2 min read

An interview with Prof. Kalyan Talluri, Imperial Business Analytics Fellow

As the old adage goes, “everything has a price”. While arguably true, determining what that “price” is, is fraught with difficulties. On the one hand, the right price is the one that maximises your revenue. On the other hand, in most situations it is quite impossible to determine what that is. Pricing optimisation models try to at least come close by using data and determine how different factors drive demand.

Prof. Kalyan Talluri explains: the key factors that go into the price equation are well known (i) the value of your product or service to the customer, (ii) the alternatives available, (iii) the income of the customer and, (iv) the degree of competition. However, all this is at a high level of vagueness and has to be operationalised to be useful. We somehow have to boil it all down into a single number, namely the price that would maximise revenue. So, this is difficult, to put it mildly.

What is new then?

Well, two broad developments:

  • One, we have a huge amount of data available in nearly real-time.  Data in digital form can be processed by algorithms and we can throw tremendous computing resources to extract more valuable information about the factors mentioned earlier.
  • Second, there is a fundamental change in the sale and pricing mechanisms themselves. Even a decade ago price meant price. That is, there was a single price for all, (with some geographic variation, but it was essentially constant). Now, however, every customer is really a distinct channel as sales transactions have become more intimate, even at scale.  Take Amazon as an example. It has millions of customers but it “knows” each of them, their purchase behaviour, their clicking behaviour, their interest in ads, their products and what other websites they visit. This is intimacy at scale.  


As a result of these changes, we now see prices fragmenting and becoming more customised to each client.  Of course, this is not presented as some form of discrimination, but instead as “offers” or bundles or as loyalty rewards.

However, despite these significant developments, there is still a persistent need to convert this information into a “right” price and, equally importantly, validate that it indeed is the “right” price. This process is modern pricing optimisation and continues to be a very active area of research among marketing scientists, operations researchers, economists and computer scientists.