Systematic Approach to Analyse Your eCommerce Data and Improve eCommerce Business Performance

Editor's Notes

• #2: Good morning.
• #3: What is your number one challenge in growing your eCommerce business?Is it your eCommercePlatform? Driving more traffic to the site? Expansion to different regions?
• #4: According to a recent survey to eCommerce professional across the world, the number 1 challenge in growing the eCommerce business to improve conversions, while the number 2 challenge is to attract more buyers to the site. It makes sense because if your conversion rate is not good enough, your effort in driving more visitors might be wasted and your customer acquisition cost may become too high to be justified.So, how would you improve conversion? The answer lies in systematic use of your data, and we are going to talk about it in the next 25 to 30 minutes.
• #5: A brief introduction about myself and the company.I’m Charles Chan, the founder and managing director of CLEARgo. We are an eCommerce agency in Hong Kong and China. Broadly speaking, we help our clients to plan, build and grow their eCommerce business.We are a Magento Solution Partner, so we focus very much on the MagentoSolutions.Our clients include brands such as Proactiv, Jebsen, Sino United Press, and pure eCommercebusiness such as Cartridges Direct.
• #6: Lastyear, New York Times reported how the US Retailer, Target, utilized their data to improve its business. There is a story:One day, a man walked into a Target store and demanded to see the manager. He was complaining about the coupons that had been sent to his high school daughter, with promotions for for baby products. He shouted “Are you trying to encourage her to get pregnant?”The manager didn’t have any idea what the man was talking about. He looked at the mailer. Sure enough, it was addressed to the man’s daughter and contained advertisements for maternity clothing, nursery furniture and pictures of smiling infants. The manager apologized and then called a few days later to apologize again. Instead, this time the father apologized as his daughter was indeed pregnant.OK, the story itself only tells that Target knew about the pregnancy of the girl before her father did. In fact, Target has developed a pregnancy-prediction model based on the vast amount of data collected from the customers. The ability to know that the customer just got pregnant and leverage such info can be very lucrative, as the purchase habit of women will shift significantly during pregnancy and after the baby is born. If Target can recommend the rights products in an unobtrusive way, unlike the one sent to the daughter. Target may well be able to convert them from purchasing just one or two lines of products into much broader ranges of product lines.This story illustrates how Target can leverage its data to improve their business. So, what do you want to get out of the data?
• #7: Of course, increasing Sales and Margin would be your goal. But to do so you can also look into other metrics.Take, customer acquisition cost. Obviously you want to reduce it. The customer acquisition cost from different channels can vary widely, and you would need to analyze the effectiveness of different channels and promotions to help you reduce the acquisition cost. Conversion rate and average order value is the holy grail of eCommerce business. What Target did to analyze and predict pregnancy of their customers was to increase effectiveness of its product recommendations, and thus increase conversion and the average order value.When you have a large product catalog especially with time sensitive items, inventory turnover becomes very very important. By analyzing the turnover rate of various products and what triggers purchase, we can take appropriate actions such as relevant promotions to boost inventory turnover.Another important metric is repeat customers ratio, and customer lifetime value. It costs a lot more to acquire a customer than to retain existing ones, thus gaining repeat and loyal customers is key for long-term success.
• #8: Before we can start leveraging your data, we need to be able to organize and use the data from the various data sources, as they are likely to be resided in various silos.For example, you may have many different traffic drivers, including search, affiliate program, eDM, banner advertising, etc. And in most situations, these data can sit in different platforms that are hosted at various suppliers.Clickstream data is another valuable data that usually sits in external web analytics systems such as Google Analytics and SiteCatalyst nowadays.Your eCommerce platform shall have some of your customer data, or in many cases the complete data sits in the CRM platform.And then your product and inventory data, and the transactions can sit in your eCommerce and or ERP platform.Apart from all these, you may also have data in your mobile app, social media, third-party review platforms and so on.Depending on what you want to get of our the data, you may need to integrate the data from multiple sources into one, or simply leveraging the reporting and analysis capability of individual platform.
• #9: With data organized and available to use, the first thing is to know what to measure and measure them.But to measure what?
• #10: There are many metrics that aneCommerce business needs to constantly monitor.At the top line, obviously you look into revenue and profit, and number of orders.Inventory turnover ratio is also an important metrics, it measures the liquidity of your inventory. A high ratio means your business is efficiently managing and selling its inventory. The faster inventory sells, the less cash company would tie up in stock. In contrast, a low ratio can mean you are holding up obsolete inventory that may be difficult to sell. You will need to examine further and plan what to do to move the stock again.On the traffic driver side, you would measure cost per visit, qualified visits, and so on. But the most important aspect is the channel attribution, which let you understand the role and true effect of each traffic source, instead of just the last click conversion. Two other metrics we mentioned earlier are the conversion rate and average order value. The cart abandonment rate is also an important measure on the effectiveness of the site.On the customer acquisition side, the customer acquisition cost and customer lifetime value are the most an important metrics in this aspect. The traditional RFM, which is Recency, Frequency and Monetary, are also useful metric to segment your customer. At the end of the day, customer satisfaction is also critical, especially in the social world full of reviews in your site and offsite, you shall also measure the Net Promoter Score in this front.
• #11: Depending on your technology platforms, those metrics can be available through various tools and means. Ideally, you’d establish a KPI dashboard with most of the metrics available so that you can closely keep track to.Now you have a good look at your own data, but how do you know if you are doing a good job or what aspect you shall improve.
• #12: You will want to understand how you stack up to, by establishing benchmark against the industry and competitors.
• #13: There are many sources for benchmark data.For example, Internet Retailer publishes a IR Asia 500 guide which has quite a lot of data on the top 500 eCommerce sites in Asia. You can find information such as conversion rate, average order value, and so on for each site.Some industry sites also published survey on the conversion rate by traffic sources, and also the cart abandonment rate.You can also look into Google Shopping for competitors who are selling the same or similar products, and their pricing.Sites such as Alexa, Compete and Quantcast also publish traffic and demographic information, though Compete and Quantcast are mostly geared towards north american sites.
• #14: An example of benchmark data here.Now you have your own data, and benchmark against competitors. You think there are areas you would like to tackle upon. Now what shall you do? What action shall you take?
• #16: Average conversion rate (visit to sale): 2.47% (ranging from below 1% to over 10%)Average % of visits leading to a basket page view: 5.76% (mostly between 3% and 7%)Basket page exits: 9.29% (mostly between 7% and 11%) [people who leave the site at the basket page]Overall checkout completion rate (basket page view to sale): 37% (mostly between 22% and 48%)Data collected:Analysed 2.5m visitors to over 27 websites.All used standard Magento ‘accordion-style’ checkout page (customised for each brand).All niche brands selling lifestyle goods.
• #18: In the traditional retail world, companies send mysterious shoppers, conduct qualitative and quantitative research to find actionable insights.What can we do when we cannot see your customers behind the screen?
• #19: To begin with, I believe everyone already has a web analytic tool setup, whether it is Google Analytics, Sitecatalyst or others. However, what I usually find is that most are not setup properly so that you can get the most out of it.Take Google Analytics as example, there are basic but important setup such as eCommerce tracking and goal setup for conversion funnel tracking. You can also use custom variables to track information such as member login, coupon code usage, checkout methods, etc.GA now has an advanced segmentation for drill down analysis, which is very useful but under-utilized.Multi-Channel Funnels in GA provides quite useful attribution analysis. This is another under-utilized tool.While web analytics tools provide some useful aggregated information, and it many cases you can already find useful actionable insights there. What it falls short is that it still can’t let you observe how individual user shops in your site, in the way that offline business can easily do.
• #20: There are other technologies that can give you such data. Nowadays there are tools that can provide full video recording of visitors’ full browsing sessions including mouse move, click, scroll and keystrokes.These tools can also generate heat map based on the mouse movement, clicks and scrolling information.With these tools, you may discover interesting findings on how your customers browse your site, your products, put the products into cart, when they take it out from the cart, in which step during the checkout funnel they may become skeptical and abandon the cart and etc.Based on the findings you can form strategies and actions to see how you can improve the customer experience and thus conversion rate.
• #21: In additions to web analytics and the heatmap tools, you can also find deep actionable insights through customer intelligence tools that integrate all your data sources.For example, there are tools that integrate your clickstream data, campaign data and transactional data and enable you to perform rather sophisticated analysis. You can slide and dice the data. From this kind of tools you may be able to discover further insight on merchandising, behavior of different customer segments, campaign performance, and so on.
• #22: Now you have a bunch of ideas on how you might improve the conversions. But can they really improve your business result?When we work with our clients, say we have identified that there may be some issues with the product detail page and we have multiple ideas to improve it, what shall we do? Sometime our client would ask us which option is best based on our experience. While we may have an opinion which option is a better one, most of the time our answer would be “Let’s test it”.
• #23: The testing technique we are talking about is called A/B and Multivariate Testing. You may compare A/B testing to a clinical trial, only instead of doctors, patients and drug treatments, it’s marketers, end users and site variations, and no one gets hurt or dies in the process.
• #24: Take a very interesting case from the Obama 2008 campaign.At Obama first election, digital channel was one of the most important channels for his campaign. And his campaign website was the primary tool to get email sign-up.Throughout the campaign, they have collected approximately 10 million sign-up through the website homepage. This wasn’t possible without intensive A/B testing.The team that was responsible for the Obama website shared what they did in a case study. At the beginning the site started with simply a key visual and sign-up box. What would you suggest to make it convert better?
• #25: What the team did was that they would test on various call-to-action buttons, as well as the key visual, whether it is a static image of him, in the crowd, or with the family. Or different videos.The campaign team mostly favored the Sam’s Video, but what turned out it was the one of the worst performing.
• #26: As you can see, the different buttons and media yield very different result. The best combination received 40% increase in conversion, while the poorest video had a negative 30% yield.
• #27: At the end this one works best with 40% increase in conversion. Meaning that if they didn’t run this test, they would have received almost 2.8 million less email addresses.Overall, 10% of those sign-up became volunteer, that means that they would have lost 280,000 volunteer if they didn’t run this test.And also, in average they receive $21 donation for each email sign-up, that means they would have lost $60 million in donation if they didn’t run this simple test.
• #28: With such experience, the Obama campaign put up more vigorous tests during the second campaign.During the 20-month process of raising US$ 250 million from the site, they conducted over 500 tests.This include simple test such as updating headline or images, or complex tests such as new donation checkout flow. At the end, they managed to increase signup-rate by 161% and donation by 49%.
• #31: http://www.smashingmagazine.com/2010/11/24/multivariate-testing-in-action-five-simple-steps-to-increase-conversion-rates/
• #32: http://webdesign.tutsplus.com/articles/general/a-web-designers-introduction-to-ab-testing/
• #33: A few tips on testing.Let go of your ego and let the data speak for itself! Instead of subjective opinion, tests give you the most objective results.Trust the data, not the HIPPO which is the Highest Paid Person’s Opinion, which is usually wrong.Focus on high conversion triggers, changes that potentially create most effects.And test continuously.
• #34: A quick summary, at first we discussed about measurement and KPI to keep track of your own key metrics. We also have talked about using Benchmarking to know how you stack up, and identify key areas for improvement.Using various analytics tools, we analyze and obtain actionable insights. And lastly perform testing to get on-going improvement.You can drive grow only if you do this continuously.
• #35: In additions to the framework, here are a few more tips on using your data for business improvement.
• #36: You have different types of shoppers and they all behave differently with different preferences and triggers. One way to segment your customers is to identify their personas. For example, you will probably have very different strategies for fashionistas vs. value seekers. Another way to segment your customers is based on their lifecycle, whether they are first time customer, active, at risk or churned, you will need different strategies and tactics for each group.There are technologies to help you segment these customers, and perform automation for e-mail campaigns, and A/B testing different offers for segments.
• #37: Another tip is to deploy a recommendation engine, especially if you have a broad range of customer base and wide product catalog.Real time recommendation engine creates individual profile based on clickstream and purchase history, it allows personalized product recommendations and cross sell recommendations.It is a simple concept, and in the past only very large site such as Amazon can do it. Nowadays there are quite a lot of Cloud-based solution and implementation of such solution has become rather easy.One of our partners, Emailvision, has the stat from their Predictive Intent product that such solution can achieve:● 35% increase in conversions● 14% increase in attributable revenue ● 16% reduction in click to buy distance
• #38: Lastly, you really should put effort in recovering the abandoned cart.Most sites have a cart abandonment rate over 70%, meaning that for every order you received, there are a few other people added items to the shopping cart but never complete the checkout process. There are many reasons for the cart abandonment, with unexpected shipping being one of the most popular one.Winning back the customers who already have purchase intention is indeed a low hanging fruit. There are two techniques. One is remarketing campaign, basically through advertising network such as Google Display Network, you can target customers who have abandoned the cart with specific win-back message.Another technique is a follow-up email. Usually an urgency message with the abandoned cart in the email works quite well. From our partner SaleCycle, in average you can get over 3% increase in total sales by simply adopting the follow-up email tactic.
• #39: We have mentioned a number of techniques throughout today’s presentation. Below are some of the tools we use.
• #40: As a summary.
• #41: Hope this presentation provides some useful information to you.Thank you for your time. Please feel free to ask any question and connect with me. Here is my contact information.