Opening Our Models to Your Insights for Greater Results

According to a recent IDC report, worldwide spending on AI will exceed $52B US by 2021. Now that AI-driven personalization is mainstream, and its value is understood by any business leader with a pulse, companies are spending an increasing portion of those billions to ramp up their data science departments and leverage any advantage their customer and business inputs can provide them.

Case in point: Many of our clients have invested heavily in big data and hired expensive data scientists to pilot data science algorithms and models for personalization tailored to their business. However, they are finding out that their personalization vendor is unable to help deploy and test these unique insights and extensions.

The RichRelevance Data Science Workbench™

The Model Behind Every Decision

While the ‘science’ of data science is incredibly complex and involves advanced math that only a few understand, every decision can be boiled down to an intelligible strategy that is the product of a model and algorithm.

For example, in a simple wisdom of crowd (WoC) strategy, like “Top Sellers,” the model can be weighted on the number of units sold. In a more complex example, such as “Complete the Look,” that looks at, say, the likelihood of customers purchasing accessories together that are assembled dynamically, the calculations are more complicated.

Not all Vendors are the Same

Most personalization vendors provide their clients with a black box of limited strategies that have fixed scores that can’t be customized or altered in any way. That means that, even if the business has a hypothesis and a data science model to test it, there’s no easy or defined way for them to deploy it. For an example of why that matters, let’s go back and look at an instance where your business is using a top seller strategy to select and order its recommendations.

Personalization Evolution

Let’s say that after a time of using a top seller strategy, your data scientists believe they’ve uncovered a valuable insight for bundling with a proprietary dataset that can improve conversion rates by 5% or more.

Now, to truly take advantage of your hypothesis, you need to be able to deploy and test the model with live production data without having to stand up an entire AI stack, which can be costly and time consuming.

The typical personalization vendor with its black box algorithm, however, is unable to extend its decisioning to your data science teams.

Data Science Workbench — A New Tool for Data Scientists and Power Users

Fortunately, businesses that work with RichRelevance have an alternative. While we’ve created over 150 strategies that work for almost any context, we recognize every company and their customers are unique.

With the recent launch of the Data Science Workbench (DSW), we’ve made it easier for business to test and deploy models from their in-house data science teams.  DSW is a next generation tool that data scientists and power users can use to quickly create and modify the models that underlie each strategy.

The new DSW dashboard makes it easy to create or change the values in each model. You can also edit the lookback period. For an OOTB strategy, that period is typically 30 days, but what if you wanted to extend it to 90 days or even six months? With most vendors, whose AI is a black box, you’d be unable to even try.

With our open AI, we have hundreds of examples of strategies that can inspire your data science team to both determine what’s important to your business and create new variations that can drive superlative results.

In my next blog, we’ll take a look at some more DSW use cases and how our clients are using it to capitalize on their experience and expertise and further the goals unique to their businesses.

Share :

This post was written by Sean Merry

ABOUT Sean Merry
As Director of Product Management at RichRelevance, Sean oversees and is responsible for all features related to Recommendations as well as other areas of the RichRelevance Platform. He works with clients and members of the field to identify top priorities, and works with RichRelevance engineers to execute on those priorities to deliver robust quality products. Sean brings over 10 years of experience in omnichannel commerce and innovative technologies within CPG and retail. He uses his digital expertise to develop products that use big data and artificial intelligence that to bring first class personalization and extensible, flexible features to clients. During his tenure at RichRelevance, he has helped to roll out personalization programs for iconic brands such as L’Oreal, Barneys New York, Michael Kors, Samsung, CDW, and has successfully launched micro-segment strategies and the Data Science Workbench.
Related Posts