4 Search Attribution
AdWords attribution is the methodology of assigning credit to different actors in a chain of AdWords search clicks leading up to a conversion. The usual attribution challenges provided by device, channel and platform proliferation are non existent as Google ties these touch points together (through it’s knowledge of signed in users) across devices. And there is only one channel and platform.
The default model in Adwords is last click. This is a gross over simplification, but it is an attribution model that has served many businesses well. Especially when you are up against competitors with a much larger market share the last click model can be very effective. You are, as of early 2016, also able to select traditional non-last click models such as linear, time decay, first click or position based. These models are known as “rules-based models”, since they’re based on a simple rule to assign the corresponding conversion credit.
No model is perfect. But all of these “rule based models” models are, in ninety-nine percent of use cases, inferior to the Data Driven model. The Data-Driven Attribution model in AdWords (still in BETA) provides an evidence-based attribution model that values all AdWords search clicks on the conversion path appropriately, based on each search click’s incremental contribution to driving conversions. Which is exactly what you (should) want.
When do you qualify for AdWords Attribution
You can always switch to any of the rule based models, irrelevant of the investment you make in AdWords. With my one euro a day account, where I import email signups as conversions via Google Analytics, I already get the option to pick another model (as per image below).
If you are keen to go for the Data Driven model you should be a relatively large advertiser with tens of thousands of Adwords clicks per conversion type (example – your track sales and leads – two different types of conversions) on Google Search every four weeks. And that is not all. You will need somewhere in the range of 1000 conversion as well in the same time period. Anyway, if it is not in the drop down menu, you will know enough.
AdWords Conversion tracking versus Google Analytics goal import from an attribution perspective
If you are serious about improving your AdWords attribution importing Google Analytics goals is not best practice. I would strongly recommend installing the (cross account) Adwords Conversion Tracking pixel if you have not done so already. This allows for the attribution logic to take all AdWords search clicks across different devices into account.
An example. Customer A has visited your website twice via Adwords Search click before he booked a hotel room. First on his mobile on his way home from work in the train. After discussing it with his partner at the dinner table, he books on his laptop at home. If you import Google Analytics goal, you will see two customer journeys, if you have Adwords Conversion tracking installed, you will be able to see this as one customer journey and connect both clicks to the sale.
A significant step forward that takes a lot of the guesswork out of your more upper funnel mobile search investment.
How to see if cross device Attribution could enhance results?
Late in July 2016 Google announced the new Cross Device Attribution reporting for Google Search. I would strongly recommend looking at all three reports under the new Cross Device Activity report. If you do not have any cross device activity, my first guess would be that you have over optimised for last click and hence paused all activity that did not perform on a last click basis, or you have an extremely short customer journey.
Should I just consider switching non last click AdWords Attribution?
You should. But, you might want to test how different valuation of certain campaigns, ad groups or keywords looks under another attribution model first. You can do this in three different ways. First one is perhaps the simplest one. You can go to the report shown above in Adwords (Tools > Attribution > Attribution Modeling), and compare the model you are considering versus the default last click model. You can do the same in Google Analytics and select only Adwords traffic.
Or alternatively, you can add columns under Campaigns > Columns > Modify Columns > Add First Click Conversions.
How to enable Adwords Attribution?
When you want to move away from a simple last click model you have another option on top of the column add described in the previous section. You can set an attribution model at the conversion action level.
If you open AdWords > Tools > Conversions > Click on the name of the Conversion Action you want to edit > Edit Settings > Edit Attribution Model > Dropdown > Select model (from drop down displayed here),
If you do this, all attribution credit will be assigned according to the model you choose. Please remember that this will not change anything if you are bidding based on max CPC. If you are bidding based on max CPA it will change your bids immediately based on the new reality of the model you have chosen.
One last piece of advice. Or maybe two. Change the look back window on the conversion action to the maximum 90 days. This will allow you to get insights on as much of the customer journey as possible. Automate bidding based on max CPA, go for the data driven model, and give the algorithm some time to adjust and some pace to learn (by not lowering CPAs too fast).
DoubleClick Search Attribution
Whereas switching to a new attribution model is a turnkey approach in Adwords, DoubleClick Search has a lot more to offer in terms of customising bidding strategies and attribution. As a result of all those cool extra options, there are a lot of things that can go wrong as well.
The first thing to nail is your labeling of the account. Group every set of keywords that has a similar function in your Search strategy together under one label. The most common way to do it is to go for a “brand”, “brand generic”, “generics” and “competitor” grouping. But it depends on the size of your business and the number of clicks and conversions you are measuring. If you are one of the large product comparison sites, you might want to have that grouping set up for every product line (insuramce, mortgages, utilities, ect ect). If you office one product and ninety percent of your investment in Search is on twenty or so keywords, you might want to stick to a simple “brand” and “generic” grouping.
As soon as those label groupings have had twelve hours to set in, you can create a Data Driven Attribution model. Make sure you select the right floodlight tag(s). If you have multiple tags for different value drivers, you can combine them in the same model and set up another specific model for each of the main ones. You can choose what models to use for what bid strategies. Your brand campaigns would want to take all value drivers into account, where as you campaigns aimed at selling “product A” should perhaps use the model created for “product A”.