This is the subhead for the blog post
Encountering sudden changes in the performance of active Facebook ads is akin to finding potholes after a long hard winter in Chicago. If you’ve run ads for any length of time, you know that sometimes things don’t go as planned. And what’s worse is that when things do go wrong, it can be hard to identify the actual issue.
How do you really know beyond “gut feeling” that you’re running up against audience fatigue or that you have audience overlap in the account? There wasn’t a concrete answer to those problems until now…
Introducing Delivery Insights
Facebook has built a new tool, called Delivery Insights, specifically to address these questions and to help us identify the root causes of dips and dramatic recent changes in ad performance. With Delivery Insights you can get deeper insights and data about your ad delivery, to help you maximize the results you get from your ads. But like every tool, it’s only valuable if you truly understand how it works.
In this post, I will take you through the basics of the tool and how to read the data it provides. I will also show you a few scenarios where Delivery Insights may be especially helpful, as well as provide cautions about using it incorrectly.
The main view is comprised mostly of the timeline and notifications sections. Delivery Insights, unfortunately, currently only shows you the last 7 days at any given point. This is what makes it useful in diagnosing RECENT issues. Thus, it may prove useful to check every couple of days to identify issues or potential problems.
The timeline area allows you to graph three items: Amount Spent, Your Goal (Conversions and CPA), and Impressions. These three items should be all the information that you need to get a gauge on where the potential problem exists. This main timeline view is extremely helpful in quickly identifying potential problems or sudden changes that you may have missed in your day-to-day operations. For example: the timeline can show you if there was an increase in CPA over the last couple of days with a spike yesterday. Being vigilant in checking the timeline should be your first step in checking, identifying, or diagnosing a potential issue.
The second area of the main view is for Performance Detail Notifications. If Facebook notified you about a potential dip in delivery or performance, it will be shown in this section. You have the ability to then dig deeper into the affected ad set.
The bottom section of the Delivery Insights tool is where the real gold lies. Here you’ll find 3 specific sections to help you take a deeper look at account changes and their effects, overlap, and saturation:
This section should look fairly familiar to Facebook advertisers. The same data here can also be found in Ads Manager by selecting “Account History” from the drop down at the campaign level.
To help give easy access to activity history, Facebook has included change logs on each individual campaign, ad set, and ad, as well as in Delivery Insights. You’ll also see the changes reflected in the timeline discussed above if you’ve checked the “Show activity” box at the top of the timeline when you first visit Delivery Insights.
One of the many items that can drastically affect overall delivery on Facebook is the amount of audience overlap. This can cause Auction Overlap, which occurs when audiences from the same ad account appear in the same auction. Although Facebook already prevents you from competing against yourself through a process of de-duping, having audiences compete in the same auction can affect distribution of your ads. If the same audience is found in multiple ad sets from the same overall ad account within the same auction, Facebook says they will simply choose to give the higher-performing ad set the delivery. This can result in an ad set with high auction overlap struggling to get distribution because of a high level of de-duping on Facebook’s end.
Within the Audience Overlap section, the first area to look at is “Amount Spent.” You should check to to see if the account is indeed spending as planned, and how close it’s spending relative to the daily budget. This may be your first sign that something is wrong.
The second section to pay attention to is the Auction Overlap Rate. This is the rate at which the ad set you’re viewing is in the same auction as another ad set in the same ad account. (Thankfully, Facebook includes the ad set ID for each ad set, and a link that takes you directly to the ad set in Ads Manager – making it easy for you to work with the specific ad sets you’re investigating).
The numbers shown here in Auction Overlap Rate are the percentage of people being removed from the auction because of the de-duping done by Facebook. In the screenshot above, you see that there are three total columns for the three ad sets providing the most auction overlap for a given day. Now, I want to make it abundantly clear: a high percentage in these columns isn’t necessarily a bad thing. The Auction Overlap Rate shows the total percentage of overlap, but in each of the three ad set columns, the percentage indicates how much of the overlap is caused by the current ad set. So you may have an audience overlap rate of 2.5% but ad set 1 has a percentage of 89.9% and ad set 2 has a percentage of 11.1%. This is simply saying that 89.9% of the overlap is caused by ad set 1, but there is only 2.5% total overlap.
Audience overlap isn’t always bad either. Consider this: you are targeting 3 large audiences and you see that you have a large audience overlap rate, but the budgets you set are rather small. In this specific scenario, you’re much less likely to have competition and see a dip in overall performance. But when you double or triple budgets, you may see a dip in performance or a spike in CPA that is probably, in part, to the audience overlap rate.
The next and final tab along the bottom is Audience Saturation. This measures how often people are seeing your ad for the first time, along with a few other things. This section is extremely helpful in gauging overall ad or audience fatigue. There have been times where I’ve noticed a very gradual decrease in performance. It would feel as though nothing I did would improve CPA or overall conversion volume. I had a hunch it was audience fatigue, but before Delivery Insights, it was impossible to know for sure. This is certainly a stat to keep an eye on and maybe even track weekly or monthly.
Most of the columns in this section should have fairly familiar stats, so we’ll just talk through “First Time Ratio” and “Audience Reached Ratio.” The “First Time Ratio” simply shows what percentage of the total daily reach saw an ad from the ad account for the first time. Like many of the stats in Delivery Insights, there’s no “ideal” here. It all depends on your goal. I would say for the most part, you want a higher percentage indicating you’re constantly reaching new people, especially in an upper-funnel campaigns.
The next column to mention is the “Audience Reached Ratio.” This is the percentage of the potential audience you’ve specified within the ad set which you’ve reached so far. Again, there isn’t an ideal percentage here, but generally speaking you may want a it to be smaller for higher-funnel campaigns and much higher as you travel down the funnel.
Delivery Insights is an absolutely fantastic tool and one that should be welcomed by any Facebook ads manager. However, I want to offer a bit of caution. The intention of the tool is to research and identify potential performance issues. It can be extremely easy to jump in and find issues that don’t really exist. You may see a high percentage somewhere and want to make immediate changes. I highly encourage you not to create problems that don’t exist. Delivery Insights’ singular purpose is to help you identify the root of a problem you’ve identified in performance.
I want to also warn against comparing percentages across the tool. It may be handy to keep track of running percentages on an account or ad set, but everyone’s accounts, goals, and performance will be unique. Stay focused on the performance and metrics that matter for you.
Last, and certainly not least, is context. Context is absolutely king here. You should be intimately familiar with your account; understanding that context will go a long way in using Delivery Insights. This is the main reason you shouldn’t be comparing everything in the tool to another account entirely. Use the context and knowledge of the account to help you further understand what may be causing a lull in performance.