Data Talk: Q&A With Jon Wilson Partner of Retail Scientifics
As Chief Data Scientist and Partner at Retail Scientifics LLC, Jonathan L. Wilson, PhD thrives on helping his clients like Petco, solve computationally intensive problems which require top-notch quantitative skills across a wide variety of business areas.
This week, he joins CPC Strategy for a Q&A on what Retail Scientifics offers, how it can impact your business and his take on the future of ecommerce.
1) What services does Retail Scientifics offer:
Retail Scientifics is a boutique data-driven consultancy dedicated to developing actionable models broadly throughout the enterprise for national and global retailers. Our most common engagements are for:
- Marketing – lifecycle management including uplift, cross-sell, welcome, and lapsed campaign via email and direct mail, media mix optimization, goal driven segmentation, and more.
- Market Planning – sales forecasting for site selection, acquisition analysis, cannibalization estimation, etc.
- Price Optimization – both for online and offline.
- Voice of the Customer (VOC) analytics – including topic modeling and sentiment analysis.
- We also provide a wide variety of other ad-hoc quantitative services.
We provide businesses the ability to implement the benefits of modern analytics immediately.
2) What are some common challenges you face working with clients?
Data organization – Often we request data, and the business must spend weeks to months trying to track it all down and aggregate it from different systems and locations.
- Data quality – Many businesses have been collecting data for years, but no one has been looking at the data let alone validating its accuracy or using it proactively. We often find substantive errors in the data which require fixing before modeling can take place.
- Socialization – Some of our clients perform basic reporting using Excel spreadsheets and have never been exposed to what predictive modeling tools like R and Python, or big data products like Hadoop and Spark can provide. The introduction and utilization of the tools and output can be daunting. We work very hard to make the process accessible and ensure the findings are actionable.
3) You work with companies that have physical locations, an online presence, and sometimes both.
How do they differ for your kind of work?
In some ways they are quite different and in others, very similar. Let me explain in the context of one of our areas of specialization: price optimization.
When we optimize prices we use a retailer’s goal (typically maximizing net profit) and build a model used to find the price at which that goal is maximized.
When analyzing consumer behavior for store pricing, we deal with a much more static pricing dynamic. Select products may get discounted on a weekly basis, the process of implementing change may be cumbersome, and reading results can be slow. We may have to wait weeks to see the effects of a test and months for the results to aggregate to meaningful values.
Online pricing is a much more dynamic. We can be much more proactive in setting prices that are implemented immediately, address competitor responses quickly and assess results very rapidly.
The push towards omni-channel solutions has made the disparate pacing of online and offline modeling even more challenging as the solution needs to address an enterprise-wide goal even as retail channels might react differently.
4) You mention you have a powerful algorithm which can de-duplicate records giving pinpoint accuracy into the behavior of your customers.
Can you explain how this process works?
Almost all retailers have a customer loyalty program which is used to collect personal and transactional data. Acquisitions, store expansion, customer mobility, and channel addition all add to the complexity of properly tracking customer behavior.
Historically different channels have tended to setup record keeping procedures in different IT systems, which gives little visibility into how customers interrelate across channels and over time.
As omni-channel retailing and personalization become ever more important, linking an individual, or multiple individuals within the same household, across all of these channels by de-duplicating the customer database provides significant value for analytics. Knowing how your customers behave across all channels and within a household unit are critical inputs to proper customer analytics.
The de-duplication algorithm we have created relies on fuzzy matching on customer text field attributes.Name, address, and email address are some examples. All field entries in the algorithm are dynamic and can be catered to the specific business at hand.
Defining when records match or do not match is different across businesses, and we provide a customized solution specifically tailored to that business’s goals.
5) What makes your services unique from other competitors?
Retail Scientifics provides customized modeling solutions for a variety of businesses. Most of our competitors are focused on selling software wrapped around a static algorithm designed to be one-size fits all.
While this approach does an adequate job for analytical deliverables, it rarely provides the exceptional findings and guidance specific to the client.
We take our predictive modeling capabilities to the best possible outcome from the start and are platform agnostic for the implementation.
6) What are 3 bold prediction that you have for online retail in 2015?
- In 2015, online retailers will come to see data science as a requirement, rather than a luxury. Your competitors are gaining too much of an advantage if they are running with advanced analytics capabilities and you are not.
- Additionally, in a reversal of recent historical trends, you will see online retailers begin to set up more brick and mortar locations. The vast majority of retail sales take place offline, and online retailers want a piece of that market.
- Finally, we think mobile share will increase its pace in overtaking standard web commerce. Mobile won’t just be something required, but begin to be used as a tool to accomplish company goals whether that be social, or loyalty based. As companies begin to roll out mobile wallet solutions such as Apple Pay, this trend should gain steam.