Is Your Facebook ROI a Lie?

Seth Newtonby Seth Newton

Home | Blog | Is Your Facebook ROI a Lie?

Share this article

 

For E-commerce advertisers ROI is the name of the digital marketing game. Paid search engines like Google and Bing provide robust conversion tracking and clear performance reporting on PPC campaigns. Thus, advertising ROI is readily ascertained. And if you’ve expanded into Facebook Advertising you’d like to think Facebook provides PPC advertisers the same level of transparency… but do they? Are you sure that your ad spend is really creating the sales reported in the Facebook UI? Based on our interaction with clients and other PPC professionals, I have a feeling some readers might be surprised.

Especially since Facebook recently launched an upgraded tracking pixel, advertisers might readily believe Facebook has finally stepped up their game. While it’s true that real strides have been made, the reporting still has a flaw that is potentially misleading advertisers. Below I’ll lay out the problem and how to solve it.

First I’ll note that there are campaign objectives and performance metrics within Facebook Ads that are more oriented around efforts like branding, increasing social engagement, and building an audience. This article is geared more towards online retailers looking for direct ROI from sales generated by Facebook Ads. So the focus is on conversions (sales) and revenue, rather than metrics such as page likes and post shares.

 

The Problem

Within the Facebook Ads reporting platform, there is one column for conversions called “Purchases.” The problem is that within this metric column, data is lumped together for both click-through conversions AND view-through conversions. On that same note, there is one column for “Purchase Conversion Value” a.k.a. Revenue. This column also contains (revenue) data from both click-through AND view-through conversions.

For those of you who don’t already know, here’s the difference between the two conversion types:

  1. Click-through conversion: Someone clicks on your ad and makes a purchase (within the attribution window – FB default for click-through conversions is 28 days)
  1. View-through conversion: Someone is served one of your ads, but does not click on it, and later makes a purchase (within the attribution window – FB default for view-through conversions is 1 day)

 

Why it’s a Problem

There is a fundamental problem in combining data for two different events of unequal value. You’d be hard-pressed to find a PPC Professional who thinks click-through conversions and view-through conversions, as they are reported, are equally meaningful. I think it’s safe to say that most of the PPC community would agree that the click-through conversion metric holds a higher value. There is no doubt that the ad was seen, clicked on, and drove the sale.

 

no doubt

 

That’s not to say that view-through conversions aren’t meaningful. In many cases they are. But the conversion path becomes a little cloudier. Even though Facebook only counts impressions that occur above the fold, that doesn’t necessarily mean the user consciously saw and processed the ad. In other words, there’s room for doubt that the impression drove the sale. So, as marketers, we need to see the two metrics separately.

 

Why it’s a Big Problem

The major issue with Facebook combining these conversion types really comes into play with remarketing campaigns. I’m of the opinion that view-through conversions from a re-targeted audience should be taken with a grain of salt. This concept transcends Facebook, as it also applies to Google Adwords and other networks. To be clear, if an ad impression truly causes a conversion, remarketing or not, it’s as meaningful as any other conversion. The issue is you don’t know which impressions really drove the sale and which ones, in reality, did not.

If you’re confused as to why we should be skeptical of view-through conversions in remarketing campaigns, all you have to do is look at the numbers… and then think about how the process works.

Lets start with a Google Adwords example. I’ll reference an account running 2 types of display network campaigns:

  1. A (Non-remarketing) Display campaign targeting users via keyword contextual targeting
  2. A (Remarketing) Display campaign re-targeting users who have previously visited the client’s website.

The below image contains data from these 2 campaigns:

google sshot vt vs ct blog

 

You’ll see that the Display – Remarketing campaign reported 110 view-through conversions, while the display – keyword(non-remarketing) campaign, despite having over 790K more viewable impressions, reported only 1 view-through conversion. That’s right, 1 whopping view-through conversion.

Yes, remarketing is almost always going to perform better than other types of targeting. But should we just conclude that remarketing does that much better in the way of view-through conversions? Consider these 2 data points:

  1. The remarketing campaign’s click-through conversion rate is roughly 2x higher than the non-remarketing campaign’s click-through conversion rate.
  2. The remarketing campaign’s view-through conversion rate is about 180x higher than the non-remarketing campaign’s view-through conversion rate.

Theoretically, the difference in click-through conversion rate between the two campaigns should be somewhere in the same neighborhood as the difference in view-through conversion rate. But these 2 aren’t even on the same planet.

Similarly, you can also look at the click-through conversion to view-through conversion ratios within each campaign:

  • Non-remarketing = 40:1 click-through to view-through conv
  • Remarketing = 69:108 (or 23:36) click-through to view-through conv

Again, it’s reasonable to expect that these ratios would be somewhat similar. This brings real doubt as to the validity of the view-through conversions as they’re reported. So to believe all reported view-through conversions in this remarketing campaign were driven by the ad impression makes no sense statistically. From this we can determine that view-through conversions in remarketing campaigns are inflated as reported, and are not truly driving all these sales.

The reason for including the Adwords example is to illustrate what could be happening in Facebook on a smaller scale. Now it’s important to note that Google Adwords uses a 30-day view-through conversion window while Facebook only uses a 1-day window. Credit should go to Facebook for keeping this window tight. The discrepancies aren’t quite as glaring, but we can still see similar data anomalies in Facebook.

Check out the 2 Facebook Ad Sets below. One is retargeting to all website visitors, while the other is targeting friends of people who have liked the business’s Facebook page (not remarketing). Both ad sets are mobile specific.

fb ss 1

You’ll see the remarketing ad set has 94 conversions at a conversion rate of 17%, while the non-remarketing campaign has 6 conversions at a conversion rate of 4%. Again, we’re not comparing apples to apples because the retargeted audience has indicated product interest by visiting the website before. Meanwhile, the other audience is simply a demographic set that we believe is worth targeting. So it makes sense that Remarketing is performing better. But is it really performing that much better?

Like the Google example above, the ratios just don’t jibe when you break down the conversion types. Of course, I mentioned before that Facebook doesn’t report the data for the two conversion types separately. And they don’t, by default. But there is a not-so-easy-to-find way to segment the data. I’ll explain how to get there. But first, let’s look at the breakdown:

fb ss 2

*Disregard the $0 revenue in the non-remarketing campaign. This client offers free samples, and it just so happens that all conversions were for free samples in this date range.

Now you see 2 columns for conversions and 2 columns for revenue. “1 day” represents view-through data, and the “28 days” represents click-through data. Notice the difference of the ratios:

  • Non-remarketing = 5:1 click-through to view-through conv
  • Remarketing = 46:48 (or 23:24) click-through to view-through conv

While the difference isn’t as glaring as with Adwords (because of the different attribution window), it’s still more than enough to be classified as suspect. These numbers go against logic and create doubt as to whether the impressions really drove the sales. And doubt is a no-no when you’re calculating ROI.

So why would view-through conversions in remarketing campaigns be inflated?

One theory is that customers were already on their way to making a purchase and got served a remarketing ad along the way. For example, picture this scenario:

A young man goes to a website to buy something, but before he can checkout… he caves into the urge and, against his better judgment, takes a peek at Facebook to see what his ex-girlfriend has been up to. If you’re re-targeting to website visitors, your ad might be served to this poor guy while he’s torturing himself on Facebook. Whether he actively sees the ad or not, it’s NOT the driving force behind the sale. Yet, you better believe it will be counted as a view-through conversion.

 

office cry

 

Regardless of the scenario, we have to be careful how we account for view-through conversion data from remarketing campaigns when trying to calculate ROI. The Google scenario is ok though, because Adwords segments the conversion types for us. The data is separated and we can easily analyze performance. You have to dig a little deeper with Facebook.

 

How to Solve the Problem

So how did I generate that breakdown in the Facebook UI? Below is a step-by-step:

  • Navigate to the campaign view within the FB ads manager and click on the Columns button.
  • Click Customize Columns in the drop down menu.

Screen Shot 2016-04-08 at 3.29.43 PM

  • Click Change Attribution Window in the bottom right corner.

Screen Shot 2016-04-08 at 3.37.53 PM

  • By default, no boxes will show as selected. Tick the box for 1 day under AFTER VIEWING AD, and tick the box for 28 days under AFTER CLICKING AD.

Screen Shot 2016-04-08 at 3.40.02 PM     Screen Shot 2016-04-08 at 3.40.23 PM

  • Click Apply.

As long as you’ve added columns for Purchases and Purchase Conversion Value… you will now see 2 columns for Purchases and 2 columns for Purchase Conversion Value. You can hover over the “i” information icon to see which metric the column is displaying – #Days after viewing or #Days after clicking. It’s sort of a strange way to display it, but this is basically saying ‘view-through conversions’ and ‘click-through conversions.’

fb ss 2

Note: You can play around with this. For example, you could only tick the box for 28 days under AFTER CLICKING AD, and you would see just one column for each that displays click-through conversion metrics only.

Now that you have the data separated, you can get a more realistic idea of the ROI Facebook Ads are yielding.

Looking back at the Facebook example above, you could calculate 2 ROI values from the remarketing campaign:

  1. Both conversion metrics together: ROI = 22.01
  2. Only click-through conversions: ROI = 10.3

In this example both present a seemingly good ROI, but those 2 numbers are staggeringly different. Depending on the advertisers return goal, this could mean you’re losing money while you think you’re turning a profit. That’s not good. Regardless of your opinion on view-through conversions, it’s still ideal to see this data broken out.

 

Additional Considerations and Tips

If you use the Facebook setting to optimize for conversions, then Facebook will include view-through conversions in their algorithm. You might want to consider optimizing for clicks, even if your true objective is to get more conversions.

Also, you might want to consider keeping remarketing audiences in dedicated campaigns, separate from non-retargeted audiences. On top of that, we recommend a streamlined naming convention that clearly indicates which campaigns are remarketing and which ones aren’t. This will be very helpful when analyzing performance statistics.

Lastly, adding utm tags to your destination urls in Facebook Ads is highly recommended. This way you can observe FB stats in Google Analytics. Keep in mind that Analytics uses last-click attribution across all channels, so FB advertising efficacy will likely be under-reported.

 

Conclusion

If you’re an advertiser using an agency to manage Facebook ads, chances are you haven’t been getting reports broken down like above. And with the prevalence of re-targeting campaigns in Facebook, chances are you’re seeing significantly inflated ROI from FB ads. I would wager that most agencies report on FB ads with default settings, so I wouldn’t necessarily hold it against them. But it might be worth a discussion about future reporting.

So is Facebook lying? No, that’s not fair. After all, it is possible to see both metrics. But it seems strange that seeing these two metrics separately is not straightforward at all… to the point that most advertisers never see them. Google automatically separates them. Why not Facebook?

Join 150+ Leading eCommerce Brands

And see how Visiture can grow your revenue online through award-winning transactional focused marketing services.

Leave a Reply

Popular Articles.

Data-Driven Marketing + Creative Commerce = Results.

Let’s Bring Our Teams Together and Connect You to Your Ideal Customer.