Knowing if your advertising or marketing campaigns are successful or not has been a struggle since long before John Wanamaker quipped about it over 100 years ago.
With no direct way to tie increased sales to a specific campaign, attribution was difficult at best.
Here’s how attribution used to work:
- You run an advertising campaign
- If sales go up, the campaign worked. ROI is assigned based on the increase in sales.
- If sales remain flat or drop, the campaign didn’t work and you fire your ad agency.
With the internet came the ability to see how a customer interacted with your brand, following their path into your digital space and through to a conversion.
This provides the opportunity for attribution and for determining the ROI of the different steps along the buyer’s path.
The problems begin when trying to decide how much (if any) credit should be given to each step along that path.
Attribution Models
Google Analytics offers several attribution models, all of which have their strengths and weaknesses. Choosing the right model can be instrumental in assigning the proper value to each channel.
Let’s take a look at your options:
Last Click / Last Interaction Attribution Model
This model gives 100% of the credit to the last interaction the buyer had in the path.
The obvious flaw here is that it ignores all of the other interactions. All of the work it took to get the buyer to that point receives zero credit.
Last Non-Direct Click Attribution Model
This is basically the same as the above model, just pushed back one step. The flaws remain the same – it ignores all the work it took to get the buyer to that step.
Last AdWords Click Attribution Model
This model seems perfect, assuming you work on PPC for your company. This model would give PPC all of the credit, ignoring the other steps.
First Interaction/First Click Attribution Model
This is the exact opposite of the Last Click Model, and equally as flawed. This is the SEO’s model of choice.
It assumes that all of the work was in getting the buyer into the marketing funnel, and that everything else is just for fun.
Linear Attribution Model
This model treats every step equally. Ostensibly this seems like a good idea – after all, the buyer traveled through each step on their way to the conversion.
The problem with this model is that not every step contributed equally. With a limited marketing budget, priorities need to be assigned and funds need to be spent where they’ll have maximum effect.
This model ignores those realities. That being said, it’s still better than all of the above models.
Time Decay Attribution Model
In this model, each point in the journey is given more credit the closer it is to the end.
The model assumes that since conversion is the goal, the steps that are closer to the end are more valuable.
The problem with this model is that it assigns the least credit to the first step, which brought the buyer into the funnel in the first place.
However, of the pre-defined models this is the best choice, coming the closest to a realistic model.
Position Based Attribution Model
This model assigns the most credit to the first and last steps (40% each by default) in the journey and gives the middle steps the leftover credit.
Logically this makes sense, since getting the buyer into the funnel is a crucial step, and none of the other steps can occur without this one.
The problem with this model is that it may be giving too much credit to the first interaction and not enough to the in-between.
This is a better choice than most, but isn’t ideal. If you can’t use the time decay model (for some reason?), go with this.
Custom Attribution Model
Now we’re talking. No preset model will give you exactly what you need, but with a custom model you can get as close as possible.
All businesses are different and will have different requirements from attribution models. Custom models allow you to use one of the above models and add your own input.
You can change the values for the distribution of credit, use engagement with pages as a factor (time spent on a page), and add a variety of other factors into the model.
This is by far the best choice, but it requires much more detailed info about your buyers and your business to be effective.
Using the time decay model customized to give added weight to the first interaction seems like a reasonable compromise for someone looking to not spend all of their time on analytics.
Which to Choose?
You should definitely be using one of the last 4 models. While they all have flaws, at least each of these is accounting for multiple steps in the path.
The time decay model is recommended for beginners, and as you get more comfortable with analytics, move into a custom model that better represents your business.
Lost in Translation
You can learn quite a bit about your buyer and your marketing efforts from analytics data, but it’s important to remember that data without proper interpretation is valueless.
Properly interpreting analytics data takes more than a Google Analytics account.
You need someone with the training and experience to interpret this data in order to take full advantage of it.
If your business is too small to have an in-house analyst (or better yet a team of them), then using a third party vendor is your best option.
Their expertise is essential to gathering and interpreting the data, and then making actionable recommendations based on the results.
Conclusion
Choosing the right attribution model is an important first step in interpreting analytics data for your business.
Attribution modeling is an important source of information for determining the effectiveness of campaigns, estimating the value of specific channels, and for identifying the strengths and weaknesses of a campaign or website.
This is information your business needs to succeed in the market today. Without good analytics data your business is just guessing. Choose the attribution model that works for you, study the results, and keep testing and making changes.