This article provides everything you need to know about lookalike audiences. From the basics of how to effectively seed lookalike audiences to how lookalike audiences work and how you can profit from the technology.
- What Is A Lookalike Audience
- When To Create A Lookalike Audience
- Tips For Building A Lookalike Audience
- Facebook Lookalike Audience
- Google Ads Lookalike Audience
- LinkedIn Lookalike Audience
- How Does Lookalike Audience Technology Work
- Conclusion & Future Developments
What Is A Lookalike Audience
A lookalike audience is a segmentation of a social network or advertising networks user base which matches criteria specified by the advertiser.
In practice this works by providing a custom audience as a seed and then the social network using the data to go out and find similar users that have things in common with the custom audience.
Lookalike audiences allow advertisers to expand and scale campaigns beyond what traditional targeting methods provide. The technology provides the ability to go out and find people like our existing customers.
When To Create A Lookalike Audience
Lookalike audiences can be built as soon as you have enough data to provide the algorithms with meaningful insights into who they are searching for.
Networks will have a minimum limit on the size of a custom audience before it can be used as a seed for a lookalike audience, this is generally either 100 or 1000 users. However the more data you have the more effective the targeting. Facebook has a minimum seed audience size of 100 but recommends between 1000-50000 users for accurate targeting.
The quality of the data you provide will effect the targeting of the lookalike audience. If you build a custom audience from 10000 website visitors this might not be as targeted as 100 customers because visitors don’t have the same purchase intent as existing customers.
Ideally you want as much relevant data as possible. Building an audience over time and increasing the efficiencies of the lookalike algorithms to go out and find like minded users for your advertising campaign.
Tips For Building A Lookalike Audience
Plan Your Use Case
Start with what the lookalike audience is going to be used for. Are you going to be running content marketing campaigns or direct response shopping ads?
This will make a significant difference to how you want to seed your list. A customer list would work great for both whereas content marketing might work well with a 30 day website visitor custom audience.
Capture Data From The Start
Even if you aren’t planning on using custom or lookalike audiences for a campaign make sure you have tracking pixels setup from the start on landing pages and web properties.
This will provide you with back-data that can be used if you want to expand into these areas a later date.
Use Email Lists
Facebook and Google Ads (min spend requirements) both allow custom audiences to be built from email lists. This means you can upload a customer list to generate a lookalike audience of those customers.
It can also be used for building lookalike lists from other email lists such as sales team contacts & prospects or a list of emails within a specific organization. This allows for a lot of creative targeting options but check terms with the social/ad network to ensure data and privacy compliance.
Start Small
Facebook ads offers the option to generate 1%-10% lists. 1% lists will be the most targeted and it’s a good idea to start here and then expand outwards. The higher the number the lower the matching threshold which usually means lower ROI on campaigns.
Other networks often have “expanded reach” options which it’s often a good idea to untick to start with to reach the most targeted audience for the initial test campaign.
Combine With Additional Targeting
Lookalike audiences can be further targeted with standard demographic, interest and behavior methods. If a lookalike campaign isn’t profitable as is, provide a helping hand to the algorithms by setting some additional targeting options. This will reduce volume but increase ROI if the targeting is effective.
Start With Tight Budgets
Even when scaling up it’s a good idea to start lookalike campaigns with capped daily budgets. The campaigns will convert completely differently compared to your existing targeting. More often than not in my experience they convert well but it’s still better to test it first with a daily budget in place.
Facebook Lookalike Audience
Facebook was the first to release this technology and is still the market leader in the field.
- 100 minimum seed size
- Build directly from custom audiences
- Generate custom audiences from pixel/email/shared data sources
- 500 lookalike audiences per custom audience
- Users in seed audience excluded from lookalike audience
- Target lookalike audiences at ad set level
More information is available here on the Facebook Adverting Network.
Google Ads Lookalike Audience
Google’s lookalike technology is called “similar audience”. It can be found in the audience manager where profiles are automatically generated.
For a custom audience to be considered for generation of a similar audience it needs to meet certain criteria. Some factors that are taken into consideration.
- Number of users on the list, minimum is at least 1000 users
- When these people joined the list and how fresh the data is
- Similarity of the users search and browsing behaviour
- Adjustments made to the original list since it was created (I’m unsure exactly what this one means or is referring to. It sounds like a loophole being blocked to be honest)
Original list users are excluded from similar audience list.
Google looks at both search and browsing behavior (via display network traffic indicators) to match users.
Similar audience list will take 2-7 days to generate after the seed audience list is defined.
More information is available here on creating Google Ads.
LinkedIn Lookalike Audience
- LinkedIn requires a seed audience of at least 300 Users
- Lookalike audience will be generated up to 15 times the size of seed audience
- Modifications to seed audience wont update existing lookalike audiences
- Lookalike audience will take 48 hours to generate and an additional 24 hours to start delivering ads
- Users are matched on member and company similarities
- Only active members that have recently logged in are added to lookalike audiences
How Does Lookalike Audience Technology Work
Ad networks make out that they are using skynet AI technology for their lookalike audiences which is probably true to some extent but for advertisers it’s actually quite simple to understand.
- The ad network will set out a list of data points such as: Age, sex, location, interests, behaviors etc.
- A baseline will be generated from the total user base i.e. 31 year olds account for 2.145% of the Facebook population. 20.543% of the Facebook population likes cat memes.
- The advertiser provides a custom or seed audience of users.
- Each data point will be compared to the baseline looking for anomalies and patterns.
- This data matching is optimized using machine learning algorithms to identify what data points are most important to campaign performance.
Conclusion & Future Developments
Lookalike audience advertising provides a huge opportunity to use this technology to expand the reach and volume of existing campaigns.
It offers a scaling opportunity to tap into new users who will convert well based on the similarities to existing customers or website visitors.
The technology will continue to develop and we will see this being rolled out eventually across all major ad networks and social platforms.
In the future we could see more data and audience sharing with marketplaces for pixel placements popping up to leverage 3rd party data.
Get creative and use lookalike audiences to scale up your ad campaigns.
If you are interested in learning more about online marketing then check out the free digital marketing course.
If you’ve enjoyed these resources could you help share this content on social media and send it to anyone who you think might benefit from it.
Thank you.
