Data and analytics are a must for running an e-commerce business – it’s been like that for decades, but in today’s world where we collect so much data, harnessing the data is even more crucial. There are a lot of good systems, components, services and tools that all claim to “drive more business”, or “boosting your conversion rate” etc. Most of them will deliver on the promise; but only if you make a roadmap and strategy before you start.
You need to be able to measure what works and what doesn’t – and that’s hard if you don’t have the right strategy and roadmap lined up.
Running an e-commerce business is complex, with a diverse set of challenges– there is no “one size fits all” approach on how to succeed. With that said, here are some of my recommendations that can help you gain the much-needed insights and awareness.
1. NOT ALL DATA COLLECTED IS GOLD. CONSIDER DATA SANITIZATION
Data is just not always reliable. In a world where decisions are made from numbers, dashboards, graphs, trends – invalid data can be dangerous.
It’s hard to validate every single piece of information, but make sure you establish a way to validate your data. Validating data and cleaning data creates the foundation for good analytics, ML and AI. This is rarely mentioned in our world of fancy ML and AI based e-commerce and web systems. The old saying "you can’ t make good system based on bad data" seems to be forgotten too often; Although it’s common sense that good data is a foundation for any good decision making. Remember to keep your feet on the ground, and consistently questioning the validity of what you see and get from the data. Some products, like CluedIn have made it their business – to gather data, clean it and match it, and finally put some metrics on how good the data is (Accuracy, Quality, Completeness, Validity).
2. START WITH YOU OWN DATA
It’s often seen that a ton of different components and analytics is added. We need “User Driven Content!”. Let’s get some integrations for that social network, product reviews, automated personalization, various AI & ML tools etc. That is all good, and sure some of all that will work in some cases, and in some cases not at all.
My basic recommendation is to start with the facts – the data you already have. This will depend on the specific context and situation – the line of business – is it a startup, for example? But focusing on all the valuable data you already have is a logical first step, but it is easily forgotten with all the fuss and buzz words on the smart tools that exist. First things first.
- Analyze you customers profiles, match them to your personas (create some if you haven’t!)
Use the data for basic personalization. You most likely have meta data/profile data that is specific to your site (type of bike, indoor/outdoor, favorite events etc.). Use this for basic personalization, and base it on “your specific data”, that is where you can differentiate best.
- Analyze customers and historical orders.
You can gain knowledge using data from previous orders. Establish segments of users based on simple metrics like (selected examples)
- 80/20 rule: 20% of the users make 80% of the purchases.
- Who buys only when discounts exist
- What discounts works for what user segment/persona
- Data and time of purchase
Data for your analysis can be from various sources, to mention a few:
- Orders from web and/or point-of-sale
- Checkout funnels
- Persona matching and profiling
- CRM related data
- Support and customer feedback
- User profile data
Analyzing your own data is where you dig up the gold. It will bring out ideas and goals for up-sell, cross-sell and other sales boosting strategies that you can try – and best of all – these would be strategies that you can measure. If you use paid advertisement, it’s crucial to be able to measure advertisement cost to the actual purchases. The same goes for other obvious channels like email marketing, product recommendations etc.
3. MAKE SURE TO OWN YOUR DATA
Most components you use on your site, like search, recommendations, reviews, social feeds will most likely generate data for analytics and personalization. The question is who owns this data? And is it free to utilize? What if you switch out a specific component for something else – invest in a preplacement or supplement? – who owns the data then? And is that data in a format you can re-use?
Not saying this is a problem, but you should be sure of who owns this data and if there are limitations on data usage.
Example: An e-commerce website uses a 3rd party email marketing system for campaigns and for transactional emails from the website. The 3rd party system obviously includes analytics to be able to track opened emails, click-throughs etc. All that is good – and expected.
But the value in the full flow from the user opening the email, click on a link, and end up placing an order – fully tracked, has a much greater value. To be able to do that the tracking data must be collected and matched, using the same analytics, or the of data integration to get the full overview.
The example is classic in the sense that businesses buy a bunch of different components, best-of-breed, and end up with fragmented overview of the analytics and lacks important view of the user journey. Don’t underestimate the value of the full data overview. If you don’t have that overview you will miss opportunities. An example could be that your using a product review component on your side; your users clicks, reads and writes reviews, but you do not get the data back into your 360 view of your customer. Valuable information that could change to actions when you reach out to those customers on any of your channels. Bottom line – make sure you have a plan and strategy to “connect all the dots” – a plan towards the full data overview. Using all this data, an especially your own data, is where you can make a competitive difference. That said, time to market is important and you most likely won’t have a full overview and needed integrations in place from day one. But make sure you’re able to do it soon and have a plan.
Always remember it is better to have lesser but quality data than lots of data that doesn’t bring value, or data you don’t have the rights to make the full use of. No matter what, you need a single view of the customer, with all data that brings business value. And you need to be able to execute on that data
4. SET GOALS AND MEASURE YOUR INITIATIVES - ALWAYS MEASURE - HOW ELSE WOULD YOU KNOW WHAT WORKS?
Obvious – but not emphasized enough – always measure when you try something new. Check the data – before/after and use A/B testing where possible. You need to setup metrics, so you know how you are doing and compare it to the various initiatives you’re doing.
- Conversion rate
- Bounce rate
- Abandoned cart rate
- Web traffic based on sources
- Average basket size
- Lifetime value of customers
- Order sizes based on
You should setup metrics that is specific for your business and personalized for your goals. Without metrics you have no control at all – it’s like driving forward aimlessly and without an opportunity to course correct.
Focus, focus, focus – and then focus some more!
- One thing at the time
You can’t do it all at once and making lots of changes at the same time makes it impossible to measure success. Be sure to focus on the lowest hanging fruit first – and on the right channels you know you’re have been doing good on. And remember #1 thought #4. You need to have a strategy and success metrics.
- Start analyzing and questioning the data you have from day one. It is the best information you have, and the only “truth”.
- Use the results from your data analysis to target users, personalize information, campaigns etc.
- Ensure you gain more and more data from your users that can help you in the next steps, such as, A/B Testing, Site Search analysis, Ratings and Reviews etc.
- Ask the users how you’re doing
Remember you can always ask the users how you’re doing – and you should use the results/feedback you get.
Focus on the quick wins, don’t let yourself be pulled in to only new channels and ideas without remembering your core channels that has a proven success. It is never either or, but be sure to constantly re-evaluate your actions.
6. DATA ETHICS
Ok, that sounds boring, I know! If you haven’t read “Ethics and Data Science” (O’REILLY) by Mike Loukides, Hilary Mason & DJ Patil, then do it. It’s short and free and bring up some discussions about data and privacy that is highly relevant. The book also outlines a set of questions, that you as a product or system owner should ask yourself about data and privacy. In Europe, with GDPR, it’s not only a pressure, it’s a law, and before you know it, a similar law will be enacted all over the world. This is a good thing, and if you’re prepared for it, it might even be a potential for your business.
A FINAL NOTE
There you have it, the 6 basic recommendations on data-driven e-commerce. For sure, there are many more, but this is where you could start. Good data is the foundation for excellent business decisions for your data-driven e-commerce.