer Match)
Uploading a customer list (hashed emails and phone numbers) lets you reach existing customers on Search, YouTube, Gmail and Display. According to Google's help center, list membership lasts up to 540 days, and a list stays eligible only if it has at least 100 members added or updated in the last 540 days. Google now recommends managing lists through Data Manager rather than the old API path. The guide to uploading a customer list to Google Ads walks through the steps.
Dynamic remarketing
For online stores, the next layer is showing every visitor the products they viewed, from the feed. It requires a product feed and commerce events with product IDs; the full explanation is in the dynamic remarketing in Google Ads guide.
Narrow or broad audience? A decision table
The "precise vs. broad targeting" debate misses the point, because the answer depends on how much data the algorithm has. This is the decision table we work with at SFB (our professional judgment, not an official platform figure):
| Account situation | What to do with the audience | Why |
|---|---|---|
| Fewer than 30 conversions a month, small budget | Relatively focused: location, age, your data and intent segments | The algorithm lacks conversions to learn from, and going too broad wastes budget |
| 30-50+ conversions a month | Gradual expansion: Advantage+ audience, optimized targeting, Lookalike as a signal | There is enough signal, and manual restrictions usually raise cost per conversion |
| Niche product or B2B | Your data, customer lists and Lookalike on converters, with filtering creative | The target audience is small and interests do not identify it well |
| Search campaign | Audiences in Observation only | The keyword already signals intent; Targeting would choke volume |
| Remarketing | Narrow audiences by funnel stage, with a different message per stage | The goal is relevance, not scale |
How to build audience targeting the right way: 6 steps
- Define the customer from data, not gut feeling. Check GA4 and your CRM for who actually buys: ages, regions, devices, sources. It is often not the audience you assumed.
- Build your data audiences before the campaign. Visitors, cart adders, lead submitters, purchasers and a CRM customer list.
- Map audience to funnel stage. New audience (interests, Lookalike), aware audience (visitors, video viewers), warm audience (cart, form). A different message for each stage.
- Exclude what should be excluded. Recent purchasers from new-customer campaigns, and employees and partners from lead campaigns.
- Choose Targeting or Observation on purpose. Search and Shopping in Observation; Display and YouTube according to the goal.
- Measure by audience and adjust. In the Google Ads audience report and in Meta's age and placement breakdowns, check cost per conversion per audience after 2-4 weeks, and only then remove or split.
Common audience targeting mistakes
| Mistake | Symptom in the account | Fix |
|---|---|---|
| Many narrow interests in one ad set | Small audience, high CPM, a learning phase that never ends | Consolidate, broaden and let the creative filter |
| Audience set to Targeting in a Search campaign | Very low impressions relative to search volume | Switch the audience to Observation |
| Same ad for customers and cold audiences | Existing customers get an introductory offer | A separate campaign or ad for customers, or an exclusion |
| GA4 audiences built without a Google Ads link | The audience does not appear in Audience manager | Link the accounts and enable personalized advertising in GA4 |
| Customer list never refreshed | The list is ineligible or very small | Update at least monthly, ideally automatically |
| Relying on old guides | Looking for Similar audiences or interest exclusions that no longer exist | Work from the current settings in the tables above |
If your account shows several of these symptoms, our Google campaign management and Facebook advertising services always start with a review of the existing audience structure. For the next step after building audiences, see the guide to managing paid Facebook campaigns.
Frequently asked questions about audience targeting
What is audience targeting in simple terms?
It is the campaign setting that decides who sees the ad: based on who the person is, what they are interested in, what they are shopping for, or what they have already done with you. The goal is to pay for exposure to people who are likely to become customers.
What is the difference between Targeting and Observation in Google Ads?
Targeting restricts delivery to people in the audience you chose. Observation does not change delivery; it only measures how the audience performs and lets you adjust bids. Google recommends Observation for Search campaigns.
Can you still exclude interests on Facebook?
No. Meta removed detailed targeting exclusions in stages during 2024 and 2025. You can still exclude custom audiences, such as existing customers or lists, and that exclusion holds even in Advantage+ audience.
What happened to Similar audiences in Google Ads?
Similar audiences were replaced by Lookalike segments, which are only available in Demand Gen campaigns. From March 2026 those also became a signal rather than a constraint by default, with an option to request an opt-out.
How big does an audience need to be?
There is no single number. A Google customer list stays eligible only with at least 100 members updated in the last 540 days, and each network has its own minimum to serve. In practice, a remarketing audience of a few hundred people will spend very little budget, so small businesses combine several audiences or extend the time window.
Broad or precise targeting, which is better in 2026?
It depends on conversion volume. In an account with dozens of conversions a month, broad targeting with strong creative usually wins. In a new account with few conversions, a more focused audience saves budget until the algorithm has something to learn from.
Where should you build audiences, Google Ads or GA4?
Both. In GA4 you build audiences from detailed on-site behavior such as events, traffic source and sequences. In Google Ads you build segments from purchase intent, interests, search terms and customer lists. Either way, they all meet in the Google Ads Audience manager.
Should you target customers who already bought?
Yes, but with a different ad. A customer who bought responds to a complementary product, a repeat order or a loyalty club, not an introductory offer. If you have no message for existing customers, exclude them from new-customer campaigns.
Sources and review date
Facts were checked on 22.09.2026 against: the Google Ads Help Center - "About audience segments" (segment types, the renaming, availability by campaign type and audience signals in Performance Max), "About Targeting and Observation settings", "About Customer Match" (540 days, 100 members, automatic inclusion in Smart Bidding and optimized targeting, Data Manager) and "Use Lookalike segments to grow your audience" (suggestion mode from March 2026, the 2.5%/5%/10% thresholds, the opt-out form and the dropped 100-user minimum); Meta for Business's announcement on removing sensitive targeting options (November 2021, effective 19.01.2022); and Social Media Today's report (August 2024) on the removal of detailed targeting exclusions and the 22.6% figure from Meta's testing. The narrow-or-broad table and the practical recommendations are based on SFB's work on advertising accounts in Israel and are not an official platform figure.
