Question for the Expert:
How do I weight different channels when setting up attribution modeling?
Answer from the Expert:
Attribution weighting can be a tricky thing–like choosing paper or plastic. They each have their benefits and drawbacks. There is no “right” attribution model for every situation. Rather, you have to analyze your campaigns and media, test out different attribution models, tweak your campaigns, and see what variation works best. Wash, rinse, and repeat.
The standard model for most Web Analytics platforms is Last Click or Last Non-Direct Click (the last visit that wasn’t a Direct visit) Attribution. This gives weight to the channel that had the most recent impact on your conversion. While this model is great for seeing how your campaigns that expect quick turnaround behave, they’re not great for campaigns that focus on branding. First Click Attribution is fantastic for determining which campaigns introduced your users to the site but generally don’t convert them. However, without these First-Click campaigns, you might lose out on conversions down the line.
The common others are Position Based, where the first and last click get equal credit. This is important for campaigns that focus both on branding and converting. Time Decay is important if you have a sale running and need to know which campaigns have the quickest impact on your bottom line. Linear is our last common model, and it equally distributes the weight to every position in the conversion path.
Data Driven Attribution is a popular one for you Analytics 360 or Attribution 360 users. It also tends to come with the most confusion, because it’s like trying to figure out one of those puzzles in Myst without having the strategy guide. Here is the short of it: DDA looks at the last four visits before the conversion, which generally make up around 75% of all site transactions. Then it takes time to purchase into account, number of visits per channel, and the general position of that channel in any conversion. DDA will, over time, create an attribution model that is specific to your data and allow you to improve your conversion rates based on that information.
There is no one attribution model that’ll tell you the true value of a campaign or channel. The best way to go about it is to compare several attribution models to one another and decide which one(s) should be used for each campaign. It’ll take some time at first to get used to, but it’ll be worth it once those conversion rates start to rise.
Want to learn more? Check out this podcast featuring our President & Chief Search Artist, Dan Golden, as he provides insights for Attribution Modeling in 2016.
Also published on Medium.