I have tried recently to connect the dots between Pricing and Big data. Opinions expressed here are personal and obviously limited, that’s why I’d love to get your comments.
As with other functions and sectors, recent advances in Big Data are disrupting the discipline of pricing analytics & insights. More particularly, data blending, i.e. the capacity to merge a variety of data sources in real-time, and data visualization, i.e. the capacity to create and share pricing analytics results and dashboards, are among the main game changers.
How to take advantage of data variety in Pricing
Pricing is at the confluent of two types of dataflow: transactions on one hand, and value on the other hand. The vast majority of Pricing initiatives either concentrate on one or the other aspects, but very rarely on both. At the same time, the maturity of solutions offered by software vendors and consultants is rather unbalanced, in favor of transactional pricing, i.e. solutions to analyze pocket margins at a granular level and reduce the variance of their commercial terms and conditions.
Variety is needed in Pricing to contextualize decisions and feed the different touchpoints of the pricing process. As an example, elaborated predictive analytics can be run by a retailer on POS data to derive elasticities and cross-elasticities, but only existing price points will be tested, and the findings will be disconnected from value perceptions. Models can find easily targeted price increases, but the impact on visit/traffic, price image and existing thresholds is more complex to assess. Transactions and Value should be blended into one holistic approach.
What are the key steps in data blending?
Variety is much more important than size for Big data to impact decision making. But how to get there? Data collection is a major problem in pricing. Data requests from IS are often difficult, even with dedicated APIs or ETLs. But solutions to access data have evolved in recent years. What we call data blending – the process or workflow designed to build a dataset from multiple sources – is different form data integration in the sense that it results in a non-permanent database, in other words a dataset built for the purpose of supporting analysis for a specific business question. It is therefore less demanding for legacy systems, data warehouses etc. as it captures only the relevant data.
Three types of data sources are relevant for Pricing: 1) Traditional data: ERP, CRM, spreadsheets, etc.; 2) Enrichment data: segmentation & research, spatial, weather, etc.; 3) Emerging data: social media, web, analytics, etc.
How to successfully visualize pricing data and improve decision-making
As per the above, the common grounding of any pricing analyze should be traditional data, i.e. transactions. This space is well occupied by traditional software vendors that have developed very powerful solutions to analyze pocket margin and pricing fairness & efficiency. But the journey to get there is often long, complex, and expensive.
Building a waterfall or a dispersion chart with transactional data at a granular level is much simpler now than it was some years ago. Of course, visualizing the data does not mean to impact business, but this is the foundational steps towards pricing maturity that many companies have being avoiding with excuses around data access and tools which are no longer valid.
As data and analytics democratization is underway, the challenge now for Pricing teams is to empower their organization harnessing multiple and real-time pricing data and communicating consistent pricing analytics to their respective stakeholders.