Category: Ecommerce | Solution: Datama Compare & Detect, Analytical Solutions | Type: Recurring | Client: Lacoste
Tags: #Conversion #Alerting #AnomalyDetection #Monitoring
Context: Weekly review for a major e-commerce website
Lacoste is a well-known worldwide player in the fashion industry. Among all their distribution channels, the e-commerce website, Lacoste.com, is an important contributor to the performance of the group.
As such, closely monitoring and improving the performance of their website has long been the crucial ongoing task of the Digital Data Analytics and CRO team.
Regular reporting and appropriate governance have been put in place to do so, in addition to sharing outcomes within the organization.
One example of communications is the “gazette” sent out to all EU stakeholders on a weekly basis, comparing performance to the previous week and the previous year, including all steps of the conversion funnel.
However, previously, preparing this “gazette” took a significant amount of time to dive deep enough into the analysis with the appropriate level of insights to make it actionable. Many KPIs and dimensions needed to be considered, and even though the analysis had the same approach every week, it never had the same answer.
Therefore, the team implemented “Datama Compare” to automate their weekly analysis and went one step further by leveraging “Datama Detect” to identify anomalies and issue alerts on a weekly basis, enabling a faster response rate to issues.
Approach: Breaking it down by steps of the marketing funnel
The “market equation” is basically the mathematical and business link between the Primary KPI (aka “North Star Metric”) of your performance and the sub-indicators.
In the classical case of an online conversion funnel, the North Star metric is the Revenue generated by the website, and it is basically the result of the product of each step’s completion rate in the conversion funnel.
A dataset for Datama is a structured table aggregating selected metrics on a set of appropriate dimensions.
In terms of metrics, the dataset to be plugged into Datama obviously needs to have all the steps of the market equation.
In terms of dimensions, the goal is to monitor performance over time, so Date is essential. The Lacoste team also wanted to be able to explain performance by a set of dimensions such as Country, Device, and Browser.
We end up with the following dataset (obviously, the data has been randomized and anonymized).
The data originally comes from Google Analytics, but to facilitate automation and extraction, the data is pulled from Big Query on the appropriate period of time with the Datama connector in Datama Prep
Once plugged in Datama, the team was immediately able to visualize and compare performance from one week to another in a fully automated way in Datama Compare.
The picture below is a good example of an increase of traffic driven by France, but which has a negative impact on conversion and especially on PageViews/ Sessions, due to a Country mix effect, because France traffic usually has lower engagement rate than other countries.
This use case was then set up to be sent out automatically by email to all stakeholders on Monday mornings, to facilitate the preparation of business reviews.
In addition, the Lacoste data team wanted to leverage Datama Detect capabilities by monitoring those same KPIs on a daily basis and go one step further than weekly monitoring by being able to proactively act on fixing main issues.
With the very same dataset and just by switching to the “Datama Detect” solution, the team was able to spot outliers in any of the performance indicators within the market equation defined above. By activating emails alerting on a daily basis, the team is now notified when one of the steps of the funnel goes wrong, and most likely also see the reasons for that change within all the dimensions of the dataset.
In the screenshot above for instance, the traffic goes below expectations based on historical observations and the most likely reason is a change of behavior in desktop traffic.
Using Datama, the data team was able to provide its internal clients with precious time savings and an uplift in the quality of the conversion performance monitoring. The e-merch/product team gained confidence in the fact that it was able to quickly identify key drivers of underperformance and therefore act faster to ultimately increase the business generated by the website.
The data team has replicated the approach for the US region in addition to the EU, and is now planning to expand the usage of DataMa to new use cases, in particular for category performance analysis using the DataMa extension for Looker Studio.
Datama’s market equation is the perfect answer