Generating Actionable Insights from Sales Analytics

The product life cycle is a concept pervasive in marketing and product management, and is basically a function describing the rate of sales of a product over time. There is plenty of literature available on the internet about product life cycles, relieving me of the need to describe it further. However, let it be said that the product life cycle is never as pretty in real life as it is in theory!

One of the most interesting points on the life cycle curve is the point at which the curve exchanges Growth for the beginning of the Maturity phase. This important point is called the point of inflection, a term you may recall from elementary calculus. I will not go into this further in this article though.

For most banking product groups (i.e. the set of all cheque account types in the bank versus a particular cheque sub type which may actually be new to the market at the current point in time), you will typically find that they oscillate between the growth and the maturity phase, and sometimes even the decline phase, with customer interest in a product continually renewed by means of periodic product tweaks and marketing campaigns, especially in mature, competitive markets.

Again, per a previous post, while this particular case relates to banking, it could just as well be applied to the telecommunications and retail industries, indeed, anywhere where a product or service is sold to many customers, B2B or B2C, and appropriate data is collected. 

---

The client: A top global bank
The problem: Demonstrate that the value of the data warehouse was beyond mere reporting
The solution: To help visualise the impact of marketing for banking products per a request from the marketing department

The data warehouse: Oracle
Data quality: Balanced and audited, with an analysis for completeness and cleanliness
The data set: Five years of product acquisition and transaction data for six millions customers

Analysis tools: Custom

The process and findings:

Using simple calculus, we can define the following:

Figure 1: ƒp(t)

• ƒp(t) = the product life cycle function for product p, where ƒp(t) is the point a product p is at on its life cycle curve at time t, practically interpreted as the quantum of sales at time t. 

From this chart, it can be seen that it is not yet clear where on the s-curve the product is.

Figure 2: ƒp'(t)

• ƒp'(t) = the first derivative of the product life cycle function, which practically speaking, is the rate of change of  sales. 

From this chart, we can see that there is no discernible pattern looking at the rate of change for each unit time. 

To gain more insight, we could group the data to align with the frequency and duration of the marketing campaigns of the bank. In this case, the data could be grouped by half-year, given that there is generally only one major marketing intervention for the product per year during the first half of the year, as has been the case for many years. For five years of data starting from January 2008, this would mean we could divide the data into ten groups to December 2012. 

Figure 3: A different perspective on ƒp'(t) using intervals

The results are very interesting, as encapsulated in Figure 3.

Group 1 & 2: Jan - Dec '08
Growth is flat, perhaps even declining in the second half 

Group 3 & 4: Jan - Dec '09 

There is growth around the marketing campaign, flattening off during the second half of the year

Group 5 & 6: Jan - Dec '10 

There is a more marked increase in growth around the marketing campaign, dropping off during the second half of the year

Group 7 & 8: Jan - Dec '11 

There is growth around the marketing campaign, dropping off during the second half of the year

Group 5 & 6: Jan - Dec '10 

There is strong growth around the marketing campaign, dropping off strongly during the second half of the year

Note that the strategy of grouping data is but one approach in the toolkit of the experienced data analyst! Other techniques could have been deployed. Indeed, more than one technique should generally be deployed in practice to verify your findings, but it is not necessarily the only way to achieve validation. 

OBSERVATIONS and RECOMMENDATIONS

1. There is generally a period of growth around the marketing campaign in the first half of the year, dropping off during the second half of the year. This shows that marketing is working, but it is also clear that the market is much too competitive not to follow through with the campaign into the second half of the year. The marketing calendar for this part of the business needs to be redesigned with this in mind.
2. The growth exhibited in group 9 is fantastic, and could be indicative of the quality of the marketing campaign delivered at the beginning of 2012, perhaps encapsulating the lessons learnt during the previous years, with a touch of something extra :). For further insight, an alignment of these findings needs to be made with other data sets, such as customer satisfaction and qualifiers such as changes in the competitive landscape. It is also important to consider the costs of the campaign incurred during 2012, and whether those costs are justified, keeping marketing ROI in mind.
3. The sales decline exhibited in group 10 during the second half of 2012 is extraordinary, demonstrating that the market is getting increasingly competitive, where any lifting of the throttle means that sales will be penalised. This finding complements the finding and subsequent recommendation made in Observation 1

In closing, it is clear that the product is mature, as expected and as the analysis shows by means of the oscillation between growth and decline. What is interesting is that the market is also saturated, yet there is a very interesting rate of growth in the product driven by the annual marketing campaigns. It is also clearly highly competitive, because as soon as a campaign comes to an end, competitor strategies come into play, negatively impacting sales.

It is clear that marketing has a critical role to play in sustaining the growth of this product in a competitive, mature and saturated market, but it should be reconsidered and extended, being sure not to take an eye off of changes in marketing ROI in the process. 

I hope documenting this case has been helpful and informative to you. If you ever need help with your analytics projects, organising your data, developing algorithms or even simply measuring the ROI of those projects, please email me.