Why use KPIs when you can use KVIs? We asked our Head of Data Intelligence Matt Mayes about value-based design, a new approach to data analysis that gives businesses a better look into why you are seeing specific results, and how to improve. It focuses on Key Value Indicators (KVI), which go a step further than Key Performance Indicators, giving businesses a clear path to act on the data mined. Read the full interview below:
What is the general way most businesses are collecting and acting on data?
Matt Mayes: Most businesses use a few different tools for collecting data about their business. For their ecommerce channel, they would likely use a web analytics tool (i.e. Google Analytics, Adobe SiteCatalyst, Coremetrics...) to collect data. They would also likely have a Customer Relationship Management tool to pull together customer data from across their different channels. In all honesty most businesses do a really poor job of acting on data. In most cases they may spend a few times a year mining through the volumes of data to extrapolate out a couple of insights that helps with site optimization and/or redesign.
Why is this not an effective way of collecting data?
It really comes down to data overload. Unless you are using sophisticated predictive analytics, have a large staff of data analysts or are paying external vendors or consultants lots of money to analyze and implement changes, you likely aren't benefiting from the mountains of data you sit on. Simplicity, although it might not make sense in the age of Big Data and with all the thousands of reporting tools out there, it makes it really easy to report, analyze, take action and measure performance.
What is a new approach that could help solve some of these problems?
Taking a top-down approach to analysis vs. a bottom-up approach. Most people take an analytics solution and force fit it into their business and try to map the KPIs in those systems to their business. A better approach would be to define all the areas of your business you would like to assess because you know they are critical components of the consumer's journey, then figuring out what measurement allows you to best analyze that performance.
How does Value-based design solve these issues?
From a reporting perspective, you are building dashboards and visualizations with the intention of having tangible takeaways as a result of how they are performing. Rather than building a bunch of reports that show flat metrics and fact telling, you build a report that is ratio based, time series and correlated. So when you view that report you know immediately what I did last week didn't work, I need to know go make XYZ change and measure. Moving away from Fact Telling in your reporting towards Action Oriented Reporting. If report reads this I know I need to change A, B or C and if it reads positive I need to continue with doing A, B and C.
Key Value Indicators (KVIs) are a new way of measuring a business. What are KVIs and why are they different from Key Performance Indicators (KPI)?
An easy way to think of this is there are hundreds of metrics and ways for a business to measure their "performance". But not all of those metrics or indicators are things the business had direct control over. For example, things like page views, clicks, time on page... These are all dependent measures which have dependencies on consumers doing other things. A good KVI would be something like engagement rate where I care about the consumers or viewers I have on a page, how much I can engage them in things like add to cart buttons, reviews, image zoom. Rather than trying to get more page views I want to focus on those page views I have and help get consumer to the next step of their journey. I can enlarge my add to cart button, move it around, change the colors to get higher engagement.
How do KVIs fit into value-based design?
You first have to start by understanding all the key value indicators for assessing your business. This requires a business to perform a holistic business assessment and come away with areas you want to improve, map those against what KVIs you can use to measure those and building corresponding reports & dashboards that help you visualize those KVIs. Doing so by allowing you to filter and correlate against all the necessary variables to properly analyze that data. (i.e. time dimensionality, cohorts and segmentation, benchmarks...)
What is a scenario where a business could gain more actionable insight from a KVI compared to a KPI?
A good one that I always reference when working with ecommerce companies is focusing on Cart Abandonment or Step Checkout Abandonment over Orders, Revenue or Conversion Rate. You may not always be able to get more people to convert, but you can ensure that those who intend to convert do not abandon. You can work on optimizing specific page elements to reduce abandonment rates and in turn, will likely have a positive impact on things like Revenue, Orders and Conversion Rate. But simply saying I want to increase conversion rate, being that it's a dependent measure, it's hard to simply say I can do XYZ to get high conversion rate.
How would a company go about identifying the top KVIs?
It all goes back to what things you are hoping to improve or measure performance. In retail, specifically ecommerce, that is really about measuring the consumer across their lifecycle. KVIs should be properly aligned to measure those things (Acquisition, Engagement, Conversion, Retention). New customer acquisition rates or Channel specific ROI measurement, on site Engagement behavior for core elements around research like search rate, bounce rates, continuation rates. Mapping these out against each of those four phases.
Once a business identifies and collects the KVIs, what is a good way to optimize and measure performance?
Starting with how to measure performance, it's best to always set proper context before defining what "good" performance really is. The best way I have found to do this is through baselining and benchmarking. Look at historical KVIs to assess W/W, M/M and Same Month Prior Year against Current Month, then looking at industry performance. There are tons of sites that offer industry, vertical and subvertical benchmarking. This way you can understand what bad performance looks like, what good performance looks like and what great performance looks like. It then allows you to start those correlation exercises around what the "great" performers do that make them great versus what I do. Then to optimize, the best way is through incremental testing. Test early and test often. This is the only true way to measure causation (what you did had direct impact). I caution people to be methodical: you don't have to test everything right away, nor does your first test decide your go forward plans. Sometimes it takes eight, nine, even tests on the same element to find that sweet spot to fully optimize.
Do you want to start optimizing on your data today? Learn more about Welcome's data intelligence offerings.
About Matt Mayes
Matt is the Director of Data Intelligence at Welcome. He has extensive experience working with large ecommerce retailers, growing overall business including topline revenue, channel specific revenue and contribution as well as offline revenue.