• Sibel Akcekaya

ADOBE ANALYTICS MARKETING ATTRIBUTION IQ



One of the most difficult part of marketing is attribution. Still most marketers do not have exact formula to understand how marketing activities or customer interactions lead to certain results. Digital era made some of the interactions measurable but at the same time multi channel, multi device world made things more complicated


Every analytics software out there tries to help companies with this problem. Today I want to talk about Adobe Analytics Attribution report. This report is one of the big differences between Google Analytics and Adobe Analytics. While attribution models are similar, usage is much more expansive in Adobe. First touch and last touch terms have been inherently built in Adobe Analytics software since Omniture. I was already using First, Last touch models along with linear model in 2007(14 years ago) I was not just using it for marketing campaigns but I was also using it for other variables. For instance Internal Search Term. I would like to know what was the first and last keywords that customer searched for. I'd also like to know all the keywords that customers searched in a visit. Adobe has this feature because marketing attribution and customer journey is not just about advertising and marketing campaigns. This is a very narrow view of marketing. Marketing attribution includes every customer interaction.


When it comes to models, I think still many companies are still using last touch attribution to evaluate marketing campaigns, if this is the case , you are really not seeing complete picture especially if you run many different campaigns and if you have multiple customer touchpoints. I think many marketers focus too much on advertising and forgetting other touch points like brick and mortar interaction, phone support, website and app usability, customer experience and after sales support etc.. Last touch is for sure a very beneficial data but it is just part of the picture.


A given customer journey isn’t linear and often unpredictable. Each customer behaves differently at their own pace; often they click around, leave the site or app, come back, create a cart, abandon, then come back through other channels or engage in other non-linear behavior. These organic actions make it difficult to know the impact of marketing actions the customer journey.


For marketing attribution one of the techniques I like to use is heuristics or trial and error. This kind of approach based on lots of testing. You keep changing things around and look at the results and then learn what is really going on. But many companies do not like to keep changing their campaigns, as they are afraid of losing some sales, so this might not be ideal method for all especially when bureaucracy runs high and optimization awareness is low.


Other method we can use is statistical and algorithmic model. We basically rely on machine to find the relations and makes us suggestions.


The biggest tech companies like Google, Amazon uses both techniques together. I think that's the best. But not all companies will have the same resources these companies have.


Both Google Analytics and Adobe Analytics try to help businesses to evaluate attribution. They both uses similar methods and models.


One thing people do misunderstand especially Google Analytics users, is that those models you see in analytics are not the real picture of your company's attribution. You need to look at different models and compare them, maybe observe them over some time. Every business has a unique attribution model, you can't just use ready models and think that those numbers are your company's reality. I actually have seen people doing that, taking those models as real. Of course we will never have a model that is 100% accurate, but we can get close to it.


And sometimes attribution is easier than that. Your team might decide to credit different channels based on a business decision. This is also another method. But usually there is a great business decision here.


There are couple things to consider before choosing the model and then taking action based on the chosen model.


1-Be aware of the biased marketing attribution reports.Attribution model decision should be determined by the whole team who has a stake in this. I have seen cases where one analyst who is responsible of paid search making the decision and representing the data in a way that credits him/her. I have seen even departments not to give substantial credit to natural search because they want more budget. And sometimes analyst is not really experienced in this area and he/she is just creating wrong insight unintentionally.


The other thing that is very critical. Are you tracking and reporting touch-points correctly? Does the last touch report really tracks last touch? Never use data without testing or making sure it is set up correctly. I have seen so many professional looking dashboards that sadly revealed wrong insight due to wrong tracking.


Here are the attribution models under Adobe Analytics. Let's review each of them. They usually mean the same thing in Google Analytics.


I will be going from the most basic to the most complex one and hopefully you will see why using very basic touch-points might mislead you. And remember all these models depend on look-back window we defined. Attribution and look-back window exist together. I will explain look-back window below.


What is The Last Touch Attribution Model?

Last Touch is the most basic and common attribution model that gives 100% credit to the touch point occurring most recently before the conversion. I would not base all my marketing attribution on this one but it does have some uses. If you are analyzing search marketing you can check if your campaigns were there before customer made a purchase. It is a great tool for micro analysis, not macro analysis though.


In Adobe we do not use marketing attribution for marketing campaigns only. This is the biggest difference between Google Analytics and Adobe Analytics. We can use attribution for many other dimensions. For instance if you have SEARCH on your site, and if you want to see which internal search keywords and results led to conversion, you can use last touch attribution here. If customers engaged with our site after searching certain keywords, we would like to see what was the last keyword. We can also check the first keyword that they searched, and see the difference between them.


What is The First Touch Attribution Model?

First Touch gives 100% credit to the touch point first seen in the attribution look-back window. I look at this channel as a journey starter and it is directly related to the brand awareness. For instance with one of my clients I have seen that most of its customer journeys started with Google Ads and search marketing. Many people needed this client's services and they were searching for it on Google and then they visited my client through Google Ads channels. So in that sense Google Ads was customer acquisition or brand awareness channel. So it is a very important channel. In order for your customers to make a purchase from you, they should know about you first. That's why looking at last touch only can be very misleading. Because in some cases without the first touch, there can't be last touch. They are directly related.


Another non-campaign example to this would be again internal search keyword. What is the first keyword customers search for? Or we can use this attribution for assessing on product recommendation effectiveness to see first recommendations customer saw.


What is The Same Touch (Linear) Attribution Model?

Same Touch model gives 100% credit to the very hit where the conversion occurred.


A helpful model when evaluating the content or user experience that was presented immediately at the time of conversion. Product or design teams often use this model to assess the effectiveness of a page where conversion happens.


What is the Linear Attribution Model?

Linear gives equal credit to every touch point seen leading up to a conversion.Useful for conversions with longer consideration cycles or user experiences that need more frequent customer engagement. It is often used by teams measuring mobile app notification effectiveness or with subscription-based products.


What is the U-Shaped Attribution Model?

With U-shaped model we are getting better at giving credit to more campaigns. Up to this point we have seen first touch and last touch campaigns. But what about the campaigns between First and Last? Are they not important? Many times retargeting campaigns fall in between first and last touch. They are usually influencer campaigns or reminders , and they might not always close the sales but they might affect it significantly. So starting with U shaped model, we started to recognize these affecting touch points.


U-shaped attribution model gives 40% credit to the first interaction, 40% credit to the last interaction, and divides the remaining 20% to any touch points in between within the look-back period we defined.


What is the J-Shaped Attribution Model?

If you have been using last touch for a long time and need a more conservative approach to change your method, J-shaped attribution model might be right for you. It gives 60% credit to the last interaction, 20% credit to the first interaction, and divides the remaining 20% to any touch points in between.


What is the Inverse J Attribution Model?

There might be a case where you need to give more credit to the first touch.In this case you can use Inverse J attribution model. This model gives 60% credit to the first touch point, 20% credit to the last touch point, and divides the remaining 20% to any touch points in between.


What is the Custom Attribution Model?

This is the model I would use most of the time. I like to define and sometimes change credits for the campaigns especially for testing purposes. With this model you can define % for each touchpoint.


What is the Time-Decay Attribution Model?

This model runs against time. It gives more credit to those conversions that are close to the conversion. This would be great attribution for specific promotional periods.



What is the Participation Attribution Model?

If you are Adobe Analytics expert you know what participation metrics means. Participation is basically gives credit to all touch-points. The total number of conversions is inflated compared to other attribution models. Participation deduplicates channels that are seen multiple times.Excellent for understanding how often customers are exposed to a given interaction. Media organizations frequently use this model to calculate content velocity. Retail organizations often use this model to understand which parts of their site are critical to conversion.



What is the Algorithmic Attribution Model?

This new model uses techniques from cooperative game theory to statistically determine which channels deserve the credit for conversion specifically. The model applies a technique known as the harsanyi dividend which is a scalable generalization of Nobel laureate Lloyd shadley's work unfairly distributing the winnings of a team of players in a game or in our case attributing the conversion events in your data to a set of marketing channels. Unfortunately not every Adobe customer will have access to it. At the moment only ultimate package customers will see the option below



Attribution Look-back Window

All attribution models make sense within defined look-back window. Sometimes this is a set agreement between the client and the advertiser. Many companies will give credit to certain advertising if customer clicks that advertising within an agreed upon time. As long as this is not a very special situation, I don't like this kind of set look-back windows.


When I work with a client I like to test out the lookback window. One of my clients only looked at campaigns within 30 day windows but once I actually tested different lookback windows, I found out that client should look at campaign within 14 day window. What does that mean?


It means that a marketing campaigns affect is around14 days, if it does not convert within this time frame, it loses.


Of course look-back window might be different depending on campaign, conversion type and the desired outcome or promotion. It might look like as we are adding more complexity to the mix, but attribution is not simple but details can give you a performance boost. In Adobe Analytics you can choose different look-back periods and compare them. You can also choose different models and see the difference.


Now go ahead and start using attribution and please add a comment below if you have questions or comments.


cheers

sibel akcekaya





If you want to get most out of your Adobe Analytics implementation you can contact me on Linkedin for consulting or training. I both provide implementation services and business analysis and training services. I have 14 years experience to implement and use Adobe Analytics with big brands like Expedia, American Airlines, Best Western International, ING Bank, Vodafone, Buhler Group. Unfortunately many Adobe Analytics experts out there only tool analysts, they do not have significant product development experience. I have been developing digital products since 1999. Adobe Analytics has become my biggest support for an insight and that's why I learned it very well. I am certified Adobe Analytics architect.

https://www.linkedin.com/in/sibelakcekaya/

You can also view my company website:

https://www.nobhillconsulting.com/


My other passion is food, visit my site to see all my recipes:)

https://www.turkishhomerecipes.com/

Recent Posts

See All

Thanks for submitting!