Loading...
As ecommerce professionals, we live and breathe data in search of actionable insights. The key is to align data analysis with business outcomes, which is no easy feat. Have you ever tried sitting down and mapping out exactly how you look at data? What do you do when diagnosing a sales dip vs. identifying growth opportunities? From personal experience, teammates often have different approaches that result in drastically different conclusions. In this post, I will attempt to explain an approach that is both objective and easy to follow.
Consider this post a whirlwind tour of ecommerce data analysis. In future posts, we will dive deeper into each one of these questions with some real-life examples.
Before initiating any analysis, there are five key questions to answer:
For the rest of the post, I’ll go into each one of these questions.
Before diving into any analysis, first understand the problem you’re trying to solve. Not aligning on the issue at hand is how divergent conclusions occur. There are typically two buckets of inquiries — diagnosing a dip in certain metric(s) OR identifying opportunities to push growth.
Knowing which problem needs solving simplifies your decision-making process. Each problem area has a unique decision tree you can use to get to insights. Why is a decision tree important? Think of yourself as a doctor and your data as the patient. Upon a patient’s first visit, they most likely don’t know what their illness is, only the symptoms. To correctly diagnose the illness and prescribe a treatment, it's crucial to ask the right set of questions.
What is the data trying to tell you? Without a roadmap, it's easy to get lost. having a decision tree is like having a roadmap you can use every time. It takes the thinking out of it and allows you to get to the visualization that provides the right clues right away.
Start by retracing and documenting the steps of your last successful analysis. Which metrics did you start with? How did you break it down? What time periods did you compare against as benchmarks?
Once your decision tree is complete, build a dashboard that helps you visualize them all in one place. While overwhelming at first, it can save you valuable time and clicks once you're used to seeing all the data at once.
All metrics can be broken down into more granular and actionable sub-metrics. The key to understanding how to influence a metric is to understand its key sub-metrics.
For instance, below are some of the fundamental formulas almost all ecommerce businesses recognize:
The list goes on. Educating your team on how metrics are calculated and their influence on different components of the formula is a powerful insight. It tells them that their actions can have multiplying effects on the bottom line — a huge source of empowerment.
Transactions, Revenue, AOV, and Conversion Rate are table stakes for any ecommerce business. These KPIs, however, are too high level for most operators to action on. For instance, how do you influence Conversion Rate without hurting AOV?
One approach is to use a composite KPI like Revenue Per Session (RPS). Such KPIs capture the complex relationships between bottom funnel metrics. Simply put, RPS is the combined performance of Conversion Rate and AOV. Have a winning experiment that results in higher AOV at the expense of Conversion Rate? No worries, focus on whether it had a positive impact on RPS.
Composite KPIs are also great as a team-wide North Star Metric. In future posts, we'll explain composite KPIs more in-depth.
Before a transaction occurs, a few other actions must take place. The user must visit your site, maybe become a lead, view the Product Details Page (PDP), add to cart, and initiate checkout. These metrics are much easier to assign ownership to. A PDP redesign should directly influence Add to Cart numbers; adding a lead capture modal should increase Lead Capture Rate.
The owner of these upper / mid funnel activities should be well educated on the metrics they influence and monitor them daily.
Dimensions are attributes that describe your data. There is an almost infinite number of dimensions for any metric, such as device category, source/medium, city, and time of day. Luckily, there are a finite number of dimensions most growth professionals look at. Knowing your go-to dimensions is what will set you apart from the rest.
Let's say you're diagnosing why Transactions are down. The first question to answer is whether it's a traffic issue or a site issue. Device category is always a good starting point. Mobile is down while desktop is flat? Break it down by browser type and see if any one browser is having issues. It's not uncommon for a key UX component to be broken on a specific browser (especially in-app browsers). Source/Medium is the next suspect. For instance, if "facebook/paid" is down while the rest is holding steady, the culprit is likely a shift in your Facebook campaign setup.
The task of growth is ultimately about beating your past performance, which is why you need a solid benchmark. A trustworthy historical benchmark is a period when everything was working as intended, without the influence of a shopping holiday or site issues. Although rare, reliable industry benchmarks are also helpful.
Relevancy is the similarity between two time periods. Consumer behaviors vary throughout the year, month, week, and even time of day. If sales were low on Wednesday this week, use last Wednesday as the benchmark. Immediacy means the period closest to the period you're looking at. Most often this just means the day/week prior. As long as you understand the intra-week cycles of your business, you can deduce what the expected variance is.
One caveat is to skip the special shopping holidays, such as Black Friday. These days have special meanings for consumers and the performance will differ greatly from the rest of the year.
Before initiating any analysis, there are five key questions to answer:
Answering these questions is key to achieving actionable insights and aligning your team on the same goals. In future posts, we will cover each one of these questions and concepts in-depth.
Extending our analytics reporting infrastructure to support B2B customers.
A webinar focused on improving LTV by conducting intentional custom-centric analysis.
