Procuring the Right Attribution Solution | Cardinal Path Blog
Blog

Assessing a Data-Driven Attribution Solution

decisions

Data-driven attribution is used to determine which marketing channels have the largest impact on conversion. While rule- based attribution can be subjective, as you assign how much credit each channel gets depending on its position in the conversion path, data driven attribution leads to more objective results. This helps to determine which channels are driving the most conversions.

If you’re looking for an attribution solution, you may not even know where to begin. Are you aware of the types of attribution solutions the different providers offer? Are you even ready for an attribution model? What types of questions should you ask internally and to prospective vendors to ensure you get the best solution possible for your business needs?

In this post, we will review some of the key questions you should be asking when gauging the different attribution solutions. Think of this as an attribution procurement ‘pep talk’.

WHAT TO ASK INTERNALLY BEFORE SELECTING AN ATTRIBUTION PROVIDER

Attribution or…?
Attribution is a hot topic in analytics at the moment, but you should make sure that you aren’t acquiring an attribution solution just for the sake of having one. It’s important to assess whether your business is ready for a data-driven attribution solution. It’s also important to make sure that you are aware of the value that attribution will provide  for your business before you dive in. Questions to consider:

  • Are you ready for a data-driven attribution model?
  • What are the big questions that you are hoping attribution will answer for your organization?
  • What level of analytics maturity does your organization currently have?

If you are just getting started with web analytics, perhaps it will suffice to stick with rules-based attribution for the time being. It is  an advanced solution meant for mature organizations with significant ad spend to ensure that the attribution solution will be able to enact large scale change at your organization.

As far as what question it is that you are trying to answer,  you may first need to gather information about the sequence of channels that will best lead to conversions. A bottom-up attribution approach is most likely the best approach, and given the right data, will provide answers to these questions. However, there are many different methodologies for attribution, and they will vary between vendors.

If granular information on channel sequencing is not a prerequisite, a mixed-media model (MMM) or top-down approach may be ideal: using regression of historic spend and revenue information, you can measure the impact of marketing tactics on sales, and use this to optimize your budget across advertising channels.

Rather than focusing on “attribution,” focus on the question you want answered, and select the provider who seems best positioned to answer it.

Actionability and Organizational Readiness
Once your attribution model is built out, you will want to ensure that you have processes in place to allow you to use the recommendations to make the necessary changes. You will first need to make sure that key stakeholders are ready and willing to use the results once they come in. For example, if attribution reveals certain channels are being under or overvalued, you need to make sure that you will be able to shift budgets between them. Attribution solutions can be costly, so make sure that the results will be used impactfully. Questions to ask yourself:

  • Are key stakeholders for different media channels on board and receptive to taking the recommendations of an attribution model into account for budgeting and other decisions?

Data Availability and Technical Readiness
The data required for attribution takes time to collect and isn’t necessarily available right away. There are two basic methodologies for obtaining the data necessary for attribution. One involves deploying tags to your paid media and conversion tracking on your website, and the other involves using historical data that your web analytics platform (Google Analytics, Adobe) and ad partners may already be collecting.

So, are you ready to implement a tag-based solution across all of your advertising platforms? Having an ad server in place to collect data across all of your online channels will often make the implementation process run a lot more smoothly.  Having a tagging infrastructure in place prior to commencing your data collection will help you to ensure that you have high quality data, which is necessary for high quality results.

Alternatively, if the approach makes use of historical data, is it readily available from all of your media channels? It can be time consuming to collect the data necessary to run an analysis, so ensuring that you will be able to get what you need from the start will make the project run much more efficiently. Questions to ask yourself:

  • Do you have an ad server in place?
  • Do you have a tagging infrastructure in place?

Do you have access to historic data for all media channels?

WHAT TO ASK A POTENTIAL DATA DRIVEN ATTRIBUTION PROVIDER

Implementation & Data Collection:
If you decide to go with a vendor platform approach (Visual IQ, Convertro, Attribution 360 or other), there may be a step where you have to deploy tags in order to collect impression, click, and conversion data. After this, there may be a data collection period that must pass before you start to see actionable results. Your attribution provider should be able to provide insight into this process. Questions to ask them:

  • How long does a typical implementation take before we get usable data?
  • Have you implemented tags with my specific display, paid search, email, etc. vendors before? What platforms have you had success with? Have you had any challenges with certain integrations that we may prepare for?

Depending on the situation, it could be a lengthy process before you see any results. Make sure your expectations line up with when you expect to receive results.

Model options and requirements:
When someone talks about “attribution,” it is worth making sure everyone is on the same page as to what is going to be attributed. With online and offline channels and conversion points in this cross-channel world, there are many different ways to tackle attribution. Also, attribution requires sufficient data to calculate quality values: make sure you know how much data is needed to get reliable results.  Question to ask the provider:

  • Is your attribution only looking at online marketing channels (e.g. paid search, display) leading to online conversions on a website? This is the classic and most common attribution problem being solved.
  • Do you provide any integration for offline marketing channels (ex. TV) or offline conversions (Ex. In store)? Is there an additional charge for this?
  • Do you execute tracking of conversions across devices, i.e. cross-device attribution?
  • What is the minimum sample size of display impressions, paid search clicks, etc. that you require to accurately calculate attribution values?

With all of the different attribution methodologies on the market, make sure your expectations line up with what results you will be receiving.

Model Validation:
Attribution can be difficult to validate. Unlike forecasting future values (where we eventually see what the true future values are and can compare them to what we have predicted), there is no “true” attribution value to verify our values against – we can only hope that the values the model gave us are the true values! The main way to verify whether the attribution values are “correct” is to see if the recommendations from the attribution lead to lift.

The frequency at which your attribution model will be updated can also impact accuracy. More recent data is often the most relevant, but large quantities of data may be needed to get solid estimates. Questions to ask:

  • Have you previously seen lift in clients after implementing the recommendations from your attribution? Can you provide a case study or another example?
  • Do you provide any metrics on model fit, or show how your model is working to give true actionable values?
  • How often does your model refresh?

Asking these questions can help you get  insight into the success your attribution provider has seen with others, and ensure that you receive an accurate model.

Deliverables:
Some attribution providers will provide support throughout the implementation process up to and including delivering insights presentations. Verify what type of deliverables you will receive and how much support you will get throughout the attribution process. Questions to pose:

  • What level of support is provided during implementation?
  • What level of support is provided for learning how to use the platform/attribution results?
  • How are the results delivered – access in a platform or summarized with actionable recommendations in a presentation?
  • How often are results delivered?

Ensure you will receive the level of support you need to learn how to first get your attribution solution off the ground, and then to achieve the most out of your attribution results.

Conclusion:
It is important to focus in on which questions you are trying to answer, rather than to just blindly choose an attribution provider. Why are you looking to do an attribution analysis? Will you be able to action the results you get from your attribution analysis?

After this, it is  just a matter of asking the right questions to ensure the provider will be able to give you the answers you need and steer you onto the track to getting the best solution for your business requirements.  

Copyright © 2016, All Rights Reserved. Privacy and Copyright Policies.