Data attribution in online performance marketing
Key figures: “Select Home Warranty, which provides warranties to homeowners for household repair projects in the United States, saw a 36% increase in leads and a 20% decrease in cost-per-conversion after making the change to DDA.”*
Attribution is a very prominent often neglected feature in digital advertising.
Why is Mobile, as a device, getting so much traffic but rather lower conversions compared to desktop? Why are my generic campaigns on Search engines recording poor results compared to the Brand? What is the exact contribution of each traffic source to the final number of transactions? Basically, which Touchpoint deserves the medal?
While answering such questions, keep in mind that what you see is not the full picture. Every possibility is to inspect carefully, from a Brand to Generic analysis to cross-device as well as a multichannel attribution through last-click, Data-driven attribution & all.
Today’s article will dig deeper into Why & How you can assess a conversion path and which features allow you to do so. The purpose is to help you keep your eyes away from single attribution models and consider multipoint conversion players.
APPROACH & TAKEAWAYS:
Why is data attribution so important? Simply because it impacts important decisions. Whether you are optimizing, reporting or analyzing campaigns. If you only have a limited understanding, your decision will surely be inaccurate. As consultants in Digital Performance Marketing, we use attribution tools to understand which elements are truly responsible for your conversions both directly and indirectly. Decisions of bidding adjustments, budget allocation and other optimizations are then implemented and a set of actions taken based on key findings.
Brand to Non-Brand:
How often do we hear our clients say: ‘Brand campaigns on Search are generating most of the transactions, generic campaigns are not doing well, let’s cut down the budget on generic’’?
Think about it! A user may search on Google Search for ‘’gluten free snacks‘’, interact with a generic Ad, visit the website and leave, coming back on Google Search, the latter would search for the brand itself ‘’The Protein Bakeshop’’ (since he kept it in mind from a previous visit), click on the brand Ad and purchase. From a ‘last click’ perspective, the conversion has come from the Brand while in reality, the generic keyword ‘’gluten free snacks‘’ has initiated the path to conversion.
By adding a label to your Branded campaigns vs. non-branded campaigns or creating a custom Channel Grouping on Google Analytics, you can run a simple report and understand the contribution of each group to the conversion. You will be surprised to see how often Generic campaigns are undervalued.
Another very interesting question is: How many users search on Mobile and end up buying on Desktop and vice-versa? We generally observe; especially in the region, a majority of traffic coming from Mobile devices with low conversion rates. Reason is often being a non-optimized Mobile site but that’s another subject. As of now we will focus on attribution. The first thing to keep in mind when it comes to devices is whether you are reading cookie based or user based data. If you are on Facebook for instance, you are Andre on mobile and on desktop, as opposition, if you are on a search engine and not logged in, your path will be lost as soon as you leave one device. Using Google Analytics, you will naturally be tempted to minimize Facebook performance since GA doesn’t record cross-device activity from Facebook. Which takes us to the next question: what is the percentage of lost conversions that happen cross device but are not recorded on Google Analytics? Hard to tell accurately but we are still able to generate an impression to conversion report on Facebook, cross vs mono device and match the data found with GA. At best, you get very decent data to compare; at worst, you get the percentage of cross device activity from Facebook and can estimate an incremental that can match the data on with backend tool.
Using the Multi-Channel Funnel tab on Google Analytics, you can choose the attribution model that you desire and read the data based on the latter. Direct conversions as well as assisted ones can be put together for a bigger picture (duplicated care to keep in mind). You can also choose the path length for the most straight-forward conversions and choose to consider a longer look-back window.
The future of attribution is most probably data-driven-attribution, A new model available on Google Stack that helps us understand which element from our accounts are driving conversions taking into consideration the good as well as the bad ‘elements’. DDA works best with smart bidding since the latter uses the learnings patterns to better optimize against your target CPA/ROI.