Succeed at Reporting Despite Differences Between Google Analytics & Facebook Ads.
How do you attribute conversions to paid social? Do you use Google Analytics exclusively or do you rely on Facebook’s ad manager to fill in the details? No doubt, you’ve run the numbers and see that they’re never the same.
Jon Loomer stated in a recent article about problems with Facebook conversion tracking, “One of the most common questions I get from advertisers is regarding the number of conversions that Facebook reports. They feel that the number is either inflated or underreported.” He continues by outlining 4 issues you could be missing that could be the reason. The one I want to delve into today is the last on his list regarding Facebook’s attribution window.
Once you get past the mechanics and fix everything that needs to be fixed, you’re still going to run into attribution issues if you haven’t already prepared for that and that has the ability to undermine your paid social work more so than anything else.
Before I dive in any farther, I want you to be aware I have theories on what the best way to find a consistent paid social attribution model is but I don’t presume I have the best one (or even a great one). I just have one that works for me and if it helps you untangle your data, then that’s a win.
First, let’s review the common attribution models and some potential scenarios you may run into if you use them exclusively while rejecting the other.
Google Analytics Attribution
Google Analytics (GA) has adopted a last-click attribution model. This means, the last place a user comes from before converting is the one that gets credit for the conversion. They have an attribution tool that allows you to see other models to compare, but it’s still not perfect because it will mis-attribute Facebook’s click-through conversions.
So when a company is deciding where to invest their advertising spend and they look at GA for bottom line success they won’t see the whole picture. This is frustrating for those who work in paid social because when we see that Facebook is working, all we want to do is shout if from the rooftops. But then GA saunters in and tells everyone that Facebook didn’t do as great as we thought it did.
Armed with the knowledge that we’re proliferating incorrect data, our peers lose trust in us, so we sneak back to our stand-up desks and pour over the numbers again. We double-check the Facebook dashboard, attribution models in GA, custom paid social reports; we analyze everything down to our UTM tags, and then put our head in our hands and weep uncontrollably because nothing makes sense.
Facebook & Instagram Attribution
Facebook has adopted what I like to call the “Facebook-only model” (28-day click / 1-day view). Looking at Facebook exclusively also mis-attributes other channel’s value because it doesn’t take in anything outside of their platform. This makes sense, since they never claimed to be a universal analytics provider, but then, the inevitable conclusion we come to is, we can’t trust Facebook can we? They can’t tell us how many touch points that consumer went through before the first link click on our ad that resulted in a click-through conversion. They don’t have a last-click metric column either, to see how many of our click-through conversions were last-click within Facebook. We’ve become accustomed to knowing that information and the absence of it can be frustrating if we’re not checking GA (Google Analytics). I’d be remiss if I didn’t mention that Facebook has come under fire in the past and even quite recently for miscalculated metrics. So regardless of where you get your metrics, just have a backup (or two), if the day comes when you need to dig into the details a little more.
So here is our dilemma. We paid social advertisers are struggling with platforms that don’t communicate effectively so we’re left as the messengers interpreting the data ourselves and in many cases, coming to the conclusion that we can’t be sure if our paid social is working or not. Regardless of the truth, the complexity has made us detractors of our own work indirectly through our lack of confidence.
The Facebook Assist Attribution Model
Do you fall into any of these scenarios below?
- You’re head of lead gen and report to a VC board or a marketing exec that doesn’t have time for buzzword nonsense like VT (view-through) or CT (click-through) conversions? They want leads and/or revenue, that’s it. You need to decide how to measure that.
- You don’t feel comfortable reporting Facebook’s numbers as conversions?
- You feel like it’s painfully transparent to other GA users that by accepting Facebook ad conversion numbers and ignoring GA you’re muddying company data or over-inflating conversions which could lead to problems?
Then you’re probably looking for one consistent number to give them, not 3 or 4 different conversion numbers that can’t go neatly into one cell in the revenue or lead column.
I get it. You want to balance 1) being honest & accurate 2) not undercutting the value you see but can’t interpret on paid social channels and 3) keeping it simple.
Here’s my suggestion: adopt a last-click attribution on all channels & count view-through & click-through conversions as assists. You will never have to wonder which channel closed the conversion, because it will always be front and center (assuming your FB & GA pixels are working correctly).
Here’s what I mean. In this table below, according to GA, we have 12 conversions this month. But if we ignore GA and look at Facebook we have 25 conversions (16 CTC & 9 VTC).
Just looking at this table you get it, right? What the heck is a decision-maker supposed to do with this? If an associate gave me a report like this, it would take me a while to even understand what I’m looking at.
If you want to expand your paid social budget, it’s beneficial to report Facebook’s numbers instead of Google Analytics, right? Well, no. Again, that leads to strong disparities between Google Analytics reports and Facebook Ad reports in the long-term and that undermines the bond of trust we seek to attain from our partners.
Those extra 13 conversions are attributed somewhere else in GA. Sooner or later those attribution problems will catch up to you. So in this alternative model I’m suggesting, your numbers will look more like this once they’re in hands of your superior:
This assumes your UTM tags tracked which campaigns converted, so you can easily divvy up what campaigns brought in last-click conversions & by omission in GA which ones then logically brought in what I’m calling assists.
Assists are simply all the 28-day click-through and 1-day view-through conversions from Facebook that weren’t last-click, and therefore, weren’t attributed as conversions in GA.
Remove the doubt. You don’t want to only report GA last-click numbers because that undercuts the other conversions (or assists) you know are the result of your paid social advertising and this is a way to reconcile that without ignoring it or burying it in the direct or organic columns where GA attributed them. The priority is pulling the paid social conversions (that weren’t last click) out of the clutter and putting them front and center where they belong.
Make sure the client or your superior is equipped with a thorough understanding of what you’re doing. This isn’t a secret. It’s a way to simplify multiple sources of information and create a template that’s accurate & simple.
In all types of sports, the one who assists is integral to winning – ball hogs don’t win games. They might score a goal and get more attention but sooner or later you need a support system. No one can do it alone. It’s the same way with advertising. If you forget about the assists and focus on the best player, sooner or later your marketing mix is going to lose (i.e. all the eggs in one basket idea). I’ve seen it first-hand too many times. Clients give up profitable paid social accounts and then months later, realize their lead gen funnel has dried up (sometimes it takes longer).
ZATO Paid Social Attribution
Which paid social attribution model does ZATO stand by?
Short answer: Last-click from Google Analytics, Facebook 28-Day Click-Through Conversions, and 1-Day View-Through Conversions.
Yes, there’s overlap. I want to dive into that more but suffice it to say, the overlap in the two disparate attribution windows are apparent in the final report. Also, each client is different. Some are happy with one number and understand the arguments we present about including click-through and view-through conversions in the final sales quantity/revenue numbers.
If the client wants to interpret this themselves, we present the numbers together so they can see the differences themselves and judge the successes and failures on their own.
We value precision just as much as accuracy and that’s why we rely on both GA & Facebook when reporting to our clients. It’s not always easy to explain but it’s consistent and honest.
Here’s what a sample internal report would look like (VT stands for view-through, & CT stands for click-through:
Once the final report is prepared it’s more obvious what you’re looking at but this is basically what we’re seeing. It shows view-through, click-through, & on another part of the report there’s a widget with last-click. Everything out in the open. It’s nice for clients that like to see the proof in the pudding. But it does still lack clarity, something the assist model provides.
I didn’t write this post to reminisce on how frustrating it can be to try to report on campaign successes and get blocked by overlapping attribution widows. Although, proofing it now, it may look like that. I wrote this because I had to. We run into this at ZATO on what seems to be a daily basis, and I felt it was finally time to address it.
What attribution issues are you facing on Facebook that haven’t been addressed? Do you agree with the assist theory? I’d love to hear your thoughts.
Tweet me @timmhalloran