Whether you like it or not – all in a sudden Facebook Inc. is now Meta Platform Inc. Though it may take some times for Facebook to become the Oasis (The immersive social network in the movie “Ready Player One”), Facebook has deployed AI in managing user engagement for years already.

As a marketing company working closely with Facebook team, Chilli’s teammates do have tips and tricks particularly in social media advertising which can significantly increase its effectiveness. Yet what you may not know is that machine learning is playing a vital role in improving targeting.

How to get the most from Facebook Ad’s sophisticated machine learning features? Here are some of them which you can try out on the fly:

Avoid Too Many Edits of Ad Set

Each time an ad set edit is made, the learning phase restarts. This can be costly for driving optimizations. To mitigate this, minimize all ad set edits to ensure that a campaign does not restart their learning phase.

Allow Automatic Placements or 6+ Placements

Automatic Placements gives the flexibility to determine which placement combination drives the best results. More practically, within each ad set, the delivery system will find the most efficient placements and double down on the winning placements automatically. Both Automatic Placements and 6+ placements are defined as solution best practices to provide the liquidity for optimal campaign performance.

Broadening Audience Targeting

Audience liquidity can be increased by targeting the broadest possible audience. The 3 options to do so are broad targeting, Custom Audiences and lookalike audiences. Broad targeting (demographic and location) targeting, detailed targeting and interest targeting are not based on conversion or event data and are not impacted by the users opting post-enforcement.

Allow Time Before Analyzing Performance

Give a campaign at least 72 hours or wait for the full length of the optimization window selected before evaluating performance for conversion-optimized campaigns. Given the nuances of delayed data and modeled reporting, daily assessment is still possible for campaigns, but waiting can help you get a more accurate picture of performance.

Niche Targeting

If you’re targeting narrow or small audiences, we recommend broadening audience size. Audience liquidity can be increased by targeting the broadest possible audience. Increasing the target audience gives the ads auction and algorithm more opportunities to find converters, lowering costs and exiting the learning phase. If you don’t see a lookalike audience or more targeted audiences perform better than broad audiences, stick with broad targeting only.

Give A/B Test A Try

A/B Test will split your target into sub-groups and Meta will keep track on the performance on both and advise advertiser focus on more effective one. The reality, well… may challenge your understanding about your audiences.

Sounds good or sounds strange? Why not give us a call and let us walk you through the details.