“In the business world, the rearview mirror is always clearer than the windshield.” – Warren Buffett
For newcomers to ecommerce, part of the learning process is coming to understand the basic statistical framework online retailers use to measure the performance (or non-performance) of a given web store.
The core aspects of a business’ performance that can be measured and reported in numbers are known as the company’s “metrics” or “key performance indicators” and the tool or spreadsheet for capturing and reporting those metrics on a regular basis so that they can be reviewed is known as the “dashboard.”
One of the more satisfying aspects of running an online business in comparison to an offline business is the relative ease of capturing a very wide range of data from sources such as Google Analytics, along with the store’s own internal reporting tools, though the value of building out a dashboard comes from the way that data from different sources can then be related to each other in a single easy-to-comprehend format and location.
Frequently I find that online retailers have a good understanding of the data coming to them though a single tool or channel, but do not have a clear picture of how the data coming to them through various channels relates to each other. Bringing key data from different sources together in one place addresses that, and opens up the potential to see opportunity in the relationship between data from different sources that might otherwise be missed.
Several years ago I was involved in a project where we brought together a massive amount of data from three separate information-rich sources, which up to that time had been viewed largely in isolation from each other. In that instance the sources were QuickBooks, Google Analytics, and a proprietary order management system that the company had designed which had a robust reporting platform. The result was a dashboard that took the form of a single set of spreadsheets combining and relating data from those sources to make analysis and decision-making easier and more reliable.
Reporting dashboards can serve another important function, as well. One danger of having such a rich field of relatable data coming from a variety of sources is information overload. This can take the form of creating metrics that don’t actually reveal valuable and actionable insights, but which do clog-up your reporting tools and make it hard to see the really useful information through the blizzard of less-valuable numbers. Really valuable reporting and management tools are often developed over time as particular metrics are added and subtracted from the dashboard based on their actual use in decision-making.
When I’m designing reporting dashboards for ecommerce projects the particulars of each tool might change a bit depending on the goals, the richness and accuracy of the information sources, and so on. But in general the data captured falls in a few basic categories.
Sales and Income Data
The point of retailing online is to sell product, and in the end much of the relational data you generate will be based on sales figures. Capturing the total sales in dollars, total number of transactions, and total number of items sold in a given period is basic to much of what follows. Many retailers break out their product sales, sales taxes, and shipping fees to provide more insight on where the revenues are being sourced. Other data to consider might include the portion of the total sales that are due to things like personalization and product enhancements, rather than just sale of the basic physical good.
From this data alone you can easily compute your average order value, and you can compute similar figures for things like average shipping cost. This data can be of value all by itself but as we will see in a minute it can be useful to a larger purpose as well.
Often sourced from Google Analytics or another similar platform, traffic data refers to the basic metrics of website usage on the part of your shoppers. Google Analytics makes it easy to capture data that relates to characteristics of your visitors, including information about the technology that they use to interact with your site, how visitors found your site, and how they use your site.
The number and percentages of users by channel (paid, organic, direct, referral, social) the overall and channel-specific bounce rate, the depth of engagement of visitors (measured in terms of time-on-site and pages per session), their preferred technology, including browser and operating system, the amount of traffic from mobile devices and the type of mobile devices used, the number of unique vs. repeat visitors, along with other key traffic metrics, can provide a wealth of insight as to how your store is being discovered and used by shoppers.
Although Google Analytics can provide conversion data itself, the conversion reports are not always accurate, but they are easy to generate yourself from the mix of sales data and traffic data. At the most fundamental level, when people talk about your site’s “conversion rate” they typically have in mind the simple formula of total sessions divided by total number of transactions. But you can produce other conversion-related metrics as well, if you have the data. For example, you might decide to try to measure the conversion as it relates to the number of unique visitors as opposed to session-based conversion.
If you purchase traffic through pay-per-click advertising with Google or Bing/Yahoo, then you’re probably very concerned to see that your ad dollars are being spend intelligently and in proportion to what they are generating for you in terms of sales.
The Google AdWords tools, along with Google Analytics, can provide you useful data that you can also harvest and relate to data from other sources. As with total sales, the basic number here is your total spend in the study period by advertising channel.
Using this data, along with traffic reporting having to do with visitors by channel, you can easily determine what you’re spending on average to generate a visitor to your site.
Much of the hard work of advertising intelligently has to do with really understanding the differences in terms of quality and conversion between visits to your website from different keywords, the costs associated with getting visits from those keywords, and the average order value generated from those visits. But all of that goes beyond the sort of overview data that a good dashboard is likely going to be able to generate for easy digestion.
Rather than go down that road, use metrics to take a longer look at what is going on. Setting aside the details of specific ad groups and campaigns, focus on metrics like (a) the average spend to generate a new user of the site, (b) the average spend to generate a transaction, and (c) the average spend as a percentage of the average order value overall. Although the particulars of what to do about a malfunctioning campaign cannot be deduced from that level of data, an overall awareness of the effectiveness of your paid marketing in relation to other study periods can be determined.
One more thing about all of this. Almost everyone who sells online is getting their traffic from a mix of paid and unpaid sources, and those percentages are changing all the time. You can track the return on investment of your advertising relative specifically to the visitors that your advertising brings to you, but you can also treat your ad spending as simply a general expense and ask yourself – ignoring the particulars of the mix of paid and unpaid sources – what you’re spending on average in the general area of marketing expressed as a percentage of your average sale.
Shipping Cost Metrics
Shipping is a direct expense, like the cost of goods sold. But you might be using a mix of shippers and packaging materials to fulfill orders. Shippers make it very difficult to relate their bills back to shipments made in a particular period, but if the study period is sufficiently large enough (like a month), then the data on your costs will get more accurate just because the pool of packages is large enough to make the data average out a bit. Costs per shipping channel, costs per shipping method, number of packages per channel or method, average packaging costs — these can all be captured fairly easily and added to the dashboard.
At that point, you can quickly see how to figure out if your shipping charges are covering your shipping costs, and also figure out your shipping charges and packaging charges as a percentage of the average order value.
Building a Business Model Around Average Order Value
I’ve brought up the matter of average order value a few times. I’m a big fan of this metric, in part because I use it as a central tool in building and testing business models. If you subtract from your average order value, your average cost of goods sold, your average shipping cost per order, and your average marketing cost per order (treating all orders as if they were the result of paid advertising even though your real sales were probably a mix), then you will see your gross-profit-per-order after all pure variable expenses, treating advertising as a pure variable expense.
That’s the amount of money per order you have to apply to your fixed costs or overhead. If you know that your total overhead is $3,000 per month, and that each order contributes $50 after all pure variable expenses (including ads), then it will take 60 orders per month to get to break-even. If you can get only one order on average per day using that business and advertising model, then the business isn’t going to work.
I’ve left out a number of other areas where you could try to harvest data and relate it to data from other sources. The details of building, testing, and improving your dashboard are, as they say, “left as an exercise for the reader.” But having a centralized reporting framework that provides you a clear and comprehensive basis for comparing your store’s performance from period to period is a critical part of running it professionally.