Digital Dan
← All posts

Audience layering in Google Ads: advanced targeting strategies (2026)

Audience layering in Google Ads: advanced targeting strategies

Audience layering in Google Ads means adding audience segments on top of your keywords or other targeting, so the same campaign knows more about who is searching: a past visitor, an existing customer, someone in-market for your service, or a particular age group. On Search you layer segments in Observation (keep reach, gather data, inform bidding) or Targeting (show ads only to that segment), combine them with combined segments (AND, OR, NOT), and keep out people you never want to pay for with exclusions. In Performance Max and Demand Gen the same lists work as signals and seeds rather than hard limits.

Want a second pair of eyes on how your account uses audiences? I build and audit audience strategies in Google Ads accounts every week: talk to me about your Google Ads.

If you need a refresher on the segment types themselves (affinity, in-market, life events, detailed demographics, custom segments and your data segments), start with Google Ads audience targeting explained. This post is about combining them.

Why layering works

A keyword tells you what someone wants right now. An audience tells you something about the person: whether they’ve been to your site, bought before, are actively comparing options in your category, or fall in an age or income band your business serves. Two people typing “kitchen remodel cost” can be worth very different amounts to you, and layering is how the account learns the difference.

Diagram: audience layering on a Search campaign. Layer 1 is keywords, everyone searching for what you sell. Layer 2 is audience segments such as past visitors, customer lists, in-market and demographics. Layer 3 is exclusions, people you never want to pay for, such as customers. Together they decide who you reach and how you bid for each combination. Observation keeps reach and adds data; Targeting narrows reach to the segment

Observation vs Targeting

Every audience you add to a Search campaign or ad group gets one of two settings:

  • Observation doesn’t change who can see your ads. You get each segment’s performance in your reports, and Smart Bidding can use your data segments as signals. Google recommends it for Search.
  • Targeting limits the campaign or ad group to people in the segment. Use it only for campaigns built for that audience.

The setting applies to every segment on that campaign or ad group, so you can’t have one segment observed and another targeted in the same place. That’s one reason layered setups often use a separate campaign for the targeted audience. The full comparison is in Targeting vs. Observation.

Which segments Search accepts

Google’s compatibility table allows affinity, in-market, detailed demographics and your data segments on Search. Life events and custom segments are not available for Search, so custom segments are a layer for Display, Demand Gen, Video and Performance Max signals. Basic demographics on Search are age, gender and household income (household income only in some countries, the United States included).

Your data segments (remarketing lists for search ads)

Past website visitors, app users, YouTube viewers and Customer Match lists are the most valuable layer in almost every account. Google’s help page now gives a minimum of 100 active users in the last 30 days for Search, Display and YouTube. Customer Match on Search can be used in Observation and as an exclusion by any account in good standing, while Targeting needs 90 days of history and more than USD 50,000 in lifetime spend. Setup is covered step by step in my guide to remarketing lists for search ads (RLSA) and in the Customer Match guide.

Combined segments

A combined segment is one audience built from others with three relationships:

  • AND (include): people in every segment you list. “Visited the pricing page” AND “in-market for business software”.
  • OR: people in any of the segments.
  • NOT (exclude): removes a segment from the combination, for example everyone on your customer list.

Google’s help page says combined segments can be used in Display and Search campaigns, and that a combined segment with 1,000 members or fewer won’t serve; campaigns or ad groups targeted only to one that small are paused automatically. Intersections shrink fast, so check the estimated size before you build a campaign on one.

To create one:

  1. Go to Tools → Shared library → Audience manager.
  2. Open Your data segments, click the plus button and choose Custom combination.
  3. Name the segment.
  4. Add segments under Include (AND) and, if needed, Exclude (NOT), using Search or Browse.
  5. Click Create, then add the segment to a campaign or ad group like any other.

You can also build one from inside a campaign: Audiences, keywords and content → Audiences → Edit audience segments → Browse → Your combined audience segments → New combined segment.

Exclusions

Exclusions are the only layer that is always a hard rule. On Search you can exclude segments at campaign or ad group level from the Exclusions section of the Audiences page. Use them for people you don’t want to pay for: existing customers on a lead-generation campaign, recent converters, staff and suppliers on a customer list, or people who already booked.

Bid adjustments vs Smart Bidding

This is where many layering guides are out of date. Google’s bid adjustments page says audience adjustments (from -90% to +900%) are supported with Manual CPC and Maximize clicks, but if you set a manual bid adjustment on a Smart Bidding strategy (Target CPA, Target ROAS, Maximize conversions or Maximize conversion value), “it won’t be supported”. In other words, a +50% on past visitors does nothing in a Target CPA campaign.

With Smart Bidding, layering works differently. Your data segments in Observation give the bidding a signal, and the reporting shows you how each segment converts. If a segment needs different treatment than the bidding gives it, the honest options are to move it into its own campaign with its own target, or to exclude it. More on how each strategy behaves in automated bidding strategies explained.

Layering in Performance Max and Demand Gen

Performance Max: signals plus exclusions

In Performance Max, audiences are audience signals on each asset group. They guide the campaign but don’t limit it, and Google says it can take up to two weeks for new signals to be fully used. The strongest signals are your own data, then custom segments built from your best search terms and competitors’ sites. Layering here means giving each asset group a signal that matches its message, not narrowing reach.

The hard layer in Performance Max is exclusion. Campaign settings now include Your data exclusions, which can exclude Customer Match, website visitor and Google Analytics lists; Google notes they apply to Search, Display, Shopping, Video and Discover inventory within the campaign. If what you want is new customers, Google recommends the new customer acquisition goal over plain exclusions, because it adds reporting and lets you value new customers differently. Demographic exclusions in Performance Max cover age and gender. The rest of the campaign type is in Performance Max campaigns explained.

Demand Gen: targeting, optimized targeting and lookalikes

In Demand Gen you choose the audience per ad group, and every segment type is available, custom segments and life events included. Two things change the picture: optimized targeting, on by default, can reach people outside your segments when Google predicts they’ll convert (you can turn it off in the ad group), and lookalike segments let you seed an audience from a customer list. Similar audiences, the old version of this idea, are gone. A layered Demand Gen setup usually means a lookalike or custom segment, plus your customer list excluded where you only want new people.

Practical layered setups

These are the combinations I use most. None of them is a template to copy blindly; each depends on having enough data in the segment.

  1. Main Search campaigns with everything observed. Add all website visitors, converters, your customer list and two or three relevant in-market segments in Observation. Nothing changes in reach, and within a few weeks you know which audiences convert better or worse than average.
  2. Broader keywords, past visitors only. A separate Search campaign targeted to past visitors (Targeting) with broader or more expensive keywords you wouldn’t buy for cold traffic. People who already know you can justify a search term that strangers can’t.
  3. Past visitors AND in-market. A combined segment of people who viewed your service pages and are in-market for that service, used on Search or Display, when the intersection is large enough to serve.
  4. Customers excluded from lead generation. Upload a customer list and exclude it from lead-gen campaigns so you don’t pay for clicks from people who’ll call your existing number anyway. For repeat-purchase businesses, the opposite: a separate campaign for them.
  5. Demographic layer, observed first. Add age and household income in Observation before excluding anything. Excluding on assumptions is how accounts lose good customers. For housing, employment and credit ads in the US and Canada, Google’s policy doesn’t allow targeting or excluding by age, gender, parental status, marital status or ZIP code.
  6. Performance Max with first-party signals and exclusions. Customer list and converters as the signal, a custom segment from your top converting search terms, and existing customers excluded or the new customer acquisition goal on.
  7. Demand Gen from a customer-list lookalike. Seed from your best customers, exclude the list itself, and decide deliberately whether optimized targeting stays on.

How to measure whether layering works

  1. Use the Audiences report. In Audiences, keywords and content → Audiences, compare each segment’s conversion rate, cost per conversion and value with the campaign average.
  2. Wait for enough data. A segment with a few dozen clicks hasn’t proved anything. I look for at least a few weeks and enough conversions per segment before acting.
  3. Compare like with like. Past visitors will always convert better than cold traffic. The question is whether a separate campaign or exclusion improves the whole account, not whether one segment looks good.
  4. Test structural changes. When you split or exclude a large segment, run it as an experiment where possible; see how to run custom experiments in Google Ads.
  5. Check list health. Customer Match lists need refreshing (memberships expire after 540 days), and consent settings shrink website visitor lists in some regions. A layer built on an empty list does nothing.

Common mistakes I see in real accounts

  1. Targeting instead of Observation on a main campaign. Traffic falls overnight to a sliver of past visitors. It’s the first thing I check when a client says Search “just stopped”.
  2. Bid adjustments on Smart Bidding campaigns. They sit there looking like strategy and change nothing.
  3. Combined segments too small to serve. Three ANDs in a row can leave a few hundred people, and the ad group pauses.
  4. Expecting signals to restrict Performance Max. They don’t. Use exclusions or the new customer acquisition goal for hard rules.
  5. No first-party data at all. No visitor segments because the Google tag or the GA4 link was never set up, no customer list uploaded. Every layering strategy depends on these.
  6. Planning around similar audiences. Old playbooks still recommend them; they no longer exist.

FAQ

What is audience layering in Google Ads?

Adding audience segments on top of your keywords or other targeting so a campaign can report on, bid for, limit to or exclude specific groups of people, such as past visitors, customers or in-market segments.

Should I use Observation or Targeting for audience layering?

Observation on campaigns that should keep reaching everyone, which is most Search campaigns. Targeting only on campaigns built for one audience, such as a past-visitors campaign with broader keywords.

Do audience bid adjustments work with Target CPA?

No. Google says manual bid adjustments aren’t supported on Smart Bidding strategies. Audience bid adjustments work with Manual CPC and Maximize clicks.

Yes. Google’s help page says combined segments can be used in Display and Search campaigns. Only segment types that Search accepts are useful there, and a combined segment needs more than 1,000 members to serve.

What replaced similar audiences for layering?

Similar audiences were removed in 2023. Demand Gen has lookalike segments, and Performance Max and optimized targeting use your data as signals to find similar people.

Need help building your audience layers?

Audience layering is one of the places where an account’s first-party data turns into lower costs, and also where settings like Targeting and bid adjustments quietly do the wrong thing. If you want me to look at how your campaigns use audiences, book a call and we’ll go through it together.

Daniel Timponi

Digital Marketing Analyst & Consultant · 17+ years crafting digital marketing that converts.

Schedule a call