Changing the data conversation at Target - a Domopalooza BI use case

Profile picture for user jreed By Jon Reed March 28, 2017
Summary:
It's not easy to change how a business uses data. Now trying doing that at Target scale. At Domopalooza 2017, Target's Ben Schein and Lisa Roath shared their data adventures from the BI team and business user view. In this installment, I look at how Schein used Domo to build momentum for real time data at Target - and how decision making is changing as a result. Bonus: some Black Friday war room gotchas and fixes.

Ben-Schein-Target
Target's Schein on stage with Domo CEO James

Ben Schein, a BI lead at Target, had a BI problem most have confronted: he wanted a BI approach that combined speed to market, scale, and collaboration. "I was tired of being told I had to choose between them," he told the 3,000+ audience at the Domopalooza day one keynote.

This was the third annual Domopalooza, Domo's user conference (see my Domopalooza early roundup). I chased down the details of Schein's back story, attending Schein's breakout session later that day. This was a chance to learn about Domo at large enterprise scale. Schein was joined on stage by Target's Lisa Roath, Vice President of Merchandising Transformation Insights.

Target's story brings a new way of looking at BI into focus - one which serves users and executives, but also challenges them to set aside the disappointment of past BI tools and give a new tool/approach a shot.

Then there is the ultimate issue: once you have that vaunted "single source of truth" in a visual format on demand/device, can you use it to improve decision making? Or is it one more sexy app on your phone you forget to look at?

Schein, whose actual job title is Director, BI and Analytics within Enterprise Data, Analytics and Business Intelligence (EDABI) CoE, built up Domo adoption from the inside as he won more employees over. Now, 2,500 to 3,000 unique users are on the system each month. (Schein's team supports Target's merchandising unit, which includes a host of retail buyers). As Schein told the audience:

The early attraction was about mobile - a lot of tools didn't work for mobile and were hard to use. It was also about the number of users Domo could support at a given time.

Why is scale so important?

What really scared me is that sometimes, we either wouldn't go after a business problem, or we would be using the non-optimal amount of data... If item-level data answers my business question more accurately, but I'm using department-level data because I don't want to wait six months - or I don't have the right tools - that's actually really dangerous.

Schein selected Domo over tools that would not be able to handle the scale of Target's Black Friday rush. With his speed to market, scale, and collaboration boxes checked, Schein pushed ahead. Three years ago, his team began building Domo "cards" and setting up user dashboards, starting with digital and Adobe data. As they scaled up, the focus shifted to: "How much data can be use, how much can we crunch, how can we do it in a way that will let us really be agile and respond to problems we didn't even know were going to come up?"

Changing the data conversation at Target - how Schein and team did it

Target's BI environment is a mix that includes Domo, Domo APIs, "Big Red," (Target's Hadoop platform), and a data automation tool (all the data is pushed through Target's automation tool from a host of data bases, via APIs into Domo for business user consumption). Schein has data analysts in disparate locations, feeding data into Domo from a range of data sources. Every day/hour, market data is loaded from Target's 1,800+ stores.

That led to an on-stage story from CEO Josh James, who was passing a Target while camping in Colorado. He sent the store location to Schein. Three minutes later, Joyce heard back from Schein: "Looks like their sales aren't doing as well as they were last summer in these items..." Schein laughed on stage, but he also warned us to get caught up in bells and whistles:

Some of that is security blanket stuff, right? Senior executives like it, and it's important. But when we can actually unlock business decisions, that's when it's powerful. Because we have the data at the location level, we have it ready when we're opening a new store.

That has a direct impact when Target looks at opening smaller stores - a key strategy to get Target closer to certain suburbs, urban locations and college campuses:

We have a team that is hyper-focused on those stores. One of those team members is here; she does a ton of work in Domo... When we're opening a new store in Tribeca, she doesn't have to reload that data, she doesn't have to reprocess it. She is on the edges with that speed to market and the scale, figuring out what's relevant to that team. They actually can make changes. They can look at the hourly sales and say, "Do we need to change staffing?" "Which categories are resonating in a way that we didn't expect or didn't see?"

Schein also hit on the viral/jealousy factor amongst executives if you do BI the right way. A number of executives were visiting a new store during the opening. At dinner, several executives were looking at the cards Schein and the business team set up. That sparked a flurry of emails asking why they didn't have access to Domo, or requesting access. Schein sees a deeper trend: the impact of data on culture.

We're excited to see that on the operational and the leadership level, we just changed the conversation of when data comes in. It used to be that once a week, once a month or once a quarter we talked about data... We're shifting how that conversation works.

That's a big change from poring over aging Excel documents and waiting for the month-end reports to circulate by email. Now, Schein's team feeds all kinds of data, including weather data, into their "Big Red" warehouse, and pushes that into Domo for near-real-time views (they typically load hourly sales at the rate of 4 million rows every ten minutes).

Schein said 7 billion rows of data poses no performance problems. "We're working at really high scales," says Schein, before cautioning the audience that he doesn't view scale as an end in itself. You still have to find the business value from there.

Black Friday real-time adjustments - waiting till next year is not an option

During his afternoon presentation, "Black Friday - Driving business decisions with data at Target," Schein talked about the adjustments his team made in real time, and the lessons learned. His talk had a football theme, so he called these adjustments "audibles." Black Friday is obviously very high stakes at Target. For Schein's team, it's all hands on deck from pre-Friday to after Cyber Monday. They set up their "war room" from Target's northern campus in Brooklyn Park, Minnesota.

Early on, they got a note from their CFO, saying she didn't understand the Domo metric she was looking at - not the person you want to be confused in the middle of your business Super Bowl. "That's fluctuating. What's going on?" she asked.

The view, or "card" she asked about had a real-time update of year to year sales. But the data was fluctuating, causing confusion. The reason for the fluctuating data? Schein's team had gone into Domo's command line interface, aka "Beast Mode," and refined the card so that time zones weren't included unless there were enough stores open (kind of like how election polls aren't called until a majority of precincts report). This was causing real-time fluctuations.

Schein's team solved the problem:

We realized, "You know what? We can use Domo to show her what's actually happening." We can help give her the view of what this metric meant.

He showed the breakout audience the revised card:

target-bi-card
Target BI card, revised during Black Friday weekend

Note to readers: this data is all sample data, but the card layout is real. That late night revision did the trick:

Without me having to explain to her or any of our executives, very quickly they could [get the info they needed, and see the time zone differences].

They got this done at 3 a.m. No waiting until next year to get it right:

We can do this quickly, so it doesn't become "Let's do this next year." It's what can we do right now.

That wasn't the only "audible" Schein's team had to call that weekend. Others included:

  • Using Buzz, Domo's embedded collaboration tool, to warn users of upstream data issues and provide updates.
  • Adjust user views by request, with new cards showing the detail those users needed. Communicate those new card availabilities via Buzz.
  • Create more detailed views by category, overlaying weather and industry trends. Add deep dives by department or item as needed. Schein said this enabled "fast follow up meetings" where the conversations could go from data to actions in ten minutes.

Quick wrap - winning over business users isn't easy

Despite all the improvements Schein talked about, winning over business users isn't easy - no matter the tool. Industry predicaments are not magically solved by data either. On diginomica, we've been analyzing Target's industry moves, and while Target has plenty of digital success to talk about, challenges with physical store sales and omni-channel strategy are similar to every retailer not named Amazon. Real-time data doesn't ensure the right decisions are made during this pivotal time - a point Schein hit on.

Still, this is a different kind of enterprise BI story than the complex data transformation/ETL projects of yesteryear, where business users had to put report/feature requests in an IT queue and cross their fingers. I'll continue this Thursday with business user views on BI change, from Target and more.