Web analysts, from those working for third-party agencies to those working in-house, dream of overcoming pain points and to make web analytics a money vehicle. However, the “true north” of next-generation web analytics begins with in-depth product-level attribution models and continuous interaction, customization, and optimization activities.
“Attribution model” is originally a concept in social psychology, referring to how individuals explain causes of behavior and events. That’s exactly what we are dying to know from our web analytics practice, isn’t it? Why our visitors purchase online, what happens before they buy, and how to spark another impulse to buy – they all rely on “interactive, customized, and optimized” attribution models of web analytics.
L Group, one of the world’s FMCG giants, decided it wanted to further study the purchasing behavior of online customers. So, L Group adopted an international tool worldwide and a local provider for its local web analytics engine in China. Using these tools, L Group is now able to realize how those highly qualified customers visit their e-commerce sites right before making a purchase, including traffic sources, search queries, etc. The team raised abundant valuable findings – all based on data. For instance, branding keywords within paid search campaigns always own the highest return on investment.
However, both conversion quantity and ROI decreased, once L Group only launched branding keywords at a time. Using web analytics tools, marketers were able to figure out the conversion dynamics among “source” keywords, “bridge” keywords, and keywords directly leading to final conversion. The team then allocated different weights on different keywords groups with different price and other competitive factors, so that L Group has helped its sub brands benefit the most from paid search campaigns.
The best practice above is one of the “attribution model” applications, and I am very proud to say, as SEMPO Greater China president, that more and more companies in the region are adopting this methodology to optimize their paid search campaigns, including Microsoft, Coca-Cola, and L’Oreal.
And we can do more, moving our attribution model forward from one channel analytics to multi-channels. Tracking of cross channels campaigns is essential in gleaning insights about your online visitors and optimizing the ROI of those seemingly disparate display ads, eDMs, paid search, even your self-produced IWOM video online. But without interactive, customized, and optimized tools and systematic work methodology, you will definitely find your team as a walk on the beach.
As we all know, Google Analytics naturally relies on last-touch campaign attribution, which means that both transactions and conversions are attributed to the most current traffic source of the visit. Although there have been lots of excellent tutorials written about different solutions to overcome Google Analytics’ limitations, setting it up for multi-source attribution can be somewhat complicated. It’s a bit of a steep learning curve, so make sure you get on your web developer’s good side with some coffee and doughnuts.
Marketers in Greater China region and around the globe are expecting to have those brilliant attribution models well embedded in a solid web analytics platform, through which they are able to:
Attribution models come from traditional psychology research, even before the Internet emerged. Therefore, we should integrate both online data and, more substantially, offline data inside the models, to more easily interact, even for traditional or junior marketers.
It is our dream to build our own paid search models, own display ads models, own rich media models, and integrate them within attribution models platform as a whole, which highly relies on customization. The more customized, the more potential the system works well.
Models are not our final destination, but results after optimization are. Armed with complicated tools with a high cost is not our goal, but we need professionals who are capable of handling the process. We truly hope the realization of such web analytics’ true north.