5 Ways to Improve Your Marketing ProgramOctober 7, 2013 By James Walker
1. Break Down Organizational Silos
Typical barriers to optimizing marketing spend spring out of the separation of functional responsibility and siloed analytical tools for assessing advertising versus CRM activities. What's needed is an integrated approach for combining all customer communications into a single evaluation model. KLM/Air France is a good example of a company that merges its research and analytics functions to combine research-based insights and behavioral data-driven insights. This creates a seamless view of understanding the market, consumer motivations and acquisition as well as experience, retention and loyalty analytics.
2. Embrace All Data
While a Google search for "big data" and "marketing mix modeling" yields little commonality, and on Wikipedia, they seem to live in separate silos, the biggest turbo-charger of marketing optimization modeling is the wealth of new data that can be included in these analyses. Many companies are now taking what was originally "engineering data" and re-purposing it for marketing optimization. For example, some telcos are looking at calling pattern data and data usage, etc. Big Data is going to transform marketing decision-making, but what we mean by data can come from non-traditional, non-marketing, sources.
3. Combine Techniques
Marketing mix modeling traditionally uses aggregate data inputs like TV GRPs or print spend to try to explain aggregate outputs like total sales. But now what we see are increasingly granular inputs and outputs being included in marketing mix modeling. Why not combine aggregate marketing mix modeling with disaggregate customer analytics? The best practice now marries the comprehensiveness of marketing mix data looking at the totality of marketing spend with what were previously seen as CRM techniques looking at disaggregate customer level data for acquisition, retention, up-sell, ARPU, customer lifetime models and so on.
4. Use Multiple Lenses
A multi-lens approach creates the greatest analytical power. For example, retailers often look to analytics for guidance on how to identify private label opportunities, or to rank strongly performing vs weakly performing SKUs. An effective multi-lens analysis for this will combine Price/SKU/Promo/Cannibalization type merchandizing analysis (ie: meta data about SKU margins, sales velocities, cannibalization, etc.) with customer analytics (ie: individual customer data showing SKUs association with high value customers) and basket analysis (ie: till or transaction data for each "basket" that shows SKUs association with high value baskets).