Audience layering in Google Ads: targeting vs observation

There’s a setting in Google Ads that decides whether an audience adds data or cuts your traffic, and plenty of people pick it without knowing what it does. It’s “Targeting” versus “Observation” when you add an audience to a campaign. Choosing wrong here is one of the most common reasons a campaign loses its volume overnight with nobody knowing why.

Audience layering (adding audiences on top of your campaigns) is one of the highest-return levers in Search, because it adds context about who the person searching is, not just what they search. But it only works if you understand the difference between the two modes. This guide covers what each one does, when to use which, which audiences are worth layering, and the mistakes that cost volume.

In 30 seconds:

• “Observation” adds an audience to measure and adjust bids, without changing who your ads reach
• “Targeting” restricts the campaign so it only shows to that audience, cutting reach
• By default, in Search you almost always want observation, not targeting
• The audiences that add most: remarketing, in-market, custom intent, and first-party data
• The most expensive mistake is setting targeting too early and choking volume by accident

What is observation and what is targeting?

They’re the two ways to add an audience to a campaign, and they do opposite things.

Observation adds the audience without touching reach: your ads keep showing to everyone who matches your keyword targeting, but now you can see how that subgroup performs and adjust its bid. It restricts nothing. Targeting does restrict: the campaign starts showing only to people in that audience. The rest of the traffic disappears.

Google’s own documentation on targeting and observation settings sums it up like this: observation doesn’t change who your ads reach, targeting does. That sentence is the whole decision. If you want to measure and bid differently without losing volume, observation. If you deliberately want to speak only to one specific audience, targeting.

When do you use observation and when targeting?

The practical rule: in Search campaigns, observation by default; targeting only when you have a clear reason to restrict.

Observation is what you want almost always because it gives you the best of both worlds: you don’t lose reach and you still get per-audience data and can raise the bid on the groups that convert better. A Search campaign with mid-intent keywords gains a lot from layering your recent-visitor list or a relevant in-market audience in observation, without giving anything up.

Targeting makes sense in specific cases: a pure remarketing campaign where you only want to reach people who already visited, a campaign with a specific message for a specific audience, or a budget so tight you’d rather concentrate it on the highest-value audience even at the cost of volume. Outside those cases, setting targeting is usually a mistake you discover late, when you notice the campaign barely has impressions. The logic ties into how you handle keyword match types: targeting an audience on top of already-restrictive keywords multiplies the restrictions and can leave the campaign with no air.

Which audiences are worth layering?

Not all add the same. The ones that, in my experience, give the most useful signal in Search:

• Remarketing and recent visitors. Someone who already visited your site and searches again has higher intent. It’s the most profitable observation audience almost always. The detail of how to build those lists is in Google Ads remarketing.
• In-market. Google classifies users actively researching a purchase category. Layering it in observation tells you whether your keyword attracts people genuinely in a buying phase.
• Custom intent. Audiences you define with keywords and URLs your ideal customer would search or visit. Useful for tuning bids toward specific profiles.
• First-party data. Customer, subscriber or buyer lists uploaded to Google. They’re the highest value because they’re yours and reflect real behavior, not inferred.
• Lookalikes of your first-party data. Where still available, they help extend reach toward profiles similar to your best customers.

The key is layering them in observation first, letting them accumulate data, and only then deciding what to do with each based on how it performs.

How do you adjust bids by audience?

It depends on whether you use manual or automated bidding, and the difference is big.

With manual bidding or enhanced CPC, audience layering in observation lets you apply bid adjustments: raise 20% on your remarketing list, lower on an in-market that doesn’t convert. You control the modifier based on the data each audience leaves behind.

With Smart Bidding (automated conversion- or ROAS-oriented bidding), things change: per-audience bid adjustments mostly stop applying, because the algorithm already uses the audience signal as one more input to bid in each auction. Here, layering audiences in observation is still worth it, but for the signal they add to the system, not for the manual modifiers. How this fits with automated bidding I cover in Smart Bidding in Google Ads.

That’s why the first step is always to know which bidding strategy you’re playing with. Layering audiences and expecting to adjust bids by hand when you have Smart Bidding on is working on something the system already ignores.

What mistakes repeat with audience layering?

Three, and all three cost money.

The first is the most common and most invisible: someone adds an audience, leaves the default targeting option checked without noticing, and the campaign that had thousands of impressions drops to hundreds. Since there’s no visible error, it can take weeks to spot. Every time a campaign’s volume drops off a cliff with no obvious budget or bid change, this setting is one of the first things I check.

The second is the opposite: not using audiences at all and letting the campaign fly blind. Even with Smart Bidding, layering audiences in observation gives the system more context to decide. It costs no reach and adds information.

Frequently asked questions

What’s the difference between targeting and observation in Google Ads?

Observation adds an audience to measure its performance and adjust its bid, without changing who your ads show to. Targeting restricts the campaign so it only shows to that audience, cutting reach. The practical rule in Search is to use observation by default and reserve targeting for when you want to restrict on purpose.

Should I use targeting or observation in a Search campaign?

In most Search campaigns, observation. It gives you per-audience data and the option to adjust bids without losing volume. Targeting makes sense in pure remarketing campaigns or when a very tight budget forces concentrating on one specific audience, accepting less reach in exchange.

Do per-audience bid adjustments work with Smart Bidding?

Largely no. When you use an automated strategy oriented to conversions or ROAS, the system already incorporates the audience signal into each auction, and manual per-audience bid modifiers stop applying. Layering audiences in observation is still useful, but for the signal they add, not for the manual adjustments.

Which audiences should I add first?

Start with remarketing (recent visitors) and an in-market relevant to your product, both in observation. They give the most useful signal with the least risk. If you have your own customer lists, upload them too: first-party data tends to be the highest value because it reflects real behavior.

Why did my campaign lose volume when I added an audience?

Almost certainly because you added it in targeting mode instead of observation. Targeting restricts the campaign to that audience and cuts reach instantly. Go into the campaign’s audience settings, change the setting to observation, and the volume should recover.

Conclusion: the small setting that moves whole campaigns

Audience layering is one of the most profitable levers in Search, but it all hinges on a binary setting plenty of people pick without looking. Observation adds data at no reach cost; targeting restricts on purpose. In most campaigns you want the first, and you reserve the second for when there’s a clear reason to speak only to one audience.

What I recommend: layer remarketing and in-market in observation from the start, let them accumulate data, and before touching anything check which bidding strategy you’re playing with, because with Smart Bidding the value is in the signal, not the modifiers. And every time a campaign loses volume suddenly, check this setting before anything else.

This is part of the account-structure work I do as a Google Ads consultant. If you want to go through your account’s audience structure and where you’re losing volume or signal, you can book 30 minutes of consulting.