Dennis McGrath on How Starbucks Tests New Ideas in the Real World

By Scott Kirsner |  July 11, 2017

McGrath says that since 2009, the Seattle-based coffee chain has been working to accelerate the way it can test ideas in just a handful of stores — in part to see what will break before they are rolled out nationwide. Now, McGrath explains, “we can have a concept on Monday and have it in the store by Thursday. We may leave it in the store for as little as three or four days, or as long as six months. …The stores that we selected [for test stores] we wanted to…be representative of the US portfolio. We wanted to have space-constrained stores. We wanted to have drive-throughs, and all the segmentations of our customer demographic.”One example of a concept that made it through McGrath’s testing protocol: the wildly successful order ahead feature that is part of Starbucks’ mobile app.

Different Innovation Teams at Starbucks

Starbucks still has a very strong entrepreneurial culture within the organization. There’s not a function or a team that is a keeper of innovation; it’s integrated throughout the entire organization. I think there are five teams. For instance, there is a beverage concept innovation team. There is a food innovation team. There’s my [operations innovation] team, and there is an IT innovation team.

Some of those teams…do not have the burden of the day-to-day business. They are charged with innovating against our [strategic] plan. Our strat plan is a five-year plan.

For myself, as well as many of those other teams, we live in 2021 as of today, and we are innovating to deliver against our five-year strat plan.

In terms of where my team sits, we’re a little unique. I actually report into two functions. [I have a] dotted line into Andy Adams, who is our global leader of real estate, facility, store design…as well as to our EVP of global operations. My time is shared equally between those two functions, but our team serves the entire organization.

The quickest way I can describe it is [that] we are the keepers of the innovation protocol. That means the idea or the program or the concept may not have been generated by us. About 20 percent of them are. About 80 percent of them are generated by other teams throughout the building…

Learning How to Test Differently

Prior to 2009, Starbucks really didn’t test anything. We may have used that language from time to time. But, basically when we were testing something, it was launched in 5,000 stores or 8,000 stores.

[That’s] a launch, it’s not a test. [And the] development process to get a program to [that point] was two or three years.

[As a result,] we were slow to market, because we had to get it perfect before we would launch it. Then… its showing up [in] 5,000 [locations.] Then we would go into this perpetual continuous improvement [process] trying to tweak it and fix it.

What started as a very small idea back in 2009 was, what if we could have direct access to a small number of stores, [and] use those stores to help us develop the program — it’s really what we call feasibility testing.

[With] feasibility testing, we’re just trying to answer [a few] simple questions: can we do this? What are the problems to solve? Now that we know what the problems are, are they worth solving? How do we prioritize the work on this? That started in 2009, and today we have a very robust catalog of stores that are available to our team.

We don’t operate the stores. The field operations team does that. These stores are available to us. We can have a concept on Monday and have it in the store by Thursday. We may leave it in the store for as little as three or four days, or as long as six months.

[Today], we have labs, we have the stores, and we are accountable for all testing in North America. A test could be as little as two stores or 600 stores depending on the test objectives and where the innovation is.

How We Choose Stores for Testing

Dennis McGrath

The stores that we selected we wanted to…be representative of the US portfolio. We wanted to have space-constrained stores. We wanted to have drive-throughs, and all the segmentations of our customer demographic. [We wanted the test stores to] be representative of how something would perform in the US.

We did spend time…training the store partners. (At Starbucks we don’t call each other employees or associates. Everybody is called a partner.) [But] the training is not on how to test things. The training is on how to get feedback, how to remain objective. Part of testing in the real live environment is getting fact-based, objective feedback, and sometimes people tell you what they think you want to hear.

We do not hand-pick the store managers or the teams. We leave that up to the field organization; sometimes we get a top-performing store manager on the team sometimes not. But, that’s also representative of how [things] may happen in the US.

In order to keep the stores objective, we never give them goals on anything. Sometimes we’ll not really tell them the true objective of the test. That’s a way to keep them objective; we give them enough information to execute the program. What they’re held accountable for is executing the program.

If that means [that] every day they’re throwing out $100 worth of food or $500, we tell them, “Keep throwing it out,” because that’s how we continue to learn. There’s a lot of that goes into setting that up and planning it.

I could tell you [about] the change management component of the first time I went to a field leader, a regional vice president, and said, “Hey, I want to take these 36 stores, and I want to have direct access, and you’re going to have no veto rights and I’m going to do whatever I want in those stores.” That wasn’t the best conversation I’ve ever had.

Launching Mobile Order & Pay

What we used to do is we would say, “OK, team. We want to launch mobile order and pay in all stores in North America,” and we picked a date.

That date became an immovable object. In that strategy, what happens is you’re making compromises all along the way just to get to that date.

What we did with mobile order and pay was different. We really started off with a question of, “What would we do if we were a startup and we decided we were going to build an app and enable mobile order and pay? What if we were not part of Starbucks?”

What we came up with was we will launch a minimum viable product at a small scale, and build to learn. That’s what we did. We launched mobile order and pay in twelve stores in Beaverton, Oregon.

We selected Beaverton because it was not in Seattle, so we could keep our senior executives out of there so the team could continue to develop it. It was close enough where we can get [from headquarters in Seattle] to the market fairly easily.

When we launched it in those twelve stores, the first couple of weeks we had some [issues]…In hindsight, looking back, they would have been catastrophic disasters had we launched in a larger number of stores.

I’ll give you one simple example that none of us on the project team foreseen, but when we launched it in the twelve stores…Up until then, store naming conventions were done for internal purposes.

For example, in California, we have a store called Point Loma. That’s how everybody at Starbucks refers to it. It’s not in Point Loma. It’s actually across the street from the Point Loma Naval Base.

In Beaverton, there’s a place called Washington Square. We had Washington Square 1 and Washington Square 2, with no reference in the app of which store is located where. We had customers placing orders, sending it to one store, and going to another expecting to pick it up.

We learned quickly in that week [that] we need a naming convention for all stores…so it’s consistent across the US.

[With] GPS [coordinates, some] stores showed up in the middle of a lake. We had stores where we placed orders, and the orders just went in to the abyss. We have no idea. They never received it, and we have no idea of where it went.

We had lots of failures on the IT side. Actually, in those twelve stores, we had programmers sitting in the stores who could make real-time code changes as we were learning things. We did that for 60 days.

What we wanted to achieve [was] really bringing in that human-centered design principle…we wanted that [software] engineer to understand and design it from the perspective of the people who are using the software, from the perspective of the customer and the perspective of the store.

Many times, you can design a great program, but if people don’t know how to use it that way or it’s not intuitive, you have a great program that nobody adopts.

Applying the Right Rigor at the Right Time

Early on, you don’t need a business case — you need a hypothesis, you need an objective. We will activate based on that.

That doesn’t mean we’re going to put something in 15 stores or 1,500 stores to go test it. That may mean we’re going to taking it to our lab and doing some feasibility trials on it.

We keep sales out of scope early on. We keep customer validation out of scope early on. Those are the things that enable you to move much faster.

For instance, in the case of mobile order and pay, going to the twelve stores, we [defined] what the minimum viable experience would be. That would be [that] a customer can open the app, find a store, browse the menu, place an order, pay for it, go the store, and pick it up.

[When the] question [came up,] “Does it have to have every single item of the menu?” The answer to that was no, because we went back to our five minimum viable experience [criteria.] They have to be able to find the item, browse the menu. We did not say, “It has to offer every item.”

…If we were to put every item that we had on the menu, with those twelve stores, we would have delayed launching it in those stores by about four months or five months.

We said, “You know what, 80 percent of the [items] customers buy, having those on it will be good enough for now.”

That’s how we moved fast. We have a saying on the team. It’s, “We want to apply the right rigor at the right time.”

How We Prioritize Projects

We have a decision logic that we use. Our first question is, “Is that in our strat plan?” If it’s within that [plan’s] five-year horizon, now we know the enterprise has decided that this is important. If it’s not in the strat plan, then it goes into Column Two. [If it’s in] Column Two, we deprioritize it.

The second is, “Is it a business imperative or business critical?” That could be, let’s say, [that] there is a new regulation that we need to address that no one foresaw. We need to solve this problem in the next six months. I’ll give you an example of that.

We used to have dipper wells at the espresso bars. One day, a customer took a video of our dipper well and posted it on YouTube. What it showed was this continuous flow of water running. He was a professor. He used it as a class project to get the students to determine how much water Starbucks was using in the dipper well.

We got a lot of bad press about it. We knew we had to solve the problem, but we also had to meet health regulations…

We prioritized that work. In 90 days, not only did we solve the problem, but [then] we actually implemented it within… 18,000 stores in about seven months. The solution for it [was] to have the water only running when it actually needed to run. It sounds simple but extremely complex and even more complex to execute that in 18,000 stores in a short period of time.

‘We Fail and We Move on’

[Founder] Howard [Schultz] is the visionary within the organization. That’s what keeps the entrepreneurial spirit alive in the company. I would say one of the things that keeps that alive is we don’t dwell on our failures.

I’ve been in other organizations [where,] when you failed, the first steps were like, “OK. Let’s figure out whose fault it was, who’s to blame, how we’re going to change that team?” It would create a lot of unnecessary reaction at times.

At Starbucks, when we fail, we fail and we move on. That culture of the acceptance of failure as part of the innovation process is our greatest competitive advantage. We have lots of sayings. One of them is, “Fail often, fail fast, and fail cheap.”

Where Ideas Come From

We still use [the] My Starbucks Idea website. As a matter of fact, if you’ve ever bought a cake pop in a Starbucks, that cake pop idea came from My Starbucks Idea. That wasn’t the exact idea that a customer submitted, but it was to have small bites of cake.

Ideas [also] come from our operations teams. A lot of that is around, “How do we solve operational problems or customer engagement problems?”

Those teams that I mentioned to you before, [like the] Beverage Concept team, those teams are charged with, “What is the beverage concept five years from now, and are we going to get there?” They may be doing the due diligence on all of those things. We partner with them on the innovation protocol.

We’re doing 350 projects a year. We learn a lot from all of those projects. We may run a project, run it all the way through a market test and everything. And realize it was a failure and shouldn’t go forward, but there were elements of that that we learned something new.

When Do You Bring in Stakeholders?

We engage our partners, customers, and suppliers early and often. If we are developing a concept and a piece of equipment is involved, we actually want that supplier to be part of the innovation team.

We’re not asking for a commitment for them to be here all the time. But we want them to understand [that] it’s about that human-centered design principle. We want them to not design a piece of equipment based on what we’re telling them — because sometimes we’re wrong.

We want them to be immersed in it so they understand how we’re going to use it — the experience we want to create, the challenges within the store with this piece of equipment. The more we involve them earlier, the better the program we’ll design.

Testing a New Store Format in Seattle

Today [April 6, 2017,] we opened the world’s first cashless Starbucks [at our headquarters.] This is a store that is built for speed and the digital experience.

It is about 275 square feet. Your typical Starbucks is around 1,900 square feet. It is compact. It is designed for ultra-high-volume capacity. It is digital orders only. We just opened it this morning at 6:00 AM. We took our first order in.

[The] staffing is different. We have introduced what we call a concierge model. You will walk up. [Since] there is no POS [cash register] transaction anymore, there’s a digital screen that tells you whether your order is ready or not. You walk up to a concierge, and they hand it off to you. You can be in and out, we believe, in under a minute.

This, for us, is a concept study. I believe it will have legs in places like airports, universities, hospitals. I could see [these] as satellites in places where, the store we have, we can’t renovate it or make it any bigger than it is — but within that area, we could open a small mobile order and pay store. Think about places like New York City, where midtown rents can be $650 a square foot. This could be a viable alternative.