Holiday Cyber Spending: Numbers in Use Are Shaky

Pick your number and take a chanceMore details are coming out about spending on Cyber Monday. Although the event took place over a week ago, the numbers are interesting if for no other reason than they suggest how consumers are shopping and buying. Know how people make decisions, and you are closer to understanding what they might want from you. But it turns out that the sources of information aren't necessarily something you could statistically project to the buying public as a whole, and many executives may be making marketing, advertising, and promotional decisions based on unwarranted assumptions.

What we had heard before was that people wanted to shop, at least in the let's-go-online-and-hit-sites manner. A number of major retail sites found themselves out of commission at some point during the day. That much seems pretty safe to observe.

But how much buying? That is a contended point if you examine at least two data sources that many news reports are using. Chase Paymentech's Cyber Holiday Pulse Index count had sales up less than one percent over last year in revenue, though it counted the number of transactions as up by 14 percent. But according to comScore, e-commerce spending on Cyber Monday jumped 15 percent to $846 million. That's some big difference.

As both companies are using samples, not getting complete reports from even all major online retailers, the difference is likely in the sampling methodologies.

Chase's index, which surveys 25 of the largest 150 retailers on its Internet payment processing network, showed that the average shopper spent $7.19 less per transaction on Cyber Monday this year over last -- reflecting deeper discounts than usual, thriftier shoppers or both.
In other words, there is some set of retailers that uses Chase to process payments and it probably does some sort of projection based on this information. On the pro side, it gets extensive data from at least some retailers and it should get all transactions through them. The con is that the information wouldn't be representative of all purchases being made online. For that, you'd need to have a sample of all those purchases, not only those made through a select subset of vendors, because people don't buy things in a mathematically random fashion. With a biased sample, you can't project the results. That means any statement about how much was spent is highly questionable.

That leaves comScore as another major source of information being quoted in the press. I directly emailed a PR representative for data on the methodology and got the following in response:

We have a 2 million person panel who give us permission to observe their online behavior. Our data is projected based on their behavior, so we cover all online retailers from which we observe transactions by our panelists.
That is a sizeable sample, but probably a self-selecting one, which means possibly not representative of overall purchasing. Additionally, I'm betting there's a chance that many of this group would be in the early adopter category, in which case it would be even less representative of all consumer activity. Furthermore, it raises the question of why the people are allowing this kind of monitoring. Are they getting something in return that might bias the sampling? On the plus side, it should be easier to do a year-to-year comparison among the group.

In short, there's no way of telling who is right or wrong and how much was really spent on Cyber Monday. The potential bias in these numbers should bring to question the reliability of any statements about e-commerce traffic or spending totals made by either company. High tech companies -- or those in other industries -- must bring a salt shaker when making decisions on where and how to market, particularly in a down economy in which every dollar has to work hard.