Google Analytics Acquisition and Attribution

Google Analytics Acquisition Attribution

Published by

May 30, 2017

Acquisition tells us how people get to the site.

Conversion tells us how many of those people purchase.

Our goal in analytics is to find which channel brings us the most visitors who are the most likely to also purchase.

Save yourself from the missed steps and trial + errors and download our guide on HOW TO SET UP Your Own Attribution Model.

We could possibly get this information manually, but that would involve going through multiple reports to get the data you need. A few days later, probably by the time your campaign ended, you would have semi-accurate reports of who came from where and how much they spent.

But there is a way to marry these two reports, and it lies inside the Conversion section of Google Analytics reports.

The multi-channel funnel (which functions only if you have set up set up proper goals) shows you the combination of conversion and traffic sources.

Google Analytics Acquisition Attribution Reports
Google Analytics Acquisition Attribution Reports

This report shows the proportion of all the channels that had a part to play in conversions. For more information on this report you can read our blog post on Multi-Channel Funnels.

In this post, we’re specifically looking at the attribution model that you should use.

What is an attribution model?

An attribution model is a calculation that assigns (attributes) the accurate value of revenue to specific channels that visitors use to get to your website.

Why use attribution?

Why do we care about this? What is all the fuss about assigning revenue to channels? After all, what we care about is the revenue we pull in.

While the revenue is important and avoiding the red bottom line is good, the question is: Is it is good enough?

Picture this:

You have an e-commerce website that sells widgets. Unless you are the only one in the known universe doing this, it is highly likely that you need to invest some money in bringing visitors (and potential customers) in. So if you are spending money to get that traffic, wouldn’t you want to know which of the channels perform best and which aren’t worth the money?

Of course!

This is what we need attribution models for.

There are a few attribution models we could use

Google Analytics uses what is called the “Last non-direct click” model as default.

This model attributes 100% credit for conversion to the last channel taken by the visitor before the conversion. The problem is, this means that any direct traffic is ignored and the credit is given only to the other acquisition channels – PPC, referrals, organic search or paid search. This method ignores brand recognition and the possibility that visitors came to your site because they know your company.

This model works best for sites that derive the majority of their conversions from paid or organic search – a situation that is pretty unlikely.

An overview of the attribution models

1. Last non-direct interaction

This is GA’s default model and, as we already have seen, totally inadequate.

2. Last interaction model

This model gives the last interaction before conversion all of the credit. The only difference from Google’s default model is that this one includes direct traffic. This model still ignores the existence of any other interaction between your site and the visitor, making it pretty useless.

3. Last adwords

This model attributes all the credit for conversion to the last AdWords click prior to conversion. Once more, a model suitable only for those whose entire marketing strategy relies on using AdWords. Otherwise, as you can probably imagine, it’s not very useful. It can be used to determine the most effective AdWords campaign, though.

4. First interaction model

The entire credit, according to this model, is attributed to the channel that delivered the first interaction of the visitor with the site. It means that, if the visitor clicked a display banner or a social media post say seven days ago, and then interacted with your site a number of times, finally purchasing some product on the seventh day, the entire credit would go to the first interaction. It is hard to imagine a situation where this would be true. But, this model is useful for determining where your brand awareness starts.

Avinash Kaushik has a memorable quote about this model:

“First click attribution is akin to giving my first girlfriend 100% of the credit for me marrying my wife.”

5. Linear model

This model attributes equal credit to every interaction in the conversion path, from the first to the last. It is highly unlikely that any real conversion path will conform to this model and it can only be considered a rough approximation. We do not recommend using this model for any purpose.

6. Time decay model

The Time decay model represents the idea that the most recent touchpoints or interactions prior to conversion have more influence on the visitor’s decision, and should therefore receive more credit. This idea is probably the closest to the actual behavior of consumers. It has drawbacks, as it assumes a linear decay and has a ‘half life’ of seven days. This means that each week that passes between the interaction and actual conversion halves the value attributed to that interaction.

7. Position-based model

In a position-based model, credit is determined based on the position of interaction in the conversion path. It is a combination of the first and last click models: The first and last interactions each receive 40% credit for conversion, while all the steps in between share the remaining 20% equally.

So what should I use?

To accurately model your attribution and give credit where credit is due, you must know the business model of the website.

As we have said (and most experts agree), the time-decay model represents a perfect first choice for most circumstances. This model is the closest to how real life works. You can adjust the half life of the decay, but you must know the approximate business cycle of the website you are analyzing.

The best possible option to use is the custom model. Custom model allows you to define your own proportion of attributions to each step in the conversion path. Using this model you will certainly get the best results. It is however a dangerous path if you do not know the business process and marketing strategy of the business.

Google Analytics acquisition and attribution practical example

How do we do this in practice?

First, we begin with the Assisted Conversions screen in our Multi-channel funnel section and look the results. They will typically look something like this:

Google Analytics Acquisition Attribution Assisted Conversions Screen
Google Analytics Acquisition Attribution Assisted Conversions Screen

Here, we can see the channels that brought about the conversion, based on the default last click model.

The numbers in the last column on the right tell us which channel visitors use to complete the conversions, and which channels are primarily marketing channels that assist in conversions. Any channel that is closer to zero is more used to complete conversions; those that have values higher than 1 are more used as assisting channels.

Based on these values, we can determine the influence that specific channels have on conversions.

That can show us which channels earn us more revenue.

Then, we can adjust our marketing budget and increase spending on more effective channels.

Note that this table is using the basic attribution model. To have a more realistic result, we will use the attribution model comparison tool in Google Analytics.

Google Analytics Acquisition Attribution Model Comparison Tool
Google Analytics Acquisition Attribution Model Comparison Tool

Here, the last column shows us the difference in effectiveness of the channels using the default model AND the time-decay model.

We can see that the referral and display channels were credited less in ‘last interaction,’ while according to the time-decay model, those channels performed 12% and 18% better respectively.

Using this data, you can carefully examine your marketing budgets and make adjustments to better reflect the effectiveness of individual channels.

Prior to doing this, make sure that all your paid marketing contains UTM parameters so you can actually track when people reach your website using those channels. Otherwise, you will be effectively blind and the channels will not receive the credit they deserve.

Google Analytics acquisition and attribution conclusion

Attribution models in Google Analytics have a clear purpose: To enable you to use your marketing budget wisely.

When you use digital analytics, your marketing budget no longer has to depend on guesses. But the key is to get your model right.

You need to observe your customers’ shopping habits closely (using analytics, of course) and once you distinguish the dominant patterns, use this knowledge to devise your own model.

Chances are that none of the default ones will exactly suit your business.

Keep in mind, the effort you spend on this will be rewarded in the long term. Also, always remember that, in order to achieve the best results, you need to import your expenses for the other Internet marketing channels (other than AdWords, that is). Importing these will enable you to see your ROI on each channel and act accordingly.

Google Analytics acquisition and attribution additional resources:

Avinash Kaushik “Multi-channel attribution modelling – The Good, the Bad and the Ugly models
Yehoshoua Coren “Understanding Google Analytics Multi Channel Funnels

Save yourself from the missed steps and trial + errors and download our guide on HOW TO SET UP Your Own Attribution Model.
Series Navigation<< Google Analytics Spam – Don’t Be Fooled: Filter Bogus Traffic in Google Analytics13 Google Analytics Tips & Tricks for Better Data >>

Published by

Edin is a Senior CRO Consultant. Edin is into Google Analytics and testing (any A/B testing tool really) and likes to write about it. You can follow Edin on Twitter.