Should you trust the Facebook algorithm? Broad vs Manual Targeting

One of the best features that have brought more and more paid social marketers towards Facebook is audience targeting. Having the ability to target specific types of people at a detailed level, by selecting interests and brands that connect well with your target audience, is the core of what built Facebook’s ability to turn over £20bn+ in advertising spend throughout 2020.  

The ability to decide who exactly your ads are being seen by at an interest level is an element of power that can work as a great starting point to find the right types of people when you have a new product or brand. 

However, there is certainly more to consider when fitting Facebook and Instagram paid advertising into your marketing mix.

Using Facebook and Instagram to target audiences

The Facebook Pixel is installed on close to 10 million websites scattered across the web. All of these websites send user activity and data back into Facebook’s algorithm. Trillions of data points feedback into the algorithm daily, giving Facebook a view of the people who engage and ultimately convert with these businesses.

The pixel tracks individual phrases and terminology which is interacted with by the user on-page. This can then be used to find audiences beyond anything that you can select at ad set level when picking your detailed targeting. It is a lesser-known power but this makes it possible to target people by using certain words in your creative. 

Instead of detailing out each of your audiences, run broader audiences that house more data. Just because you’re using an open targeting audience doesn’t mean Facebook isn’t going to find lookalikes within it anyway. 

It works the same from an audience overlap point of view. With broad audiences, Facebook is smart enough to avoid the same set of people within each version without you having to split everything out and tell it to do that for you. Going granular will only hinder your performance as you restrict the algorithm from selecting people who could convert, even if you don’t think they will. If the algorithm doesn’t think they have a chance, the ad won’t serve. 

In search, you prod and probe around finding the sweet spot in terms of the bid to get as low a cost per click (CPC) as possible among people who are already in the market. In contrast, Facebook and Instagram are experiential channels. These platforms, therefore, need to be treated separately from demand-based channels such as search. This means that getting ads in front of as many people as possible is the main piece of the puzzle. This will open you up to find those conversion opportunities within the feed. The cheapest way of doing this is to get your cost per thousand impressions (CPMs) down as far as possible by centralising data and going broad. It’s that simple.

The Facebook algorithm is always ready to learn

Facebook at core runs from a big data algorithm which, like any other algorithm, means it needs to be taught how to achieve what you want to get out of your ads. A common mistake lots of advertisers make when advertising with Facebook is to split out their account structure into incredibly granular detail, fragmenting data into individual campaigns and ad set strands. This may seem logical, as it allows you manual spend control over every element you could want. 

The reality, however, is that you will never be smarter than the machine. Facebook works optimally when your data is centralised into a few campaigns, with a couple of ad sets containing ad variations. Centralising your data allows Facebook to learn, gathering data quickly to get through a learning phase and deliver results as quickly as possible.

Why you should trust the Facebook algorithm to deliver results over manual audience targeting

The algorithm is smarter than you and there is no way past that. While you may know your customers from a demographic and interest affinity point of view, the pixel understands the data your customers are leaving behind with them and simply goes off to find more people just like them. Quite rightly, you can give this a gentle nudge in the right direction with some very basic demographic and audience targeting but make sure that your audiences are broad enough to let Facebook do its thing as no other platform does. 

Facebook knows your customer journey, and that it takes 4 - 5 clicks to get to the checkout screen. You taught Facebook this when setting up your conversion events. Now the algorithm knows that it needs to find people who have converted after 4 - 5 clicks. Facebook knows who is clicking on a call to action button similar to yours, so now it knows to focus on and be more aggressive in serving ads to those people who have clicked on a ‘Checkout Now’ button.

For example, Facebook scans your landing pages and realises that you are advertising a specific flavour of your soft drink brand.  It will then find people who have hit other websites with the pixel installed and browsed items with similar flavours.

Final thoughts

Marketers need to trust the algorithm and use it to their advantage. Doing so can open up your campaign to audiences you may not have ever considered, allowing more opportunities for conversion and growth.

We’re big on innovation and have been experimenting with this technique to test its effectiveness for our clients.