Last updated November 2nd 2023
There are three main ways to use audiences: target, observation only or exclude - in the following we go through the main use cases for each of these categories.
Targeting an audience means your ads will only be shown to users within the audience(s) you target, making this a more precise and narrow targeting than using audiences for observation only. This is not true, however, if your campaign is opted into audience expansion and optimised targeting, which will allow the campaign to go beyond your targeting and show your ads to users outside of your targeted audiences.
“Cold audience” targeting refers to targeting audiences which have not previously interacted with your brand or page. These can be demographics, in-market, interest, custom segments (combining other types) or any other type of audience outside of your customer lists.
Cold audiences can be in different stages of the marketing funnel, and a cold audience does not automatically mean an upper funnel one, rather it indicates its newness to your business.
Cold audiences can be good for prospecting, if you do not know exactly who your customers are, and are looking to find fresh traffic or generate brand awareness. It is generally less effective at generating immediate conversions, but can be layered together with other restrictive measures such as more specific keywords in tighter match types, to ensure you are spending money on relevant traffic.
Remarketing audiences are available for YouTube and Display, and can be a great tool to build rapport with customers and make sure you leverage your initial investment in ad spend into a sale or lead, once the customer is ready for it.
Avoid wasted spend on users who will never convert
It is recommendable to exclude bouncers from your remarketing campaign - these are likely to be users who are not interested in what you are offering, and are unlikely to convert. Following this logic, set up and ad frequency cap on your campaigns, to avoid showing the same ad over and over again to users who are simply not interested
To set up a successful remarketing campaign, it pays off spending some time to analyse and get to know your conversion path. Knowing how your converting users typically interact with your page and ads will show you which conversion actions to target with remarketing.
Consider setting up multiple remarketing tracks
When setting up remarketing the most common strategy is to set up a high level branded ad for all website visitors. There is however a lot to be gained by adding granularity to this strategy. Remarketing can be a way to create meaningful conversations at scale with your users:
Your users communicate with you through their behaviour on your site. You can set up appropriate "replies" by creating several remarketing tracks to direct the most relevant message to each user. For example, "If a user has seen page x but not y, show this ad" or "If a user has seen page x and y, show this ad".
For the user, it should feel like a dialogue between their behaviour on your site and the ad content you respond to them with.
A fitness equipment e-commerce website has a blog with articles about health and training. They sell a variety of products, including supplements, weights, and training equipment.
To set up effective remarketing, they should show users supplements or weights that help build muscle mass if they have viewed an article on that topic.
They could also target users who have viewed the products page, but not the blogpost with an ad leading to the blogpost on the benefits of the product, and an ad group that provides education on the benefits of the product.
This can be done by showing a specific ad copy or ad group to users who have visited the product page but not the blog post.
If users have viewed both the blog post and a relevant product page, they are likely further down the funnel and ready for ad copy that shows them why they should buy from this specific retailer, such as price points and time limited offers.
Analysing your conversion time lag is a valuable investment to determine how long you should remarket to a user. For example, if the average cycle from a user's first engagement to a purchase is 20 days, this will inform the limit you set for remarketing.
Using the average conversion time window will cut off some users with longer conversion windows, but starting small is best practice and gradually increasing if CPA/ROAS allows for it.
The remarketing list for search ads (RLSAs) is similar to display remarketing in that it is a cookie-based remarketing mechanism. RLSAs differ from display remarketing in that a user needs to both be on your list, and searching for your keywords in order to be targeted.
RLSA can be a powerful tool for securing engagement with customers who found you on the search network. These users are likely to browse some more before making a purchase, signing up or whatever your goal is. You can leverage RLSAs to ensure you are there when they search again - this is especially important for businesses with a high average order value and a correspondingly longer conversion path.
Your client runs an e-commerce website that sells men's apparel. To increase their chances of converting high-value customers who click on their ads but don't complete a purchase, they use RSLA with a bid adjustment of +30% to target users who have abandoned their carts. These users are shown personalised ads with an offer for free shipping.
It's crucial to ensure that your client's website appears in the top search results when these users search again. It's also profitable to pay extra to be there.
You can create RLSAs based on Website visitors (in relevant segments such as ‘all visitors’, ‘visitors to specific final URL’, ‘converters’, ‘cart abandoners’), customer emails, app users or Youtube users.
Don’t retarget users who are not interested
Users who bounced or spent a very short time on your page are probably not interested in what you are offering. To avoid wasting spend on retargeting these users with your RLSA campaign, exclude those who bounced or spent less than five seconds on your page
You can add your RLSA for observation or targeting. If you choose to target your RLSA, your ads will only be shown to users within the specified audience. Adding RLSA for observation only means your ads can also be shown to users beyond the list.
If you are using manual bidding you will have the option to add negative or positive bid adjustments on an observation segment, while a smart bidding algorithm will take the list into account as a signal when determining how much to bid in an auction for a query from a user on your list.
Consider pairing RLSA with DSA to target high value segments
You can pair RLSAs with audience targeting and DSAs to create highly personalised search remarketing for your most valuable segments. The benefit of this approach is to leverage the DSAs to avoid manual keyword work, while keeping targeting strict enough for it to be profitable.
To use this strategy efficiently it is important to only target your most high value segments, a broader targeting is likely to make this approach unprofitable. It is also recommendable to add your brand as negative keywords.
Deep dive into your historical data to find your most profitable audiences and add these to your search campaign. Make sure to set to ‘target’ to ensure you will only show your ads to this audience, and proceed to adding DSA ad groups - working with one ad group for all destination URLs or one URL per adgroup for easier performance comparison.
Dynamic prospecting and dynamic remarketing both bring together customer information and product information to combine ad tailoring and targeting. Both require access to a product feed to create personalised ads, which are then served either to previous site visitors, or new prospects.
Dynamic remarketing tailors your remarketing to the individual user, by showing previous visitors to your page ads containing the specific products or services they have viewed on your site, served in personalised messaging. This is a powerful tool that allows you to cover your full product inventory in your ads in display, performance max and app campaigns.
It is worth noting that a dynamic remarketing ad may contain several products of which the user has only viewed one - Google will use factors such as popularity to determine which products to pair up and serve in an ad.
Use dynamic remarketing with Pmax
Performance max without dynamic remarketing will be less powerful in its remarketing efforts than smart shopping was. Enabling your performance max campaign to use dynamic remarketing can help shore up your full funnel within that campaign.
Dynamic remarketing is baked into performance max when its set up through Google Merchant Centre, and your site is tagged for remarketing. If your campaign is not linked to GMC you will need to link a feed to your campaign from the ‘dynamic ad feeds’ field in the campaign settings.
Some advertisers chose to run a separate display dynamic remarketing campaign for better control of budget allocation to remarketing, however according to Google performance max will take priority over display dynamic remarketing campaigns that cover the same products.
Dynamic prospecting is a great tool for acquiring new customers. It catalogues your users according to both demographics and online behaviour, to find common features among users who have converted on specific products. It then looks for new users who share these behaviours and demographics to serve your ads to.
This analysis happens on the product level, Google will look for users whom it has enough data to make a specific product recommendation. In short, it draws on the vast bank of highly granular data Google has on all its users, to combine what it knows about a user’s behaviour and their demographics, and find high intent new users for your ads.
Think of dynamic prospecting as advertising to an audience of people completely new to your page, but within a silo of highly relevant lower funnel users, who are likely to be interested in one or more specific products you offer.
Use dynamic prospecting with standard shopping
Dynamic prospecting can be a great tool for e-coms not using performance max. You can pair it with standard shopping to leverage some of Google’s vast bank of user data to find new lower funnel users interested in your products, before they are even searching for it.
An advantage of this set up is maintaining control over how much budget goes into the display channel.
Dynamic prospecting is available for display campaigns, and is most likely part of the architecture of performance max when it has access to a product feed.
“Observation only” audiences does not limit targeting of your campaign to users within the selected audience segments, but it does go beyond only collecting data, as the name implies. Depending on your bidding strategy, you can use observation only audiences to:
- Gather data
- Add audience bid modifiers
- Signal for smart bidding
For manual bidding observation only audiences passively gather data and do not influence targeting, unless you add audience bid modifiers.
For smart bidding any audience added to the campaign will function as a signal for the smart bidding algorithm, meaning it will try to increase bids for the observation segments to see if it can find cheap conversions within it. If, however it does not find good value conversions, it will not keep bidding up on the audience.
Note that observation only audiences does not restrict the campaign to the added audiences, leaving it free to spend wherever it finds traffic, according to your other settings.
Use observation only audiences to gather information
Adding audiences as observation only can work great when you are looking to better understand your audience, and find profitable segments. It is useful to add new audiences as observation only to start with, and once the campaign has gathered some data you can add bid modifiers or even ad the audience as an exclusion.
You can add audience bid modifiers to any audience segment added as ‘observation only’. Use this to reduce bids on segments you know to be less profitable, and increase bids on segments with better performance.
A pet supply retailer uses bid modifiers in Google Ads. They decrease bids by 15% for audiences interested in "pet grooming tips" to save money. They increase bids by 10% for "premium pet food" to attract customers looking for high-quality pet nutrition.
Smart bidding algorithms take into account audiences which are added as ‘observation only’ to inform bidding and targeting - audiences added as observation only will not necessarily be targeted by your campaign: the algorithm will try to find converting users within them, but if not found might not bid especially for this audience.
To increase newsletter signups on your health and fitness blog, you create a segment for users who visited your site but didn't sign up. Using this segment as a smart bidding signal with maximise conversions, the algorithm tests your list by increasing bids on users within the list.
If profitable conversions are found, it will keep bidding extra on the segment. If not, bids will stop increasing or may even be reduced if the segment has more expensive conversions.
Add audiences to test their value
Add industry relevant audiences to test out audiences which might be of value for your business. Once you see how they perform you can chose to keep them in, letting them inform the smart bidding algorithm, or to exclude them, should they turn out to perform below average.
Resources & Further Reading
”Google Ads Audience Strategy: Targeting vs. Observation” by Stu Edwards
Audience signals for Performance max fall into the ‘observation’ only category, as they work similarly to smart bidding signals. As this is the only targeting type available for Performance max, however, there are specific strategies for working with audiences on this campaign type.
Audience signals in Performance Max are used to jumpstart a campaign, and continues to inform it throughout its lifecycle. The machine learning accumulated based on an audience in a Pmax campaign will continue to inform it, even after removing the audience.
It's important to keep in mind that audience signals for Pmax are only suggestions. They are used to steer the algorithm in the right direction, but not for actual targeting. It's like telling Google to "try over here." If Google finds profitable converters there, then it will, but the algorithm is not confined to your audiences and will go beyond them.
Focus on Audiences with the Highest Conversion Potential First
Audience signals can be divided into four categories, here listed from more specific to broader:
1. Converters (people who have already bought from you)
2. Your data segments (which include website visitors, add-to-carts, and initiated checkouts, etc.)
3. Custom segments and keywords
4. Demographic and interest-based targeting
Begin by selecting the most specific high-converting audiences and gradually broaden your reach, starting upwards on the above list and ensure you are feeding your Pmax campaign with quality first party data.
To effectively identify your target audience, segmenting them by asset group is crucial. Google doesn’t directly show you performance data on audience signals, it only shows you which asset groups are performing well through the asset group details and the listing group metrics.
The only way to find out which audiences are over- or underperforming is by segmenting your campaign into one audience signal per asset group.
Structuring your campaigns in this way additionally allows you to down the road separate high-performing audience signals into individual Performance Max campaigns, a strategy which gives you better control over your budget allocation and targets.
Spin off high performing campaigns from your Pmax
Google recommends structuring your Pmax asset groups based on sections, themes or categories on your website, adding several audience signals per asset group. However, having several audience signals per asset group makes it impossible to know how your different audience signals are performing.
Instead, segment asset groups by audience signals to get more insightful reporting, and to identify high performing audiences. It’s crucial to create the segmentation from the beginning of setting up the campaign to get clean performance data.
Once your campaign has run for a while identify well performing audiences, based on your asset group segmentation, and create a separate Pmax campaign just for the top performer(s), deleting the asset group(s) pertaining to the audience signal(s) from the original campaign.
Having the top performers on a separate campaign will allow you to restrict their targets to drive a higher ROAS / lower CPA and have control over the budget allocation between top performers and the rest.
Note that audience segmentation is just one layer in your Pmax set up, depending on the size of your account you may want run separate Pmax campaigns for each product type, and layer the audience signals segmentation on top.
Your client is a large apparel e-com retailer. You start with one Pmax campaign per product group: women’s clothing, men’s clothing, footwear, accessories. You segment each campaign with one audience signal per asset group.
Once the campaigns have run for long enough to gather some data, you analyse the performance of the different asset groups in your ‘footwear’ campaign and see the asset group with the audience signal ‘running enthusiasts’ has a ROAS of 700% compared to the campaign average of 500%.
You create a new Pmax campaign for footwear and add an asset group with the audience signal ‘running enthusiasts’, while deleting the asset group pertaining to this audience signal from the original campaign.
Your new campaign is likely to be able to get you a better ROAS than the original, and you can now both restrict the target to squeeze out better performance, as well as funnel more budget through this segment.
Resources & Further Reading
”How to Structure Audience Signals in Performance Max Campaigns” by Joan Porio at Solutions8
Exclusions can be just as useful as target audiences. They allow you to avoid spending budget on users or audiences that you know will not be profitable for your business, or are not within the audience you want to reach with a particular campaign.
You might want to exclude certain segments from one campaign, which you target in other campaigns, in order to control targets and budgets with more granularity.
Exclude converters and bouncers
Exclude your converters if you are generating leads or selling a product that people tend to buy only once, leverage your RLSA as an exclusion for these campaigns and business types. In the case of remarketing campaigns you might do well by excluding converters, as well as users who spent less than 10 seconds on your page, and those who have visited your page multiple times without converting (depending on the average length of your sales cycle).
A pension provider has a free calculation tool on their site, as well as a sign up form for capturing leads. Once a user has filled in the sign up form, it is probably not profitable to keep targeting them, as the marketing efforts now move into following up on the initial lead. Likewise if a user has visited the page 15+ times (to use the free calculator tool) without converting, they might not be worth investing more adspend on.
Resources & Further Reading
”4 Google Ads Audience Exclusions You Need to Try Next” by Joe Martinez