I am a firm advocate for measuring click based conversions when it comes to performance focused online marketing with various attribution models applied depending on the goal. With that said however, I’m also open to the fact that like real world billboards; display and remarketing campaigns can generate view-through or post impression conversions. For those that don’t know, a view-through conversion is where someone has seen your advert at some stage and then has come back to your website or landing page direct or via another source and converted into a lead or a sale.
The main issue with measuring view-through conversions however is that you can’t actually tell if the prospect seeing the ad was the reason they came back and converted, or if it was another marketing campaign (the one they clicked perhaps) that influenced their decision. They may have even come back and converted if there was no marketing turned on at all. Without a clear and accurate view of the sources, campaigns, ads or targeting that are delivering actual results it becomes hard to correctly attribute the conversion and then optimise and improve your overall marketing results.
There are two main types of marketing that use view-through conversions:
Non Behavioural-Based Marketing ie. A banner placed on a specific website
With non behavioural-based marketing such as placing your ads on a specific website or targeting ads towards a specific demographic on Facebook, chances are the prospect doesn’t know your brand so when they see your ad and later convert there is a good chance the advertising had an impact. A view-through conversion is generated and you can then attribute part of the conversion to that advertising channel based on whatever model you are using.
Behavioural-Based Marketing ie. A remarketing campaign targeting past website visitors
With behavioural-based marketing such as a remarketing campaign, measuring view-through conversions accurately becomes a bit harder. The prospect has already been to your website, they know your brand and then you show them an advertisement. Later if they come back and purchase, did the campaign you were running cause them to come back? Possibly, but its not as likely as a view-through conversion that has been generated from non behavioural-based marketing.
Measuring this type of conversion for both types of marketing is never going to be 100% correct as there is no accurate way to determine what effect viewing the ad had on a prospect however behavioural-based campaigns such as remarketing, are in my view, where the main discrepancy lies. In the following sections I will look at two major issues with measuring this type of conversion for a remarketing campaign and a possible solution to more accurately attribute these conversions and thus improve your overall results.
Issues With View-Through Conversions & Behavioural Campaigns
There are two main issues with tracking view-through conversions on a behavioural-based marketing campaign such as remarketing. They are:
Issue #1: Have They Really Seen The Advert?
Currently most of the third party remarketing platforms report on view through conversions like this:
With the default setup these platforms report on both the click-based conversions and the view through conversions they have generated off each ad impression. Seems the correct way to do things but the issue is with the view-through conversions the user may not have seen the advert at all:
Yes, that’s right, the default settings in these platforms count all impressions served as having influenced the conversion whether the user has scrolled down far enough on the page to see the ad or not. In the scenario where the ad has not been seen it will not matter if your offer was good, your banner was designed well or if the website it was on had high-quality traffic the ad will have had absolutely no impact on the end conversion as it wasn’t seen and therefore the numbers reported in these platforms are not correct and cannot be relied upon. There are ways to tweak third party platforms settings to only count banners shown above the fold but this is not the default and still doesn’t guarantee the ad is seen by the user or had an influence.
So how big of an issue is this? Well back in 2014 Google released a study that showed that only 43.9% of its ad impressions were actually being seen by prospects. Most third party remarketing platforms claim 40%-60% of banners are seen but despite this they all seem to count 100% of the view-through conversions and as seen in the Adroll example above, use these conversion numbers to calculate the return on ad spend, artificially inflating the results.
There are steps being taken by some networks to address this issue but until every platform only counts a conversion if the ad is actually seen it is a major issue and one that will incorrectly skew the attribution of your conversions.
Issue #2: People Who Visit Your Website Are Likely To Convert Anyway
The audience who are most likely to convert into a customer for an online business are people that already know of and have engaged with your brand. It’s why, on remarketing campaigns we see lower cost per conversions than non behavioural-based marketing campaigns and why it is such a popular way to advertise. If a prospect has seen and clicked your remarking advert then you know for sure it has influenced them and helped get the sale. If however they have only viewed it, should you count the conversion 100% of the time?
Lets take a look at how the conversion process would work with a remarketing campaign turned on and turned off:
It is clear that with the remarketing campaign either turned on or off both scenarios are possible and even with remarketing off it is likely a portion of people will come back and purchase.
If you do have remarketing turned on, chances are that very quickly after that first session at least one of your remarketing ads will have been served to the prospect on one of the many websites they browse and from that point on they will be counted as a view-through conversion if they later come back and buy, regardless of which path they take. Essentially, that means if someone is served the remarketing ad but didn’t see it or weren’t affected by it and simply went back to the website they had already visited in the past to purchase (the same process as when the remarketing campaign is turned off above) – the conversion is claimed incorrectly.
It is possible to see the number of people coming back to your website a second time and converting using Google Analytics by checking the path length report. This will show you how many people visited your website once, then came back and converted later. It is generally a pretty significant proportion of your conversions whether you have remarketing turned on or off and there lies the problem.
It’s likely you have optimised and spent a lot of time on your website or landing page and have a good offer so chances are a big proportion of the prospects would have converted anyway – remarketing seen on or not. Counting 100% of the view-through conversions that are generated from a remarketing campaign is thus a major issue that can lead to you allocating a higher budget to your remarketing than the actual results show.
With two major issues in the accuracy of view-through conversions being counted for remarketing how can it be measured?
A Possible Solution: Split Testing A Blank Banner Set
Not being happy with accepting the third party platforms view-through conversion numbers for remarketing and attributing part of a conversion to these sources when they may have had no impact I started to brainstorm what methods could be used to more accurately determine the actual impact. Unfortunately, I have found that there is no 100% accurate way to track this type of conversion – even asking every single person in interviews would be inaccurate due to the subliminal effect banners can have. However there is at least one way to measure this conversion type more accurately that will help you determine what results you’re actually getting out of your behavioural-based campaigns – split testing your ads against a blank, unbranded banner set.
I hate spending a clients money and off the bat, spending your advertising dollars to promote a blank banner set sounds a bit crazy. In the long run however accurately working out exactly what level of conversions a network or certain type of marketing is generating can save you a lot of misallocated budgets and will improve your results overall.
This is a pretty simple test to setup, all you’ll need to do is get your current banner set & landing page and run it in a split test with your chosen remarketing audience against a blank set of banners with a different destination URL. Depending on what network you’re using you may need to get creative as to how the blank set of banners looks as plain-white banners can possibly be disapproved.
What you’ll find is that both sets will generate view-through conversions due to the issues identified above, so you will need to keep running the test until you have statistically significant data. Once you do, the difference in view-through conversions between the two (standardised so each has the same impressions) will give you some insight into the actual impact the banners are having on the end conversion.
For example, if your branded banners with a targeted website and landing page generated 100 view-through conversions and your blank banner set generated only 60 view-through conversions then the actual impact of running your campaign was generating 40 view-through conversions, not the 100 that most third party platforms would report.
This method may cost a bit of money in the short term however it will allow you to solve both of the issues in tracking view-through conversions mentioned above and give you much needed insights into which of your networks are actually working best with your banners actually seen by your target market and having an influence on the final decision.
Of course the current way of tracking and this method don’t fix cross device attribution at all (which is another blog article entirely) but it will get you closer to accurate. If you have any other ideas on how to accurately track view-through conversions for remarketing, please let me know in the comments.