Cross Device, International, Custom Attribution: Analytics and Data Science Case Study
This case study demonstrates how Search Laboratory provided continuity in attribution reporting from our client’s historic data, and made it customizable for future reconfiguration.
The client is a boutique luxury fashion retailer based in London. With over 25 years’ experience in high street retail they have grown substantially in recent years, using their digital asset to service over 170 countries worldwide. Digital marketing forms an integral part of what drives this company’s growth.
Understanding their customer journey
As digital marketing is such a key component of the client’s business strategy, and luxury fashion can be a much more considered purchase than fast fashion, it’s important that they understand their customers’ journey. Understanding which channels their customers came from meant that they could use it to work out where to spend their marketing budgets and achieve optimum ROI for their spend.
For many years, the client used Rakuten DC Storm for their attribution. It used a model that assigned scores based on where a certain channel appeared in a user journey.
Migration to Google Analytics
Due to several reasons, such as product familiarity, extra functionality and growth of website traffic, the decision was made that the client would migrate all their analytics tracking and reporting to Google Analytics 360. This product would provide a more accurate and robust solution going forward, however this caused reporting continuity issues, as this was a completely different platform to what they were using previously.
The aim was to attain an understanding of the full picture, in terms of new customer acquisition and repeat customer purchase behavior. The client needed to see not only the full path to conversion but also how this browsing behavior played out across multiple devices.
Although the Analytics 360 user interface already has an excellent data driven attribution solution, it could not provide the continuity of reporting that mimicked what was produced in DC Storm for a couple of reasons:
- Using the UserID view, Analytics 360 does provide cross-device reporting, however a user must be logged in for this to show up in the UserID view. This meant that any sessions prior to signing up or logging in were not visible in the reports.
- Using a standard view (which uses the ClientID to define a single user) allowed the client to see the full path to conversion on a single device, however this did not include sessions by the same user when browsing on a different client/device.
The client engaged with the Google Analytics 360 implementation specialists at Google, however they were not able to solve the problem. The internal development team at the client’s headquarters couldn’t solve the problem either, so they then engaged with Search Laboratory to execute an appropriate solution.
Search Laboratory were able to export Analytics 360 data to BigQuery and unify the ClientID with the UserID, using a 30-day lookback window. This meant that the full user journey prior to sign up or login was visible for each user, whether it was cross device or not.
Search Laboratory were then able to build the previously defined attribution models from DC Storm into BigQuery. These were then ran against the data on a daily and weekly basis to produce a table with attribution figures for each channel.
The final piece of the puzzle was to visualize this data from BigQuery using a customized Data Studio dashboard. This dashboard consisted of a top-level overview page, followed by subsequent individual pages specifically for each acquisition channel in detail; broken down by country and by device.
The Business Benefits
The overall solution allowed the client to have an up-to-date true view of channel performance, based on definitions set by their previous attribution tool. This also allows them to react better, redistributing advertising monies accordingly based on this performance.
The BigQuery calculations were also engineered to look at a Google document to provide the attribution model weightings. This allowed the client to have full control to amend channel/position weighting as they see fit, without any reliance on development and changes to the calculations in BigQuery. They can now optimize and fully control their attribution model to maximize profit.