for brands on a mission
Do we really trust the algorithm? How Facebook’s latest updates will impact your campaign performance

With a number of new features being rolled out in 2019, this seems to be the year of change for Facebook. From algorithm changes, to encrypted private messaging and its crackdown on inauthentic content such as purchased likes, for better or worse the platform is changing based on the consumer’s evolving needs.

One of their most recent announcements is the removal of manual budget allocation between ad sets. This means Facebook will only allow you to set a minimum and maximum campaign spend and rely on its algorithm to shift budget between ad sets to drive overall campaign results. One would assume Facebook knows its own platform the best, but there are a number of reasons why this change will fundamentally impact the way brands design their paid campaigns.


This change will fundamentally reduce your campaign control. With the inability to manually shift spend between your campaign ad sets, you can’t decide where budget will be allocated. This is a positive for the less-experienced user who spends a small budget to expand on page followers or drive additional leads. Without the time to upskill yourself on identifying positive campaign metrics and optimising budget accordingly, Facebook’s algorithm will now carry this responsibility. Don’t be fooled: Facebook is definitely not an easy tool to understand, and if you’re not spending a few hours a day learning, wrong decisions can be made. Taking away this type of users’ control can limit the chance of error and mean better business results with less time spent.


On the flip side, reducing control for the more experienced user could cause headaches. For example, if positive campaign indicators highlight a potential opportunity, you’re not allowed to optimise ad set budget to further enhance performance. This leaves the onus to Facebook’s budget algorithm. If you’ve used Facebook’s campaign budget optimisation feature before, it is anything but flawless. When the feature is switched on, the algorithm will automatically allocate budget between ad sets based on the set objective and optimisation goal. This leads to certain ad sets receiving more spend based on the algorithm’s campaign learnings.

Like all AI powered tools, results improve over time as the machine optimises based on campaign data. The more data the machine is fed, the higher level of campaign performance. We tested this feature last month and discovered that even though the overall campaign performance was good, there were ad sets which were being neglected. This lead to less budget being spent on superior ad sets and worsened the overall campaign cost efficiency. This is where the frustration lies. Though many people think that artificial intelligence will be making huge business decisions in the future, at the moment, human intervention is still essential to ensure the best ad set is getting the majority of the budget.


If you’ve worked on a number of different Facebook campaigns at the same time, you would know that optimisations can be quite time-consuming. An experienced Paid Social Specialist knows that optimising is not only switching adverts off or distributing spend among ad sets; a significant amount of time should be spent on campaign analysis. This can include identifying audience micro-segments that are driving a high level of performance or understanding insight which highlights why one creative is outperforming others. Spending just a few hours a day analysing your campaign can help boost overall performance. It can also help uncover segments which could be hindering your campaign’s success. For example age or gender specific segments that are driving a high cost per result. So with the new feature allowing less time spent on day-to-day optimisation and allowing more time for analysis, this may lead to better overall campaign performance.


When setting up a campaign for the first time, Facebook usually recommends splitting the adsets based on different testing variables. This can include device, age segments, gender, placement and many more variations. This campaign structure helps you gather insight on segments which drive a high level of performance based on simple A/B tests. If one ad set is driving a very low cost per result, you can then allocate budget accordingly.

With the implementation of these new changes, this still can be done. However, you lose control on the adset level and can no longer allocate more spend to one adset manually. An alternative is to create multiple campaigns, targeting the same audience but segmenting each campaign based on different testing variables. This still allows you to conduct your own tests, however this will leave the account quite messy and incredibly tough to manage with so many campaign variations.


Ultimately, we need to ask whether Facebook knows its own algorithms as well as it thinks? Daily optimisations wouldn’t be needed if this was the case. In a perfect world, the campaign is switched on and spend will be allocated based on the algorithm’s learnings.

Unfortunately, this is not a perfect world and this can lead to budget being spent in the wrong places. Sure, they created the platform and have been making tweaks and improvements to the algorithm for years. But do we trust Facebook enough to control our campaign budgets and to allocate the correct amount of budget to the best performing ad sets? Based on what we have seen so far this can lead to average performing ad sets receiving large amounts of spend and the better ones left neglected. This is definitely the question a lot of you will be thinking when launching some of these campaigns for the first time.

We recommend monitoring each campaign closely to make sure the spent budget is achieving the best possible results. Only time will tell whether this new feature will attract the right kind of publicity for Facebook, or whether it will still take a few years to perfect. No doubt, this feature will eventually bring value to businesses around the world, but it may take a bit longer than expected.

By Social & Digital Specialist, Gidon Jacobs