Successful workshop on Promotional Modelling and Forecasting

Julian Bridle (SAS) presenting at the LCF workshop in London on promotional modelling.

20 October 2015

The Lancaster Centre for Forecasting (LCF) organised a successful and highly stimulating workshop on "Promotional Modelling and Forecasting". This was held on 7 October 2015 at BMA House in London and was free to attend for all practitioners and academics in forecasting and marketing analytics. 

The workshop was formally opened by Professor John Boylan, followed by an introductory lecture by Dr Gokhan Yildirim (LCF), who highlighted some of the challenges in marketing analytics and the difficulties in identifying causal relationships.

The first speaker, Julian Bridle (SAS), highlighted the fact that demand was becoming more volatile and unpredictable as promotions became more frequent and more varied. To address this problem, John Spooner (Head of Analytics at SAS) advocated the following principles: i) get scalable solutions, ii) manage by exception, iii) have automatic exception monitoring, iv) incorporate events and other demand drivers, v) build up extensive modelling capabilities and vi) automate your solution as much as possible.

Next, James Oates (Nielsen) stressed the importance of promotional modelling, given that spend on food and drink had risen from 25% of total spend during 2000 – 2008 but had risen to approximately 33% from 2009 onwards. He also discussed the complexity of the “path to purchase” (eg a customer may visit a store before buying on-line). James recommended measuring all marketing at the point of influence which should increase accuracy, granularity and flexibility of measurement. Wherever possible, it is desirable to work at store level rather than at higher levels of aggregation.

The final speaker was John Dawson, from MarketingQED, who explained that his company’s goal was to provide software which included features such as regression analysis and machine learning but did not require command line programming. John discussed a case-study with a UK insurer, bringing together internal and external data sources and replacing time series with causal forecasts. This led to the Marketing Division being recognised within the company as a centre of excellence for forecasting.

The workshop concluded with a panel discussion chaired by Professor Robert Fildes (LCF) in which the four speakers responded to various questions form the audience. There was a lively discussion on the difficulty of recruiting staff to positions in Marketing Analytics, and the scope for greater co-operation between companies and universities which specialise in business forecasting.


Companies represented at the workshop included AG Barr Soft Drinks, Bear Nibbles, BT, Cambridge University, Centrica, Epson, Federal Mogul, GfK, L’Oreal, McBride, MarketingQED, Milner, Morrisons, Nestlé, Nielsen, Openreach BT, Primimrose Consulting, R Twining, SAB Miller, SAS, Specsavers and Lancaster Centre for Forecasting.