As part of my ongoing series on Marketing Decision Management solutions I recently caught up with Granata Decision Systems. Granata was founded in 2012 by folks from the University of Toronto who specialize in optimization and AI. GDS’ product is a marketing optimization solution targeted at organizations with a relatively sophisticated customer analytics environment who want to move their marketing effectiveness to the next level. These marketing organizations are often faced with competing objectives and constrained marketing resources that must be allocated as effectively as possible. Without some cross-product, cross-division optimization the danger is that the same customers will be targeted over and over again while decisions about channel and mix are made in an inefficient, judgmental or uncoordinated (rather than analytic) way.
GDS offers marketing portfolio optimization across spend, mix and targeting. The solution manages trade-offs and optimizes across campaigns, improving campaign coordination and execution. The intent is to reduce customer contacts over all, focus those contacts on an optimal interaction and so improve overall results.
The system takes customers/leads, scores and other data, campaigns/offers/channels and various budget, constraints and objectives as input. The campaign portfolio that is output allocates budgets and contacts to specific campaigns and selects channels for campaigns. The campaign portfolio optimizer connects to various data warehouses and analytic environments as well as to existing campaign management tools. This allows a marketing department to reuse its existing assets, marketing processes etc.
The campaign portfolio optimizer also powers a scenario analyzer that allows what-if analysis to see how changing budgets, objectives or constraints impacts overall results. Actual results can be monitored and resources (e.g., contacts, budget) re-allocated in real-time using existing infrastructure as campaigns are executed.
Architecturally the environment beings in customer records (attributes and contact history for instance), predictive analytics, campaign-specific models, campaign parameters and campaign status. Analytic models can be brought in as scores or as models (using PMML) so that they can be executed live as part of a scenario optimization. All of this can be loaded from existing systems through a set of APIs. This data can be encoded and transformed and then run through a cloud-based optimization engine. The optimization engine is highly performant, especially when it comes to handling the explosion of targeting segments and attributes common in multi-campaign environments. GDS claim a unique, proprietary approach to dynamic segmentation that allows true optimization to be accomplished quickly and at very large scale. The optimizer produces a customer targeting policy to be loaded back into the campaign tools. In addition the engine interacts with the what-if analysis tool to allow things like budget, capacity, objectives, etc, to be changed.
The nature of the way this works is that the results are campaign-focused and thus focused on batched marketing campaigns (rather than interactive next best action selection). The optimization can be used to focus a real-time conversation by developing an optimal response for each state change or action on the part of the customer. One of the key features of the optimization engine is the ability to optimize across sequences where an optimal action must be selected when each action has a large numbers of potential next actions.
You can find more information about Granata Decision Systems here.
Comments on this entry are closed.