NEW Facebook Feature – Value-Based Lookalikes
Adapting is Leading
Facebook marketing is constantly evolving, as new features and updates are added on a weekly basis. Missing out on even the smallest opportunity to ride an upcoming trend can set you back for months. As Facebook marketing experts, yellowHEAD has set the goal to seize these opportunities and become the industry’s earliest adopters.
Introducing Facebook Value-Based Lookalikes
One of Facebook’s most powerful offerings for marketers is Lookalike Audiences. The Value-Based Lookalikes new roll-out is taking that feature to the next level and letting you assess current and new customers by how valuable they are to your business.
According to Facebook, “LTV is a value associated with your customers based on how much and how often they spend with your business over the course of their relationship with you.” By telling Facebook how much your current customers are worth to your business in terms of Lifetime Value, their smart algorithms can help you find more people who are similar to your customer base.
Here’s how to create a Value-Based Lookalike Audience:
- In Business Manager, go to Assets → Audiences
- Upload your Custom Audience in the “Add customer file” stage (add a name and description for your convenience)
- Select your Customer Value from the dropdown menu under “Choose customer value”
- Under “Edit Data Mapping”, preview and map your data and then click “Upload & Create”
- You will reach the final stage after your Custom Audience is uploaded, and then you just need to click “Create Lookalike”
- To complete your LAL audience, choose your desired Location and Audience Size, and click “Create Audience”
To get started with your first Value-Based audience campaign, Facebook suggests the following test:
- Normal LAL — For the control LAL campaign, the seed audience should be the customers who the advertiser would have typically uploaded for a normal LAL audience. This is usually the advertiser’s highest valued customers. Seed Audience Minimum: 10K.
- VBLAL — For the test campaign, the seed audience for Value-Based LAL should include a wider representative base of customer quality. This means that advertisers who only use the highest value customers for their LAL seed should begin to include medium and lower value customers in their VBLAL seed. The VBLAL ideally will include the same users in the LAL seed + lower and medium valued people + the numeric value attached to them all. Seed Audience Minimum: 20K.
Both campaigns should have the same budget, in order to properly compare performance.
It’s preferable to use the same creative, and only one, within each campaign. This eradicates the problem of delivery skewing on a specific creative which can confound what drove the test’s results.
A few more tips from Facebook…
- Use dollar values only. Assign a dollar value for each customer and don’t include ratings or rankings, for example.
- Include a full range of customers, from low to high value. This allows Facebook to be able to “hone in on what might distinguish an average customer from a great one.”
- Don’t use negative values to signify undesirable customers. Facebook won’t count those.
- Make sure you’re using the same currency throughout the list. Facebook will otherwise assume you are using the same currency.
- Use decimals for cents, but no other punctuation.
We first tested Facebook lookalike audiences for one of yellowHEAD’s clients, a top gaming app. Since its launch 4 weeks ago, this new feature drove outstanding results for the client, reaching new people similar to high-ROI customers and leading to better campaign results at a lower cost.
Below you can find the results of Value-Based lookalike campaigns vs. other quality spenders lookalike campaigns, broken down by OS and GEO.
Client’s Performance KPI: D7 ROI > 25%
Android D7 ROI performance increased by 311%, whereas iOS D7 ROI performance increased by 169%.
We can see a significant improvement of Average D7 ROI for the tested GEOs.
Bottom line, the Value-Based Lookalikes performed much better than all other quality lookalikes in the past 4 weeks in terms of D7 ROI, which is our main performance KPI.
An additional interesting test would be generating Value-Based lookalike audiences for each region, as we know that different regions have different user characteristics, and we should address each audience specifically.