Attribution Modeling in Google Analytics: Why is it so important for marketers?

What do we mean by attribution?

A marketing ROI calculation needs to have a method of assigning credit for a goal conversion (sale, conversion, form submit or any such goal) to one or more touch points in the user path. A user path usually has more than one touch point. It gets interesting as many of these touch points are supported by marketing dollars.

It is common for users to research and get information. During this research, the user might

1. Click on a Google ad
2. Come back to the site using a Facebook post
3. Find an organic search result to click and reach the site
4. Reach the site by directly typing in the URL or through a bookmarked link.

In such situations, it is very important to attach credit correctly. An incorrect attribution would lead to loss in ROI. Common questions are:

1. What should the parameters be in the attribution model?
2. How long should the user be tracked for?
3. Should all touch points get equal credit?
4. Should there be a weightage for recency and/or frequency of user visits?
5. Are there standard models or can there be custom models depending upon the nature of business and even seasonality?

As you might realize by now, it is important to understand the effects of each touchpoint on the user journey before deciding upon a model.

Attribution Modeling in Google Analytics helps you do just that. It helps you evaluate the effectiveness of your different online marketing channels and study the impact of various channels (and touchpoints) on conversion.

DIFFERENT ATTRIBUTION MODELS

Last Interaction Model

1. 100% conversion value attributed to the last channel with which the customer interacted before conversion
2. Use: Transactional business or sales cycle that does not involve a consideration phase. For example, when buying FMCG products, people usually know what they want and directly visits the store to make purchases.
3. Perhaps the most outdated model

Last Non Direct Click

1. 100% attribution to the last channel that the customer clicked through before conversion
2. Default model used for non-Multi-Channel-Funnel reports.
3. Useful benchmark to compare results with other models
4. This model can work well if you are a small business owner who is new to digital marketing. It will help you remove the direct visits from being valued as last interaction.

Last Adword Clicked

1. 100% conversion value attributed to the most recent Adwords ad that the customer clicked before conversion
2. It is useful to identify the Adwords ads that closed the most conversions

First Interaction Model

1. 100% conversion value to the first channel with which the customer interacted
2. If your business is more focused on creating awareness through ads and branding campaigns, this model can help you identify the channels that introduced the customer about your brand.

Linear Model

1. Assigns equal credit to each
   touch point in the conversion path
2. This model is for big campaigns designed to maintain contact and awareness with the customer during the complete conversion cycle.

Position Based Model

1. 40% credit assigned to first and last interaction. The remaining 20% is assigned evenly to the middle interactions
2. When touch-points which created awareness and final touch points are most valuable for your brand, go for this model.

Time Decay Model

1. Based on the concept of exponential decay and gives most of the credit to touch-points nearest to the final conversion.
2. Perhaps the most practical model which can help you in measuring the promotional periods when you are running campaigns of short duration.

Custom Modeling

You can actually create a model specific to your business needs where you can decide how much credit you want to assign to a particular channel. Google Analytics allows you to apply multiple weighing rules based on site engagement metrics from your visits. You can also apply custom credit rules. But how much credit should we attribute to which marketing channel? This is what we call as attribution problem.

The best way to answer this problem is to run control experiments to build your own custom model. You can then apply multiple rules based on your experiments and assumptions.

Another solution is to ask your customers who complete the conversion process.You could ask them about how did they find you website? And how many times did they visit you website before making a purchase decision?

Learn more about Attribution Modeling-

https://analyticsacademy.withgoogle.com/course01/unit?unit=6&lesson=4

http://moz.com/blog/set-up-meaningful-custom-attribution

Which Google Analytics attribution model works best for your business? Let us know in the comments below.