eCommerce Analytics Guide, Getting Started.
October 13, 2020 | By .
All roads lead to Rome. When it comes to eCommerce analytics, there are almost too many available routes. You might be juggling insights from multiple service providers like Shopify, Facebook or Google Analytics; number-crunching manually in Excel; or bending and twisting tools like Mixpanel or Supermetrics to fit your unique business needs. These routes are time consuming, error prone and costly as you scale.
At 173Tech, we believe data should be easy and actionable for everyone to help drive more sales, keep customers happy, and reduce costs.
In this post, we will show you how to get your analytics stack up and running quickly that helps you grow in the following areas:
- Ability to centralise and analyse data from all customer touchpoints.
- Acquire new customers cost-effectively.
- Increase sales and customer lifetime value (LTV).
With an agile and scalable infrastructure in place, you can then model additional data sources and build algorithms to boost your own data-driven competitive advantage.
Kickstart: Data Warehouse + Sales Data + Reporting Tool
Your full stack will continue to evolve, as pictured below. To build a quick and solid baseline and start generating analytics ROI, we recommend the following steps:
- Choose and set up a data warehouse most suitable for your business.
- Automate eCommerce data extraction, e.g. Shopify sales via API.
- Model key sales components in your data warehouse: orders, transactions, customers, products, stores and refunds.
- Connect a visualisation tool to your data warehouse and build an automated Master Dashboard. Metabase is an open source solution that is quick to set up and robust enough to satisfy your early stage reporting needs.
- Create a data dictionary that captures standardised metric definitions with both business and technical logics, ensuring accurate and consistent reporting. Keep it updated as your business evolves.
Once you have this solid foundation in place, you can plug and play any additional data sources seamlessly. The ultimate goal is to connect all relevant information for the most comprehensive view over your customer journey, and to build data science solutions to optimise it at every single step.
The proposed kickstart project typically takes teams at 173Tech between 10 and 12 weeks depending on your data maturity.
Organise Sales Data
To get your sales data organised quickly so you can easily analyse it through multiple lenses, follow these steps: automate data extraction, transform and visualise.
Automate eCommerce Data Platform
Extract order level records with all the details. Most eCommerce platforms provide an API where you can plug in and pull raw data at your preferred time interval (e.g. daily) into your own data warehouse. If you built your own eCommerce solution, we can connect directly to the operational database and automate data load.
Transform Revenue Data
Next connect all the data dots and arrange them in a meaningful way so you can easily build and tailor reports from any angle. Below is a sample data diagram of key fact (e.g. Orders, Order_Products) and dimension (e.g. Customers, Products) tables linked together by unique identifiers (e.g. customer_id), which will be the bedrock of your reporting.
You might have noticed some columns in blue which are not directly available from raw eCommerce data. These are examples of modelled fields that can dramatically increase your analytics capabilities and speed.
Integrate Cost Of Goods Sold (COGS)
A simple way to model COGS is to record it in the Products table then model into Order_Products, so you can easily aggregate and manipulate in the same way as sales. For more sophisticated COGS tracking, check out our eCommerce Analytics Playbook, details at the end.
With revenue data in place, you are ready to build automated dashboards that are tailored to your business and easily digestible. Start with a Master Dashboard covering key growth indicators for a quick and up to date bird’s-eye view. It will evolve as you integrate more data sources.
Increase Customer Lifetime Value (LTV)
Customer LTV is the most crucial intelligence for any eCommerce business. It can be used to allocate and optimise digital spend, build high value audiences, segment and tailor CRM strategies and more. To increase customer LTV, optimise on its two key components: basket value and repeat purchase rate (aka customer retention).
Analyse your basket data to get a good understanding of what product(s) customers are buying and the distribution of spending. Then try the following ideas.
Examine currently purchased-together product groups for any discoverability issues, if obvious cross-sell opportunities are missed.
Build algorithms to optimise the display of relevant products dynamically.
Create A/B (or multivariate) testing funnels on:
- Copies describing the benefits of multiple products working together.
- Discounts on product bundles or up-selling to a larger size or quantity.
- Timing for cross-sell and up-sell recommendations. We usually see this being right after a product is added to the basket or during check out.
- Timing for prompted discount codes. Try postponing it to later stages of the buying funnel. If a new customer is willing to buy at full price, discounting it too early will sell yourself short.
Repeat Purchase Rate
The more frequently customers come back and purchase, the higher their LTV. Analyse the distribution plot of how long it takes for a customer to make the next purchase. Also examine if this duration is shortening over time or among cohorts, and what percentage of customers never made a second purchase.
If the repurchase pattern is fairly consistent among your customers (e.g. every 3 months), when do they become ‘sticky’, i.e. customers are loyal and their behaviour is predictable? Identify this loyalty moment and you have found your North Star.
You can also borrow the retention concept from app analytics to track customer retention over time by cohort. More details in our previous post.
Providing a subscription service is a great way to increase customer retention and LTV. It saves customers time and effort going through repeated buying processes, and ensures a more predictable stock level and revenue forecast for eCommerce businesses.
If you operate a monthly subscription model, retention analysis and churn prediction and prevention are your top priorities after customer acquisition. Your North Star metrics are undoubtedly:
- Monthly Recurring Revenue (MRR)
- Month X Retention, X being the point that correlates to customer loyalty and long term stickiness.
Your modern analytics stack ensures a consistent and accurate view on these critical KPIs over time.
Predict Customer LTV
Based on historical data, you can build prediction models on customer LTV at the moment of their first purchase. The more details you have on your customer (e.g. acquisition, demographic, purchase details etc), the more accurate your results. Among many use cases, predicted LTV can be used as an early signal to optimise your digital spend.
Integrate Marketing Data
To optimise digital spend towards LTV on all campaigns and channels, automate data extraction from marketing platforms via API connection, and unify tracking standards across all channels with UTM parameters.
From here, you will be able to allocate spend data to individual transactions based on UTM attribution, and apply LTV prediction at transaction level and aggregate for campaign ROI. You can also build algorithms to generate optimisation recommendations at your desired frequency to guide marketing budget allocation.
A comprehensive discussion on maximising marketing ROI is available in our eCommerce Analytics Playbook. More details below.
Now you have an agile and scalable data foundation, you can integrate more data sources such as marketing, inventory management, and customer support; and start building data science models for customer segmentation, retention and churn predictions, text analytics etc. If you want to read more on these topics in our full version eCommerce Analytics Playbook, or have a chat on data in general, please ping us on email@example.com.
Remember, your analytics possibilities are endless!