Are you still using last-click attribution in your AdWords account?
If you just answered “yes” or “I don’t know” then you definitely-probably are.
It has been the default attribution method since AdWords was in diapers. I personally don’t have a problem with it, and it is not the end of the world if you want to keep using it. But the benefits of doing so are roughly:
- Your conversion numbers are pretty because they’re whole. (It’s nice to see 7 conversions, instead of 6.7.)
And… That’s about it. Google has been pushing agencies pretty hard lately to change their attribution model.
They have a point.
While it’s true that it may mean that some of your conversions are going to show in various combinations of decimal points, this is for a very good reason.
At a recent conference, an enthusiastic Googler showed why last-click doesn’t make sense, regardless of how unintuitive it can be to see that one of your campaigns contributed 0.1 conversions this week.
They asked members of the audience to pass a pen on from person to person all the way from the stage, to a lucky recipient right at the back of the room. The Googler then proceeded to profusely thank the person who made the last pass to the recipient. Everyone else who had formed the human conveyor belt to achieve this great feat, was ignored.
That is last-click in a nutshell. The last campaign, ad group, keyword and ad gets all the credit, no matter how many others were involved in turning a curious searcher into a paying customer. Then we make sure to give that campaign more budget and up our bids to keep that happening.
I don’t know about you, but that sounds a little crazy to me. What if I see a campaign that hasn’t converted for the last month? I’d want to cut its budget or switch it off completely and pretend it never existed.
But what if, I turn it off and the hero campaign that was delivering 100 conversions a month is suddenly only producing 50?
It’s an extreme example, but in this scenario, perhaps the campaign with “no” conversions was actually a key part of the research process for half of our eventual customers. While it wasn’t the campaign sending searchers to the “buy now” button, it was a key part of the customer journey to those sales.
You and I would want to know that before we make a huge mistake!
Last-click attribution can’t tell you this until it’s too late. The “unsung hero” campaigns in your account are labeled as worthless. The heroes get all the credit, and marketers don’t get a full picture of performance within their AdWords account.
Something has to change, which is why Google has been pushing their alternative attribution models with renewed vigour recently.
What will happen when I change my attribution model?
In the previous example, an “unsung hero” campaign that was responsible for 50% of conversions got none of the credit. This is not a realistic picture of what you would see when switching to an attribution model with a more holistic view point.
What you are likely to see is that brand campaigns were a little bit less responsible for conversions than you thought, and that mobile campaigns were a little more responsible for conversions than previously seen.
The chance of seeing a seismic shift in your data after applying a new attribution model is very low. The insight generated from this more holistic view-point though, could be invaluable to your PPC approach.
The great thing is, you don’t have to commit to a change in AdWords attribution models blindly. Using the attribution tool in AdWords, you can compare different attribution models before deciding on the model that best suits your account!
AdWords attribution tool
You’ll find this handy tool in the “tools” menu of your AdWords account.
When you first open this up by clicking the “attribution” link, you’ll get a nice overview of some average conversion statistics.
The one that I find most interesting here, is your “Days to conversion” number. In some accounts, this could be 4-5 days, which definitely makes you think about the amount of consideration time searchers are taking before making that leap to a conversion.
In the left-hand menu, you’ll see “Attribution modelling.” This is where you can compare what your conversion data would like using different attribution models before committing to a change in the way your account reports them to you. I would suggest comparing last click to at least position-based and time decay models. If data-driven is available in your account, I would highly recommend taking advantage of this!
AdWords attribution model options
So what options do we have? Which is right for your account?
There are several, from first click (which is pretty much as bad as last click) all the way through to the gold standard, data driven attribution.
The question of which one is right for your account will depend on the goals that you have, and how much consideration your customers usually require before making a decision.
Last click attribution
The whole intro of this piece was dedicated to how limited this model is. Last click is better than not tracking any conversions. It is worse than just about any other model that gives you a better view of your performance.
First click attribution
This model, as it implies, gives all of the credit for a conversion to the first click of an AdWords ad. There are situations when this may be favourable to last click, but overall it still subscribes to the outdated mantra that “there can be only one.”
Okay, now we are starting to get somewhere. I don’t personally use this model, but it does have something going for it that is at least a little better than first or last click. This model shares the conversion EQUALLY with any click that led to the conversion (within the conversion window).
Let’s try it out with Jimmy, an eco friendly romantic who’s keen to visit Wellington with his girlfriend. Maybe Jimmy searches “accommodation Wellington” on the first of month and clicks your ad, but doesn’t convert.
Then he searches “best Bed & Breakfast in Wellington” a few days later and clicks your ad, but still doesn’t convert (he’s hard to please).
Finally a week or two later, Jimmy searches “Celia Wade-Brown’s Green B&B” after thinking about those sweet looking green smoothies he saw on his first two visits to the site. He makes the booking.
Celia welcomes all guests to her Green B&B with a stirring rendition of Pokarekare ana
If last click is still your model, brand search just got another win. You might think about stopping your ads on “accommodation” searches, and maybe even on “B&B” searches because people just seem to love typing the name of your B&B directly into Google and making a booking. It must be those print ads you ran in the Coffee News…
Or we look at it with another model. With linear attribution, each of the those three clicks will be given an equal share of the one conversion. .33. .33. 33. Simple. Elegant.
Maybe a little too simple.
While it’s great that we are seeing more than click as part of the conversion path, is it really fair that all of the clicks get the same amount of credit? The answer usually is:
Probably not, but not always for the same reasons.
Time decay attribution
With time decay, the last click gets most of the credit, with every click before it getting a smaller share based on how long ago that click occurred.
For our Jimmy situation, this probably isn’t the best option. The last click would have received most of the credit, even though the first two clicks were pretty important In his decision making process.
Time decay attribution may make sense when the usual time it takes for a conversion to happen is quite short. In that case, it’s only logical to give a click that happened 28 days ago less credit than the one that finally delivered the conversion. The searcher may not even remember visiting the site the first time around.
Position based attribution
Position based attribution is a personal favourite of mine.
If data driven attribution is not available in your account, I would generally suggest that this is a pretty good option.
Position based attribution gives more credit to the first and last click but also gives credit to the clicks that happened in between those two. To be more precise, it gives 40% of the conversion to the first click, 40% of the conversion to the last click and distributes the final 20% of the conversion between any other click in between. If there are no other clicks and between then it will assign the first and last click 50% of the conversion respectively.
If we considered Jimmy and his bed and breakfast conundrum for a moment through the lens of position based attribution, the ad that he initially clicked when he search for accommodation would get 40% of the credit for the conversion.
This seems pretty fair. While that first click didn’t get him to book immediately it obviously made an impression. He began searching simply for accommodation and then decided that bed and breakfast seem to be the way to go.
The second click would get 20% of the credit for the conversion. This also seems quite fair. Jimmy had clearly been enticed by the concept of a bed and breakfast from the first search, which had brought him back to the site for a second time. But this second click also didn’t get him to convert, so it makes sense that it gets less credit than the final click.
The final click here would receive 40%. This is probably fair. It is certainly a lot better that this click be given 40% of the conversion as opposed to the 100% that it may get in last click attribution. But I can’t help feeling that it may still be getting a little too much credit.
This is where the Holy Grail of Adwords attribution comes into play. Now it’s pretty unlikely that someone running a bed and breakfast in Wellington is getting enough conversions to be eligible for data driven attribution, but for sake of argument let’s pretend that this is the most popular bed and breakfast and all of New Zealand (and possibly the world).
Data driven attribution
Data driven attribution takes attribution modelling to a whole other level.
All of the models that we have discussed so far use totally fixed rules to assign credit for a conversion within your AdWords account. What data driven attribution tries to do, is to analyse conversion paths and evaluate the contribution of each individual click on its own merits.
The reason you need so many conversions for this to be available, is that the algorithm must compare conversion paths against each other in many different ways to assess the value of each click.
In the case of the B&B, the data driven algorithm will look for conversion path patterns and share the conversion between the clicks based on how similar clicks have lead to conversions in the past. So if there is a click on an “accommodation” ad and a later click on a brand ad, Google will consider the importance of each and give more weight to the initial click than other models might if the data suggests its a key step.
We’ll leave it to another post to discuss data driven attribution in detail. If you’ve made it this far, you deserve to know the limitations of all of these models!
Limitations of AdWords attribution
One thing that needs to be made clear, is all of this is only talking about attribution modelling within AdWords. There is a lot of debate going on in the industry about attribution in general and nothing that you do with your Adwords attribution modelling is going to change how difficult it is to assign conversions across channels.
More specifically than this, these attribution models only work for search campaigns within AdWords. This is one of the most limiting aspects of these attribution models, and I hope it is something that Google can address at a later stage.
For example, if you have a display remarketing campaign and someone clicks on one of your remarketing ads before searching for your company several days later, that click on the display ad would be given no credit whatsoever for the conversion regardless of which attribution model you have chosen within AdWords. This particular conversion will be no different to a last click situation.
If you would like to know more about Google AdWords attribution, try Google’s documentation on the topic.
And as usual, if you’re looking for help navigating the murky waters of attribution modelling, AdWords or otherwise, get in touch with the team at Uprise Digital.
Friday, 22 December 2017