E-Commerce Analytics is HARD. If you are an E-Commerce player or a D2C brand, you know what we’re talking about.
E-commerce businesses have multiple moving pieces - product listing on one’s own store and multiple marketplaces, ad spends, social media presence, promotions, abandoned carts, payments, order placement, order fulfillment, logistics, post-sales support - all in real-time, every single day.
It’s essential for every e-commerce business to be on top of their most important ecommerce KPIs in real-time and not manually refresh the KPI dashboards every hour to stay on top of their business. However, this is not that easy.
Research from Experian Data Quality, suggests that 56% of organizations report losing sales opportunities due to bad data, and 83% report that poor data quality impacts their business initiatives.
Inability to use their data correctly & meaningfully is one of the key reasons why e-commerce businesses are struggling with customer loyalty and experience.
Why is E-Commerce Analytics Hard
- Data Silos
Most e-commerce data lies in various places. For example: for e-commerce metrics like daily sales, you need to gather data from your own Shopify store and marketplaces like Amazon, Walmart, Etsy, and multiple other marketplaces where you have listed your products. These sales numbers then need to be reconciled with the payments you’ve received from your own Payment Gateways to the commission-adjusted payments your receive from the marketplaces. For a mid-size player, just calculating daily sales & payments requires one to coordinate with at least 10 different data sources. We haven’t started talking about the data lying in your inventory management system, CRMs, Google Analytics, Ad Accounts like Google Ads, Facebook Ads, and many other parts of the business.
- Real-time coordination
E-Commerce is a fast-moving, real-time business. As soon as you receive an order, all wheels must turn to get the order delivered to the customer asap. This requires real-time coordination between different departments. Lack of coordination can result in substantial cost wrt to customer experience. An example would be the marketing team running a promotional campaign for a certain product, without taking into consideration the real-time availability of the most popular SKU leading to a spike in stock-outs/poor conversion and a massive waste of Ad spend.
- Big Data
As an e-commerce business scales, the amount of data generated increases significantly, and even answering simple questions like the impact of Seasonality on sales can require investment in expensive resources like in-house data scientists. Most brands do not invest in data & e-commerce analytics capabilities and hence continue to run their business based on guesswork rather than data.
Common Anomalies in E-Commerce
The costliest implication of not having the right data & analytics capabilities is missing out on detecting anomalies in your most important e-commerce KPIs. An anomaly (also known as an outlier) is when something happens outside the norm.
The most common anomalies that are seen in E-Commerce businesses are:
Sudden spike or drop in conversion rate
Conversion is one of the most important Revenue impacting e-commerce KPI. There could be multiple reasons for the change in the conversion rate which go undetected before it’s too late :
- A new feature release leading to a spike in payment gateway failures
- Increased traffic on the website leads to slow loading speed and poor website performance
- New catalog push leading to broken images
- Data error leading to wrongful stock-out displays for your top products
These errors are detected too late when the businesses are reviewing their weekly KPIs followed by weeks of root cause analysis.
Anomaly Detection in Marketing Campaigns
Digital marketing is one of the most important drivers of revenue for online businesses. However, as we discussed in our previous blog, not monitoring your marketing spending and effectiveness can lead to millions of dollars of wasted ad spend even for giants like Uber & Airbnb.
The most common anomalies detected in e-commerce marketing KPIs are:
- A sudden spike in CPC for certain keywords due to some external factors
- Spike in Ad conversion rates in certain geographies going unnoticed
- Drop-in website traffic due to poor website performance or outage leading to poor ROAS
What are players doing?
Most players are all too familiar with these outages happening daily and the implications on their revenue and customer experience. However, the current BI tools are severely limited in their capabilities to provide any real-time monitoring or alerting capabilities.
In the lack of relevant tools available, most businesses are currently monitoring their core e-commerce KPIs manually or learning about it when a customer complains about social media or there is a spike in call center tickets/calls. Manual detection is often late, incomplete, and time-consuming.
- Internal development
Data-driven businesses realize the importance of real-time monitoring and invest upfront in building their own in-house analytics tools - which are often time-consuming and expensive and can only be afforded by a few players. A classic example is Amazon, which has used automated Anomaly Detection for monitoring its business process & customer experience for over 15 years now.
What’s the way forward for E-Commerce Big Data
When it comes to big data & analytics, it’s clear that the current Business Intelligence tools are woefully inadequate to meet the demands of fast-growing modern businesses. Businesses are actively looking for tools that help them monitor their thousands and often millions of KPIs at scale and be alerted in real-time for any usual trends in their business.
Chaos Genius is one such tool that’s helping D2C brands and E-Commerce companies automate their e-commerce analytics by 1) Automating their ETL processes towards consolidating data from multiple sources 2) Offering multi-dimensional analytics to automate slicing & dicing of data 3) providing multiple automated anomaly detection models to perform outlier detection and be alerted of production instances in real-time.