Predictive Models
SMI in consultation with key corporate instructors from some of the leading Southern California colleges and universities has developed a new Data Experiment methodology that can increase a company’s ROI.
​
These data experiments are designed in part on current certificate programs in data science and predictive analytics that are used to train senior analysts and managers at companies such as Cisco Systems, IBM and Southern California Edison.
​
Data experiments on average can provide an uplift from 25% to over 100% above standard predictive modeling created without data experimentation.
In fact, the top data driven companies in the U.S. (e.g., Google, Amazon, Capital One) conduct data experiments on a regular basis to augment and improve their predictive models.
​
However, outside of this top tier of thought leaders in data science, too often models are created and deployed without proper data preparation, business understanding, data experimentation and validation methods.
​
In response, SMI has created an affordable Model Fitness Test for small-to-medium size businesses that can be used to grade your predictive model and campaign effectiveness.
Applications of SMI’s Predictive Analytics programs include:
​
1. Predictive Cart Recovery
Recover carts by providing the right incentive for each customer, taking into account likelihood to buy, lifetime value or cluster personas. Coordinate cart campaigns across channels to ensure you don’t remind customers who already bought.
2. Predictive Email Retargeting
Double site conversions by reminding customers to come back. Use one-on-one content such as abandoned search terms, items browsed, and recommendations. Then add the right incentive for each person based on likelihood to buy and lifetime value.
3. Repeat Purchase Program
Get customers to repeat purchases and double sales with post-purchase offers and recommendations or by sending replenishment reminders for consumables. Use likelihood to buy predictions to add the right level of discount or incentive.
​
4. Cluster Based Targeting Program
Triple conversions and increase lifetime value by targeted content using predictive product, brand or behavioral clusters. Also add personalized product recommendations to any of your programs – online and offline.
5. Loyalty Appreciation and Retention Program
Increase retention by showering your best customers with attention. Increase lifetime value by rewarding each customer with personalized loyalty rewards if they buy just a little more, or a little more often, than they usually would.
6. Customer Reactivation Program
Reactivate customers with content based on cluster personas, a personal recommendation and an incentive that is right-sized using likelihood to buy and lifetime value. Also reduce email opt-outs 40% by adjusting email frequency by segment.