Why prescriptive analytics matters - the view from Profitect customers

Profile picture for user jreed By Jon Reed May 1, 2019
Prescriptive analytics has my attention. Most users don’t have time to assess data – they need exception alerts and next best actions. Here’s what I’ve learned talking to Profitect customers.

Yehiav (left) and Adams at NRF 2019

Now and then, a terrific story falls through the cracks. But sometimes the topic is worth fighting for. That's the case when it comes to prescriptive analytics.

Analyzing the data is one thing. Taking smart action on the data is where the real juice is. The last few years, some of the best conversations I've had on modern analytics have been with Profitect CEO Guy Yehiav. Profitect certainly makes prescriptive analytics sound appealing:

Our award-winning software automatically interprets your data to find valuable opportunities, alerting you in near-real time how to improve profits and margins.

Profitect's CPG and retail customer logo list is deep, indicating they are field-tested. So when I had the chance to swing by the Profitect booth at NRF 2019, and hear hands-on stories from customers, we made it happen. I'll spare you the post-interview delays to publication, but here we are. First up? Chris Adams, VP Category Management Services at Retail Business Services, an Ahold Delhaize company.

How Retail Business Services empowers users with data

Retail Business Services, LLC, is the services company of Ahold Delhaize USA. In that capacity, Retail Business Services provides services to six East Coast grocery brands, including Food Lion, Giant Food, GIANT/MARTIN'S, Hannaford and Stop & Shop, and last but not least Peapod, the country’s largest online grocery retailer. Taken together, the companies of food retailing giant Ahold Delhaize USA comprise one of the largest grocery retail groups in the United States.

Ahold Delhaize USA isn't holding back on retail tech either, as evidenced by the use of Marty the Robot in some of their subsidiary stores. I caught Marty the Robot in action on video at NRF - see The informal NRF demo awards - a countdown of retail tech in action. So what does Retail Business Services do, and how does analytics fit into the picture? It's really about helping employees serve customers better. As Adams told me:

We have a lot of extra store users taking action now, doing the execution and that to me is the really key thing. I don't want to have 5,000 people in the stores staring at a big spreadsheet. I want preferably everybody out on the sales floor - like I wanted to be when I was in the store.

To Adams, prescriptive analytics is about getting away from cumbersome report building:

When I think about big data and what we're doing with prescriptive analytics and AI, it's about how do I take that repetitive activity, crunching the numbers, creating the report, and spend less time on that - and more time on taking action?

During our NRF chat, Adams said that Profitect has been rolled out to 800 stores. Most of their 5,000 Profitect users are taking action based on the patterns Profitect identifies. A smaller subset are under the hood, helping to tweak the system. Adams' team started small, with a pilot, but the adoption of Profitect grew quickly.

Adams told me Retail Business Services provides a range of services to Ahold Delhaize USA, spanning information technology, merchandise and marketing, financial services, and more. Turning data into insights their stakeholders can use is one common thread.

Turning insight into action - analytics lessons for retailers

When it comes to AI and prescriptive analytics, Adams advises retailers to ask and answer a few key questions:

  • Does it save time? Can I go faster?
  • Does it increase sales in a profitable way?
  • Does it help me focus on the end consumer?

He adds:

This space is about using patterns and automation to drive action. It's all about empowering associates to take action based on the data.

But what about the end customer? Adams says he's starting to see that impact as well:

Certainly, our starting point and our focus is very much in the shrink and the loss and the inventory space. But thinking about the sales implications of that, if I have that product that's sitting in the backroom and doesn't make it to the sales floor, I can see that the vendor didn't service it or get it out. Based on similar stores, the patterns are showing me that you would expect the store to be doing something different. There's that double-sided benefit of getting the sale - and getting to less shrink.

Some examples of putting prescriptive analytics to use?

Operations and execution - Adams used the example of a dairy markdown: how do they make sure all stores are actually executing on it? They don't want to retrain all 700 stores on a new process. And they want to act while perishables are still in the store:

It might be that fifty stores are high volume, and are not seeing the kind of sales on those markdown items that you would expect based on their peers. So that's benchmarking and metrics, and being able to know right away while the product is still in the building, and we can take action.

On the operations side, an effective question is: "How can I see exceptions and go after them faster?"

"It's always that blend of art and science"

Adams told me as their use of Profitect matures, they are looking at upstream benefits.

One thing we're looking at is ORC (organized retail crime)... If stores along route 95 are where I'm seeing a loss, I might be able to see that faster than those individual stores would see.

One Profitect breakthrough: faster measurement, with better benchmarks.

One of the biggest challenges we've always had in the execution side is measurement, and especially fast measurement. So I'd say one of the things we got out of Profitect is we got some benchmarking and visibility too.

Though Profitect identifies patterns and recommends next best actions, Adams emphasized it's not about automating everything. The human with retail expertise is still in the middle:

Data-driven is not the same as data making the decision. We want to use the data as a better start for decisions, and continue to leverage the expertise of our people.

Data is not just "science":

It's always that blend of art and science. No tool, no matter how good is going to replace all the experience you get between the ears of those category managers. We give them a way better starting point, and a lot less of wading through the data to get to that place to make a good decision. That's the art and that science mix.

Here's a sample screen shot of recommended actions for a marketer:

We talked about the challenge of cajoling users away from their tools of choice, or getting them to embrace new ones. With 5,000 Profitect users and counting, adoption has gone well for Adams and team:

If you can't show people that it's going to make their life easier and save them time, they're not going to use it.

The wrap - it's not about perfect tools, it's about the right partnerships

Understanding roles is key. Analysts using Profitect might still want to dig around in Excel from time to time. But a store manager relies on the quick actions Profitect surfaces:

As a user or store manager, you might spend three, five minutes a day in the tool. You're going in there to look at the opportunity. In many cases, we're pushing it right to your phone or right to your email. You're going in there to say, "Hey did the action, that prescriptive suggestion that came out of it, did it work?

No tool is perfect. A good vendor partnership is about facing challenges and improving the result:

Are store manager's reacting to the opportunity that we're seeing, and closing that loop of bringing the feedback back to say basically, "Was this worth it? Was this a pattern that actually delivered value to you?" And if it didn't, then that handful of super users back at our office, partnering up with some of the Profitect folks, can tweak and adjust it - or get rid of that pattern and replace it with something that lines against their needs.

That's the right note to end on:

Lining up against customer needs, both internal and external, is the key to success.

This was more of an impromptu discussion than the end user case studies we typically do. Hopefully, it paves the way for more formal use cases in this important area.