MQnomix® – Consumer Products Analytics

 

Are you an online Entrepreneur? Do you sell products? Then fight for your products with SAS!

I don't know much about successful ad campaigns these days, but I do remember one that honed in on men who should fight for their ...hair, while they still have it. Obviously, this should be a matter of principle. As entrepreneurs, fighting the good fight for perennial values, especially in these tough times, could keep our sanity and wealth in check.

Let's face it: extreme impact events have always "changed the world forever". And since they are rare enough, having a low frequency of occurrence but also being able to generate a huge impact, we can call them "Asymmetric Impact Events". They are rather different from "Black Swans" and have nothing in common with similar notions found in linguistic science. In effect, they convey that there's no need for 10,000 (similar) events in order to get an extreme impact, one or two would be enough. Therefore, they can and should be considered asymmetric and paramount when it comes to dislodging societal equilibria.

And yes, having a long history of ignoring signals, we have no solid contingency planning. That's why, when talking about the negative impact of asymmetric events, people always had one option: learn fast (in the aftermath!), constantly adapt (while learning) and have the upper hand at implementing the "emerging model". Those getting quickly but thoroughly through all of these steps are usually part of the winner's pack.

Fast-forwarding to our present time, we came to realize that there is no substitute for social distancing and digitalization. For most businesses, and in particular, for small and medium-size ones, digitalization became a matter of survival.

Albeit important, establishing an online presence by just checking the box won't do the job. Imagine the myriads of firms that are joining the ever-growing online club and the ensuing hyper-competitive environment generated by this process.

Moreover, it is well known that big online platforms are winning big, but that does not automatically translate into a spillover effect for the majority of players. No wonder that entrepreneurs are left with quite a few unanswered questions.

For example:

how important are the ratings for the product's popularity (for the platforms where such a sales-related index is provided)?

what changed in the online competitive landscape, in terms of who are my competitors (compared to the brick and mortar)? Who are my direct and indirect competitors? Are they the traditional ones?

what other products are in the same cluster my products belong to? Does the finding make sense?

what discount policies do my (direct /indirect) competitors apply over any stretch of time?

what are the most important observed variables related to the products sold in the segment the Entrepreneur is interested in?

based on these variables, how do my products stack up against those of my competitors?

In the classic Brick & Mortar setting, we can hardly have access to information in a timely manner. The process is cumbersome at best and the outcome of product & competitor analyses are easily becoming obsolete. The point is that making decisions based solely on past behavior and data is pernicious.

However, the seemingly ubiquitous online setting is a completely different ball game. It provides abundant information and, if properly used, it represents a truly decisional pillar. Consequently, proper answers can be found to previously asked questions.

After using specific tools that amass relevant product data, we can use PCA (principal component analysis) to determine the minimum number of variables that should be used in cluster analyses. In the "Proportion" column (see table below), PCA determines the proportion of the variance explained by each component. For example, the first three components explain 77% of the variance in the data.

Eigenvalues of the Correlation Matrix

perform (product) cluster analyses, by means of unsupervised learning algorithms,

determine the groups of products having similar characteristics with those from your   portfolio,

find the direct and indirect competitors,

get the pricing & discount policies of competitors.

If only two variables at a time are used, the resulting graphs tell us a partial story about clusters' relative position. For example, we can get clusters position in regard to "Rating" and "Discount" or "Rating" and "Power".

However, if all relevant observable variables are used at once, as they should be, we'll get the entire picture of competitors' landscape. Below it is a so-called, Panel graph. It displays the average values of all relevant observed product variables. Moreover, we can determine which are above and below average in each cluster.

What else ?! Well, by choosing any of the clusters from the Panel above, you'll find what products are in that particular cluster. First and foremost, two types of clusters would stand out: those of your direct competitors and ...your own.

But there is more! It makes sense to learn some lessons from these graphs in conjunction with others, like profit or market share for the same time period used for the analysis. Draw your own conclusions, time and again, and learn. After a while, you'll learn which scenarios are more favorable and under which circumstances. Then, (try to) replicate them...Sometimes we'll get it right sometimes we'll not.

But definitively we'll learn a whole lot more in the process! 

And one more thing: in which cluster you would like to have your products? Would you be better off in Cluster2 or in Cluster6? Why ?!