Main Attribution that Search Marketers Need to Know in 2019
23 Jan, 2019
Main Attribution that Search Marketers Need to Know in 2019
Biddable media is highly technology-driven.
That’s why it’s so awesome, but also why it’s so confusing at times.
To really excel in this industry, you need to know what to ignore and what to pay attention to.
This is certainly the case when it comes to attribution.
In this post, I’d like to give you a helping hand in the filtering process, based on my own company’s research into this issue.
I’m going to focus on attribution from a paid search perspective. So naturally, I’ll be talking about attribution models within Google Ads, and Google Analytics.
I’ll also consider a few independent analytics companies’ approaches to attribution, and finally the newcomer, Facebook Attribution.
I am not attempting to seriously critique any of the software referred to in this post. My aim is to provide guidance on the right questions to ask about attribution this year.
Google’s Attribution Software
Data-Driven Attribution in Google Ads
To be clear, this is a specific feature of Google Ads, rather than the general concept of data-driven attribution – blame Google if this has led to any confusion.
Data-Driven Attribution (DDA) is one of several attribution models available, including:
I’m sure you already know about the first two on this list.
Time decay gives more credit to clicks closer in time to the conversion.
Position-based gives 40 percent to first and last clicks, and the rest is evenly distributed across any clicks in between those two.
DDA, which was rolled out to all advertisers in October 2018, is the latest addition to the list. And it’s really cool as an idea.
Unlike the others, it is not predetermined how much weight this model will place on each click – it depends on the unique data attached to the account.
DDA looks at all clicks – converting and non-converting traffic – in order to estimate the value of a particular ad, or keyword, or ad group etc.
Theoretically, Google’s machine learning should be able to identify how much each ad in a click path contributed to the conversion
And it does kind of work.
From the tests done at Brainlabs so far, we’ve found that the adjustments DDA suggests are often useful – but we’re talking about pretty marginal gains here.
Comparing DDA and Last Click, for example, the variation between the models in attributing credit is only as much as 5 percent. This might sound like a lot, but once you convert this into optimization gains, the benefits are less impressive.
I’m not saying that DDA isn’t a good idea, nor that it isn’t useful to some extent. It just isn’t a priority.
Considering the time and effort needed to test DDA, there may be more pressing aspects of your paid search strategy to attend to first.
Google Analytics 360
Should you invest in the paid version of Google Analytics?
This is a far more pressing question than which attribution model to use for Google Ads campaigns.
Google Analytics 360 was launched in 2016, so there has been plenty of time for consideration following its release. However, a few important changes have taken place since then that re-open the debate.
Firstly, there’s the integration of GA360 with the customer relationship management (CRM) specialist, Salesforce.
Data from Salesforce’s CRM and Marketing Cloud can be integrated with data from GA360, all of which can be exported into BigQuery (part of Google’s Cloud Platform). This is a pretty lethal combination.
And then there’s the Google Marketing Platform (GMP), launched in June 2018.
GMP itself didn’t add many new capabilities to GA360; it was more of a symbolic act to demonstrate Google’s focus on integrating its many different marketing and analytics products.
However, it did achieve one important thing: greatly enhancing our ability to attribute display.
GMP has strengthened the integration DoubleClick and Google Analytics, essentially by making that integration easier to execute.
Prior to GMP’s release, GA 360 users were able to draw upon a native integration between DoubleClick and Google Analytics.
However, GMP has served to strengthen this integration, and provided a strong foundation for continued data-sharing across platforms.
GMP enables impression data to be integrated into the GA attribution model – a huge step forward for multi-channel attribution.
Whether GA360 is worth it ($150,000/ year) is entirely dependent on the individual case.
It’s worth noting here, as well, that GA360 has far more benefits than just better attribution: this post is focused on attribution, but there are a number of analytics capabilities that GA360 enhances.
For an excellent consideration of this question, I recommend reading Derek Gleason’s article published at the end of 2018.
There will undoubtedly be more discussion of GA360 as the year progresses, and I would recommend all enterprise and SME businesses to give it some serious thought.
Independent Analytics Companies & Facebook
The dust has not really settled yet since Google’s Ads Data Hub maneuver last year.
I know it was almost a year ago, but it’s still unclear whether Google’s barring of access to DoubleClick user IDs has caused the damage people anticipated.
In other words, has the lack of independent verification of DoubleClick data led Google to manipulate results in their favor?
Nobody knows, it seems.
A good test, then, for those with sufficient motivation: start making some comparisons between Google and non-Google attribution platforms.
There are some promising independents out there – Visual IQ, Flashtalking, Datalicious are just a few that have caught my eye.
And then there’s Facebook Attribution. Early responses to this seem very positive.
Simon Poulton has dubbed Facebook Attribution as being devoid of “platform bias,” based on tests in which Google Ads triumphed as the most important channel in user conversion paths.
Facebook’s tool will also benefit from the world-leading supply of user data the platform has amassed, and its numerous marketing science features.
So I would definitely be keen to experiment with Facebook Attribution, and especially to start making some comparisons with Google’s own measurement of Google Ads.
Look at what the independent analytics companies suggest, too. I’d be fascinated to see the results.
Don’t Over-Do Attribution
What is the impact of OOH and TV on sales?
This is clearly beyond the range of attribution software – even the latest generation of it.
It will never be truly measurable, because OOH and TV are for branding, and the long-term sales value of brand equity is simply incalculable.
So I’m a little skeptical when I see news of various TV companies offering their own attribution solutions for the value of TV.
When TV is programmatic, then there’ll be greater scope for it, but even then it would be ignoring the most important, long-term effects of this sort of marketing.
I’d also say that even within the highly measurable world of PPC, there is such a thing as over-doing it with attribution.
Better measurement can optimize performance and help to sharpen strategy, but it’s only one part of the puzzle.
As I hope this post has made clear, to be a PPC superhero, you need to choose your battles wisely. Optimize with discretion, folks.