What Mark Twain Can Teach Us About Ecommerce

What Mark Twain Can Teach Us About Ecommerce

“There are lies, damn lies and statistics.”

Most often attributed to Mark Twain, this quote could be modified just slightly and applied to the world of ecommerce. If Samuel Clements had been born 150 years after his actual birth in 1835, he might have opined in a more modern fashion:

“There are data, damn data and analytics.”

We really emphasize to our clients how analytics can improve their ecommerce businesses. FastPivot has developed more internal resources like our Google Analytics Home Run service. We preach every time we talk to retailers about how much of a difference numbers and percentages can make to building their bottom lines. Larger companies are often more receptive simply due to having more resources to analyze data, test hypotheses, draw conclusions and make longer-term changes. But smaller companies, too, should understand that a little bit of data mining and analysis can make the proverbial world of difference. (Here comes the caveat).

Back to The Guy With the Moustache

The relevance of the 2st century version of Twain’s commentary on the veracity of data is particularly true for ecommerce analytics. Just because the numbers are in front of you, doesn’t mean you need to like them or attribute more import than they merit. The data may be correct but the conclusions drawn from them may be faulty. Or the data may be incorrect and your conclusions are equally incorrect. Here are three examples:

Don’t Set Yourself to Fail

This applies in so many areas of life. Make sure your data collection processes are set up correctly. I can’t emphasize this enough so I’ll say (copy and paste, actually) it again. Make sure your data collection processes are set up correctly. Ecommerce tracking is all about behavior and attribution. If you’re looking at your bounce rates, make sure that you’ve configured/enabled cross-domain tracking. If it isn’t, someone can land on one domain, click through to another (perhaps a registration page) yet be recorded as a bounce. Test your recorded transactions against your inventory/sales management system to ensure that you’re recording every transaction. Use IP filtering for site traffic to ensure that you’re not recording traffic from employees or a partner’s office, for example. #ConfigureThenConfirm.

Beware of Ecommerce Whales

This is a problem encountered far more often by smaller online stores. When you have fewer total sales, the “importance” of each order is much greater than at a large Internet Retailer Top 500 site, for example. We’ve seen with some of our smallest clients that large orders, or repeat orders from a single customer can skew the overall picture. Large orders and dependable volume from these “ecommerce whales” is great, but make sure you’re aware of the atypical nature of this sales volume. Average order value will be skewed. Two options are to remove outliers when calculating averages, or use median data to give you a more realistic perspective.

Compare a Macintosh to a Slightly Changed Macintosh

I’m not talking about silver laptops; l’m talking about doing real, disciplined A/B testing. A/B testing means altering a single variable. (If you’re going to change more than one, that’s multivariate testing). Let’s call it “almost apples-to-apples” testing. The reason you need to be disciplined and patient is that you want evidence/proof of causality. Go slow and test one thing at a time. When you see behavioral changes in your research group compared to your baseline, then you can attribute the cause to the effect.

Data is your friend. Just ensure you’re gathering and analyzing the data that will help you make the decisions that will grow your business. Don’t get distracted by shiny objects among the numbers. If you have any question about how to use data more effectively, give us a call.

Clip to Evernote
Read More About: ,

Leave a Reply

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>