Improve online ROI - use the Finance toolkit


Marketing, in particular online marketing and eCommerce, has made tremendous strides in focusing on performance. Professionals in the area of eCommerce share a quantitative skill set that is more comparable to that of finance professionals than of marketing professionals of old. Much attention, or at least lip service, has been given to using ROI as a measure of effectiveness of marketing spending. But how do you derive ROI and how does this measure and many other eCommerce KPIs interrelate?

I needed a comprehensive overview of this subject as introduction to a review I was writing about research of what works in eCommerce. As I could not find this overview online, I decided to write this blog. I draw on my background in Finance to highlight how the ROI breakdown from the Finance toolkit helps to identify actionable improvements and how to apply the method meaningfully to eCommerce.

a thorough ROI analysis has helped me time and time again to identify operational and managerial improvements 

ROI analysis, also called DuPont analysis where it originated in the early 20th century, allows the results of a business to be compared across various years, or against other businesses. In a previous life, a thorough ROI analysis has helped me time and time again to identify operational and managerial improvements by following the money; following the financial outcomes back to the choices and practices that led to the outcomes. I’ll outline the mechanics of ROI tree and its branches and leaves and then move on to the artful part: how to compare and interpret comparisons.


Let’s start with ROI. A useful way to define Return on Investment is by dividing the profit you got from an investment by the size of the investment. This way, you can compare various types of investments, businesses or projects. Now, imagine you are a physical store selling physical goods and you ask yourself how can you increase ROI? Well, the simple answer is to either increase your profit or decrease your investment. In reality the two parts cannot be moved independently of each other. You need to break the ROI down in the components that mirror the business mechanics that drive profit and required investment. For a physical store, you need to invest in inventory. That inventory you sell at a profit and part of the proceeds you use to replenish inventory and the other part you retain as profit. To increase profit over investment, you can either increase your profit margin or increase sales. Higher sales means that you can keep investment in inventory constant but earn the profit more often. Say you can increase annual sales from 4x the inventory to 5x the inventory. Your ROI will increase from 10% to 12.5%. Note that: 


Similarly, ROI of an online venture, channel or campaign can be broken out and parts that can be compared, analyzed and influenced.


In online business, the common driver both of revenue and costs are the number of visits during which goods are bought. Therefore, online KPIs are typically expressed per visit. Conversion is an important metric and multiplied with average order value, you get the revenue per visit. Note that costs of goods sold – most suppliers will charge you for the goods they have supplied you – are many times absent in online KPIs. Even though online professionals rarely influence the costs of goods sold, they can influence upsell, but need to understand if that will improve profit margin, not only order value.

You can choose any type of headline result and any path to breakdown, as long as the math is sound, e.g.


An online business that relies on repeat transactions with each customer or periodic renewals, e.g. insurance, may opt to focus on customers rather than visits.


Just remember, you want to work with KPIs that:

  1. are meaningful to track your business and to compare yours with others;
  2. point to operational decisions you can act on.

Lets take a moment to discuss both points.

The power of a ROI analysis is in the comparison and understanding how business drivers and decisions work their way through to headline results. In the ‘old days’, the biggest limitation was really to make sure your comparison proved causality. Unless you had many stores or had the fortune to have two stores in the same city, it could be rather difficult to argue that a change in marketing lead to higher sales rather than a turn of the economy in general. With online business, if you have a sufficiently large customer base, you can do A/B or multivariate tests in which every condition is kept constant except for the one you are testing. By the way, there are many ways to screw up A/B tests

The second point is about ensuring that your analysis leads to something actionable. That is the key of the breakdown. Unlike your boss who only moans about underperformance, you can actually articulate what can be done about it. This determines both in which components to break down and how far to break it down. If results are obvious, you are probably not looking at the right layer of granularity. When all you see is that water is much higher twice a day, you are observing the tides but not the waves. An electronics store that sells both $2000 TV screens and $10 cables needs to device KPIs to avoid clouding the performance of the more profitable cables. Average order value can be broken down further into number of items and value of items. That way, you can compare the performance of merchandising efforts to promote cables together with the TV.

The intention of this write-up was actually to introduce a review of research about what works in e-commerce - a meta-analysis of 6700 online experiments. This introduction has grown a bit beyond its original purpose but hopefully it is a useful standalone blog about online ROI analysis. I will post the link here to the blog as soon as I finish the review.