In the ad, a fictitious encyclopedia company begins to see a spike in their web traffic and orders. The two employees quickly interpret the data as a sudden sales spike and escalate their findings to management who immediately contacts his supplier stating, “We’re back!” This leads to a chain reaction of production and shipping of the company’s encyclopedia. The ad ends with a couple watching their young child paying with their laptop and repeatedly hitting the buy now button. The ad concludes with the question “Do you know what your marketing is doing?” and a case for Adobe’s marketing cloud suite of services. The ad is quite funny and I am sure I’m not doing it justice with my description.
So what can we learn from this ad?
Before we can make a decision based on the latest data, we need to understand the data and be able to validate it. In this case, a sudden spike in sales should have raised a flag to investigate further; however, the commercial would not be as funny if this happened. Any abnormality in your data such as a sudden spike or decline could be a result of a data or system error. For example, the user could have selected an incorrect date range or applied a filter that removed some for the data. Alternatively, there could be a system issue such, which is omitting or duplicating data in your raw data output.
Whenever I am analyzing data, I prefer to compare my data to my previous results to determine if there are any trends or abnormalities. I also recommend reviewing the raw data prior to manipulating it in Excel for potential errors such as duplicate data or values outside of the normal range. Excel offers several options for assisting you in validating your data.
In the ad, the small child is clicking repeatedly on a banner ad. Let us assume that the company just released a new marketing campaign and experiences an immediate increase in sales. The company could assume that the two events are related – the basic cause and effect logic – new campaign released and our sales increased so they must be related. However, we know that correlation does not imply causation. To confirm the two items are related, the user will have to eliminate multiple variables including internal factors such as a pricing change or an improved ordering process and external factors such as a change to the competitive landscape, regulations, or scarcity of resources.
My last point is if something appears too good to be true, it just might be. If you are not confident in the data or your results, ask someone else to review the data. Another set of eyes could discover the error or provide insights on the abnormally. While the ad is funny, it is a reminder for us to understand and verify your data before taking action on it.
What are some of the tools/tips you use for validating your data?
Link to Adobe’s Marketing Cloud Service