James Phillips: Convergence 2015

ANNOUNCER:  Ladies and gentlemen, please welcome Corporate Vice President Microsoft Power BI James Phillips.

JAMES PHILLIPS:  Good morning.  I am very excited to be out here.  I’ve been looking forward to this for quite some time. We’ve got an hour set up for you.  It’s amazing, the energy in the room, following a rock band, a dance troupe, and I’ve really been set up.

I’m going to talk to you about our data platform, the Microsoft Data Intelligence Platform.  This really seems to have turned into a data conference, if you listen to what Satya and Kirill and others have communicated, the power of data to empower, to disrupt markets and to transform your business is the topic that I’m going to go a little bit more deeply on.

We’ve touched on it.  You’ve seen Power Bi.  You’ve seen pieces of the platform.  I’ll take it a little more deeply, and I’ve got two great customers that are going to come out, J.J. Foods out of the U.K., and Wash Laundry, to show how they’re using the technology and the platform to transform their businesses, and both actually have disrupted their markets.

If you walk into Kevin Turner’s conference room — Kevin Turner is the COO of Microsoft — and on the wall in his conference room in about 600,000 point font are the words, “When we’re faced with a tough decision, we need only ask ourselves what’s best for the customer and then the decision is simple.” Being customer-focused, gathering enough information to make those decisions really is the power of data.

If you’re going to act on behalf of your customer, if you’re going to make a decision based on what’s best for them, you have to know them.  And historically, short of being with your customer and getting to know them, it’s been very difficult.  Particularly as our markets have broadened, as organizations have gone global, this gap has widened between organizations and their customers.

Over the last 30 or 40 years, software has been sort of the great gap closer, if you will, bridging the gap between businesses and customers.  We started using software initially primarily to run our businesses more effectively.  Satya talked about systems of record, systems of engagement, running your business. Let’s say you’re an auto manufacturer, a system of record may be capturing warranty work that’s been done, automating your inventory management, your manufacturing processes, and so running the business and in some cases capturing information about your customers.  If you understand how the warranty work is being done, you’re starting to build a pattern of information about customer behavior.

With the emergence of the Internet, we were given a conduit that connects us directly with our customer.  And the software systems that have resulted from that, if you look at Web applications, websites, you’ve got the ability to engage in a bidirectional conversation with your customer, and it really does begin to close that divide.  Now you’re not only reacting and storing information when your customer comes to you physically, but your customer can come to you now virtually and it provides more opportunities for engagement and certainly more information as a result of that.

Mobile computing takes it probably 10 steps further.  It was the case with the Web that your customers, when they were at their desks, were able to surf to your site or to your application and interact with you.  With mobile applications, you are in your customer’s pocket.  Think about the opportunities there.

Let’s say you’re a retail grocery outfit.  Historically if you wanted to learn your customer’s habits, desires, needs — and people actually did this — you would have someone that would follow a customer physically through the store, watch the aisles they walk through, the products they’re looking at, and capturing that information.  With an application in your customer’s pocket, you can follow every single customer through your store, understand their shopping patterns, behaviors, learn, combine that information with their actual purchases, and so the power that mobile computing provides, again, bridging the gap.

Perhaps the most powerful and certainly the fastest growing and I think it’s really going to transform businesses over the next decade, is the ability to embed software right in your products and services.

So every single time your customer uses your product or your service there is software there that can both provide better experiences and provide more information.  If you’re a shoe manufacturer you can literally be in the shoes of your customer watching steps, understanding calorie burn, discerning patterns of behavior, phenomenal.

Now, what’s most interesting, I believe, from all of this software is flowing a torrent of data, a torrent, massive amounts of information being is captured that if mined, if harnessed, provides you with the information that you need to truly walk in your shoes, in your customer’s shoes, to understand what they care about, what they don’t care about, to give you information that you can use to build better products and services, to increase your efficiency as an organization.  And so mining this data is what I’m going to talk about for the next 45 minutes to an hour.

So how do you get from data to something that moves the needle for your business, and we’re going to use this sort of value flow that I’ve got up on the screen here to guide our conversation over the next hour, combined with the previous chart that I showed you with the data collecting from software systems.

Here we’re showing a flow from events.  Let’s say you’re an auto manufacturer and the oil pressure reading emits from a sensor in the car.  It’s collected into the cloud.  Some intelligent analytics system converts that raw data into something meaningful that gets delivered to a human being, who may have an insight that then leads to some action.

And so if you look at data in this flow it’s consistent across all types of systems and this is what we’re going to use to frame our conversation.  We’re going to go deep on two of these things, data and intelligence.  Microsoft has the most complete data platform on the planet and the most powerful set of analytics systems used to turn that data into intelligence.  And we’ll go deep on those two things.

But, before we get into the technology it’s important to understand that probably the most critical link in this formula is the connection between intelligence and people.  It’s sort of like the old saying, if a tree falls in the forest and no one is there to hear it, did it really make a sound?  And data is very much like that.  You’ve got a large collection of data, but if it doesn’t land in some way that a human can take advantage of it, then you really don’t have anything of value.  And we’ve seen multiple times here this morning Power BI provides that critical link connecting this very powerful data platform, intelligence platform, with people who are in a position to do something with that intelligence.

So if you look at Power BI you can view it as sort of coming in front of your data, providing a cockpit, if you will, that takes all of that data and delivers it in a form that makes it useful to humans.  Now, Power BI is what we have called a third-wave business intelligence platform, or a business analytics platform, and we characterize the third wave as allowing business people, non-technical professionals, to come sign up for the service in mere seconds, to begin connecting with their data in minutes, and to begin getting insights as a result of that connection in five minutes or less.

This is very, very different than traditional business intelligence where technical professionals analysts had to connect with data, massage it, model it, create reports, all this work that had to be done before any person could actually get value out of the data.  We’ve taken that complexity and that time and we’ve reduced it down to almost nothing from a complexity and a time perspective.

And so I want to show you Power BI and what I really want to land, I mean the most important message here is this is simple, it’s easy to use, it’s easy to get started connecting with your data and getting value from it.  So what you’re looking at here is a Power BI dashboard.  You’ve seen a number of these this morning.

There are a couple of concepts I want to highlight in Power BI.  So if you look here, there are dashboards and there are reports.  I use a medical analogy to talk about these two artifacts or these types of information you can get out of Power BI.  Think of a dashboard like a patient telemetry monitor in a hospital setting.  When you walk into a patient’s room, there’s a patient telemetry monitor hanging next to the bed.  It gives you heart rate, oxygen levels, blood flows, the health of the patient.  And dashboards in Power BI are exactly like that.  It gives you a one-shot overview of the health of your business or whatever it is that you’re trying to track.

Going back to the hospital analogy, reports are a more diagnostic tool.  So again, if you are in the hospital setting, you see a patient, it’s got a heart rate issue, blood pressure, some test comes back bad, roll them into ultrasound, MRI, perhaps an X-ray, some sort of diagnostic tool.  And that’s what reports are like in Power BI, and there’s a connection between these two.  Tiles on this dashboard are always backed by a deeper exploration experience.  And so if you see something that looks off, you can drill in and begin asking questions.

So here we’re looking at a Dynamics CRM dashboard.  What I want to show you is how easy it is to get started with your own data.  So you saw me click “get data” in the upper left-hand side there.  Power BI comes with a wide variety of connections to SaaS services and other applications that allow you to immediately connect and start getting value out of your data.  And I want to demonstrate how one does that using Microsoft Dynamics Marketing.

So I’ve clicked on this, you’ve got an overview.  I’m going to connect.  Now, I need to enter the URL for the Dynamics Marketing instance.  So this is right on my Dynamics Marketing configuration pane.  And so it’s connecting to Dynamics Marketing and it’s bringing to me an applet that allows me to connect to the system.  And it’s going to make a connection back to my Dynamics Marketing instance.  Here it’s logging me in, because I’m already on the Microsoft platform.

It is now making the connection.  So what’s happening right now is we are connecting to Dynamics Marketing.  We’re pulling the data in.  It’s being modeled in a set of out-of-the-box dashboards and reports are being created.  You can see that happening right here.  So there’s a new dashboard that appeared.  You can see it is loading.  This process takes roughly on the order of about 10 seconds to a minute depending on the system, and I’m going to Martha Stewart here because I’ve already done this, and so I’ll bring this up.

And this is what I will see.  So if we had given it a few seconds, this would have shown up.  And so I’ve got a picture now of all of my marketing campaigns, the contacts that exist in my marketing database, the email campaigns that I’ve been running.  And it was just that easy, no connecting to data, massaging, modeling, creating reports.  It’s out of the box.  And that’s the power of Power BI.

And, of course, maybe not of course, but I’ll tell you, it’s fully customizable.  So as a business user, I’ve got the ability to move these around.  If I want to customize this dashboard to something that’s optimized for the way I want to look at the world, I can move these charts.  I can delete them.  And I can drill in, as I said before.

So here I’m going to click into one of these deeper diagnostics reports.  And you can see that I’m given a view of that same data, but now it’s interactive in form.  When I click, you can see a response from the other charts as I start to ask questions.  If I want to drill down more deeply into a campaign or a particular customer, or a certain kind of potential target customer, I can do that.  You can see it’s a multi-page report, and so there’s a lot of information backing this dashboard that allows me to begin asking questions and trying to more deeply understand my business.

One of the cool things about Power BI is that, again, as a business person I can begin to build up that dashboard on my own.  Here you can see a pin that comes up when I click on a tile on a report.  And if I go back to my dashboard, you can see now that that has been added.  And so I can get into this iterative process where I look at a dashboard, I drill in, I ask a question, I decide I want to operationalize that view, pin that back to the dashboard, and no technology knowledge required here.  It’s something that anyone can do, very, very powerful.

In addition, we’ve got the ability to ask natural language queries.  So number of contacts, for example.  And you can see that it does a great job of spell checking and correcting or not, by country.  So you’ve got the ability.  And, again, everything that I’m doing here is pinnable.  So as I go back to the dashboard, you can see that I’ve got this new view that I’ve pinned, and I can move it up.

And so very, very easy system to use that allows me to make connections to my data, begin interacting with it, both from a monitoring perspective as well as a deeper analytical question-and-answer perspective.

Let’s go back to my slides.  Now I want to move on now.  So we’ve talked about the power of Power BI and the simplicity.  Now I want to start digging into the platform itself.  We’re going to focus, as I said before, on data and intelligence, that left-hand side, not useful if you can’t get it to people.  We’ve made that part simple.  Now I want to start highlighting the completeness of the data platform and the power of our intelligence system that sits on top of that data.

So let’s move Power BI out of the way and let’s go back to the data, and let’s start drilling in.  Now if you look at the software systems that exist above the data, these systems that are producing data, there are different requirements for each of these types of software systems that requires a different kind of data processing technology.  If you go back to traditional systems of record, very transactional.  I just had an order.  I made a sale.  I checked something into inventory.  I started a manufacturing process.  These transactional systems require a relational database management system that’s really good at hard-core transactionality and the ability to ensure that those transactions are executed quickly and reliably.

As we moved into the Web, there was a different kind of pressure being placed on these databases.  And if you think about a Web application, you may have a relatively small set of data, maybe your product catalogue or other information that you want to deliver to your customers, but if you’ve got millions of customers coming to your website or your Web application, that’s putting a different kind of pressure on the database.  It’s no longer simply transactional from time to time, but lots of pressure to get at that data, and it requires what I’ve called a big audience database system, optimized for lots of users coming and accessing data.

You may or may not have heard of NoSQL database technology and it really was invented and optimized for these big audience systems, allowing lots of users to have a great experience against what may or may not be a large data set.

In addition, if you look at mobile applications or as I’m about to describe the IoT, sensor applications, software embedded in products, they tend to generate a tremendous amount of data, maybe sampling every second, 10 times a second, 1,000 times a second, emitting massive amounts of information that, again, requires a very different kind of data processing technology, technology that’s optimized for ingesting massive amounts of information and allowing analytics to occur on top of that.

As you move into the IoT, the information flow can become so great that no system can hold all of that information or it becomes uneconomical, and you want to process that information as it flies in.  So streaming analytics and the ability to aggregate and process that data in flight.  All of these various types of data processing technology are required if you want to have complete coverage across all of these various types of software and business systems.

Now, in addition to the type of data technology, often data will be located in different places and you need to be able to support that.  The cloud and I put this on the right-hand side intentionally.  If you think about the cloud and you think about its applicability to the data problem, it really does start to light up, if you will, or become incredibly valuable on the right-hand side of the spectrum.  Let’s think about this for a second.  Let’s say that you are a shoe manufacturer and you do want to walk in the shoes of your customers and you’ve embedded sensors into sneakers and you sell these sneakers all over the planet.

The ability to ingest all of that information, to do that in real time, to store it where these sneakers may be in Bangladesh, India, Russia, the United States, South America, requires you to have a scope and scale that is almost impossible for any one organization to build on their own.  And Microsoft has taken a huge bet in building out this infrastructure for you to make it simple.

We’ve invested billions of dollars of capital over the last several years and we’ll invest billions of dollars in capital over the next several years to lay down a global footprint of datacenter capacity that is designed to allow you to never drop a byte of data on the floor.  We’ll capture it for you.  We’ll hold it close to your customers.  We secure it, ensure privacy, comply with government regulatory requirements, and we wake up every single day with one mission and that is to drive the cost, the marginal cost of storing that next byte of data, as close to zero as we possibly can.

As a partner we want you to view us as an infinitely large place to capture all this data in regulatory-compliant manner, to hold it, to secure it, to keep it private, to make it available for these intelligent systems that I’ll talk about in our next session.

Now, in addition to the cloud, Microsoft has a very long, multi-decade history on-premise.  Sometimes it makes sense to hold your data in your datacenter behind your firewall, particularly for transactional systems where the scale of the systems and the scope of the audience using those systems is better served by on-premise technology.  And, of course, with Microsoft database technologies that we’ve been in market with for, again, a couple of decades, we support that, as well.

So having both the cloud and on-premise solutions gives you ultimately complete coverage and Microsoft is the only vendor on this planet that covers this full gamut.  We have transactional, big audience, big data, streaming analytics.  We cover you in the cloud.  And we cover you on-premise.  Now that’s important.  If I am a car manufacturer and I’m collecting events out of my car, if it’s telling me, for example, that this particular customer drives 322 average miles per week, that’s interesting, but it really doesn’t provide me with anything actionable.

If I further collect that this car driver has 40 driving habits, and if I combine that with my warrantee or maintenance history that’s stored in my transactional systems, and if I reference that against my Web application, my car configurator, that a user, the driver of that car may be using, I start to get a more complete picture and having that complete picture, which requires complete coverage on the data platform side, is really what delivers the power.  Here we can infer from all of this information as operated against by our analytical system that this is a customer that has an expected chance of buying a new car at a certain level and I predict that they will spend a certain amount of money on maintenance, which allows me to craft an offer, or to proactively reach out and spend money, frankly marketing and selling, based on good knowledge, data that this will be a customer and that that investment will pay off.

So I’m going to bring out our first customer with Wash Laundry.  This is an incredible story.  You heard from John earlier this morning already, we’re going to go a little bit more deeply into what they’re doing with the Microsoft data platform.  But, they have made laundry cool.

John, thanks for coming out.

JOHN BUCCOLA:  It’s great to see you again.

JAMES PHILLIPS:  Yes, absolutely.  So, John, tell us a bit about your business.  What do you do at Wash?

JOHN BUCCOLA:  So we’re an outsource provider of laundry services.  We’re the No. 1 provider in Canada, No. 2 provider in the United States.  And essentially our product is our people and our customer service.  So what we do is in a multi-family housing setting, where there is a laundry, shared laundry type of a setting, our customers use that equipment and we, on behalf of the owners of that building, will collect, service and install that equipment for them.  So we maintain it.  We’re essentially a facilities management and service company.  We are a billion-dollar company, valued at over $1 billion today, and we’re on our way to becoming a $3 billion company over the next five years, all through the application of Microsoft’s technology.

JAMES PHILLIPS:  So if I live in an apartment, or if I’m living in a dorm and I go down to the laundry room, that’s your stuff.

JOHN BUCCOLA:  That’s our stuff.  So most of the universities in California, as an example, Stanford, UCLA, we’ve been doing laundry services at UCLA for over 50 years.


JOHN BUCCOLA:  So our average customer tenure is 23 years.

JAMES PHILLIPS:  That’s a long time.


JAMES PHILLIPS:  So a lot has changed, by the way, over 23 years.

JOHN BUCCOLA:  It has, absolutely.  Over the past few years we embarked on a journey, as I shared earlier, to really going from being a laundry services provider to being a technology and services provider.  We think of ourselves as a broker of technology and service.

JAMES PHILLIPS:  And how have you used the Microsoft data platform to make that happen?

JOHN BUCCOLA:  So our Microsoft data platform, it really starts with Dynamics for us, Dynamics CRM, Dynamics GP.  We have a very rich implementation of WennSoft, as well.  We have a great partner channel in TMC and others who have implemented our technology.  And on top of that we have a very rich data warehouse.  So our data warehouse, which was implemented a couple of years ago, really started coming to life over the past couple of years with the emergence of Power BI and the other suites around the Power BI platform.

JAMES PHILLIPS:  Got it.  And so we’re talking in this section about data, more data, the ability to bring all your data together, and I guess that’s what the data warehouse does for you.  What kind of data do you keep in there?

JOHN BUCCOLA:  So our enterprise data within our four walls, if you will, our information about equipment that’s installed, the service transactions that we’ve performed, and the revenue cycle transactions that we’ve performed, as well.  And then the external data that — Power BI, we’re going to show in just a minute — really kind of tells a more rich story about our business.  And what the data has done for us, James, is it’s allowed us to grow our business more profitably than we otherwise could have without these insights.  So we have really been now a pioneer in how we think about pricing and how we measure usage to get the optimal pricing for each of our customers in each of our 70,000 locations.

JAMES PHILLIPS:  Let’s go see what a high-tech laundry business looks like.

JOHN BUCCOLA:  Let’s go check it out.  So a year ago we licensed Power BI very quickly and the first kind of tool we pulled out of the box, the first toy we pulled out, was Power Map.  So we have over 70,000 locations and before Power Map our ability to kind of take our data and put it on a map was pretty limited.  We were able to do this within a few minutes, with no programming at all.  And so it was really exciting.

So what we’re looking at here is actually our service data.  And we just, again, drag and drop this onto the map here.  And this is actually showing us the repairs per machine by manufacturer.  So you’d expect that similar manufacturers would have a similar repair rate, or reliability rate.  This allows us to normalize that and put it right on the map.

JAMES PHILLIPS:  So what were those columns that were poking up?

JOHN BUCCOLA:  So those columns that were poking up are actually our locations.  So those locations let me just kind of go back to that, each of those 70,000 columns is a location and the height of the columns indicates the repairs per machine.  So if it’s higher it’s no bueno.  We don’t like that.

JAMES PHILLIPS:  Something is going wrong up in the Northeast there.

JOHN BUCCOLA:  That’s right.  And it should be normalized across manufacturers, so there could be a house problem.  We have house problems.  We have user error, I hate to say it.  But, then we have traditional machine problems, as well.  And so we may have to do some user education problems and the Power BI story we’re going to show in a minute will help us drill beneath this data to understand what those problems might be.


JOHN BUCCOLA:  So that’s service data.  We also think about revenue cycle.  So what we’re showing here is actually the revenue per machine.  And again, this is very easy for us to just take drag and drop that information right onto the map and then show very quickly the productivity.  These heat maps are showing money.  This is showing revenue per machine that we’re gleaning out of these machines.

Again, our customers, our residents in these apartment communities, they’re putting either quarters in a machine or they’re swiping a credit card or using some electronic payment system in every one of these 500,000 machines.

JAMES PHILLIPS:  And do you have to choose one is coin, one is card?

JOHN BUCCOLA:  Typically there’s one or the other, but we also have some that actually have both.  So you can see these pink bars are actually going to indicate coin and card.


JOHN BUCCOLA:  So about 60 percent of our revenue comes in from traditional means, so it’s coins or dollar bills.  That 40 percent is coming via electronics.  We’re payment system agnostic. We’ll put in whatever makes sense.  But what I shared in the earlier session, which I want to kind of showcase today, is we really have gleaned some unexpected insights about our business as it relates to payment systems.

Traditional thinking is, hey, just move everything electronic.  You’ve got to be able to swipe a card.  OK, well, let me show you what that looks like.  So what we’re showing here is this is coin versus card revenues by location.  Now the height of the bar is a good thing.  This is the amount of revenue we’re getting from that location.  And the color of the bar, again, indicates the payment system itself.

Now we’re going to overlay that gross per machine, so this is how productive these locations are.  And so when I overlay that on top of this map, I’m starting to get a real picture here for at a location level how productive our machines are and how productive these locations are.

This is Southern California.  We’ve got 50,000-plus machines in Southern California.  We have 500,000 in North America.  But now we’re looking at Long Beach.

Let me tell you just a very brief story about Long Beach.  It’s the fifth largest city in California, has over 500,000 people, and we’ve got 8,000 machines in Long Beach.  And most of them, you can tell from this light blue color, are coin operated.  I also have some card-operated machines that are here as well.

But what you’ll notice is that the productivity of the coin machines is actually higher than the card machines.  And the reason for that is essentially it’s a demographic preference.  The residents in these communities just simply prefer cash.  That’s how they want to interact with our service, so we let them do that.

We can collect either way.  We can install either payment system.  But in this particular example, that’s what they prefer.

JAMES PHILLIPS:  And that impacts your revenue.

JOHN BUCCOLA:  That impacts our revenue.  So in some cases, when we install the wrong payment system, we’ll see revenues actually go down.  Obviously, not a good thing.

So that’s what Power Map has done for us.  It’s a great tool.  It’s visual.  Just by its nature, it runs off of Excel.  So when we think about distributing those insights, we think about Power BI.

This is what’s really exciting for us is now taking the data that we just saw, along with all of our other enterprise data in our Dynamics ERP platform along with CRM, along with our other bespoke systems, our customer systems, teaming that data along with other data that’s external and bringing that all together.

JAMES PHILLIPS:  The heart-rate monitor of your business.

JOHN BUCCOLA:  It is.  So this is a heart-rate monitor for our business.  So let’s take service as an example.  So here’s our dashboard.  We can now drive into service.  Remember those tall bars in the Northeast?


JOHN BUCCOLA:  Let’s go check out to see what the heck is going on there.

So as I click on this, it’s going to take me to a page that’s going to get deeper on the insights of the machines themselves.  So now we’re going to look at service calls by resolution, and we’re going to look at the reasons for service calls normalized by manufacturer.

JAMES PHILLIPS:  And what did you click there?

JOHN BUCCOLA:  So what I just clicked here was — we masked this because we want to protect our manufacturer partners.

JAMES PHILLIPS:  Fair enough.

JOHN BUCCOLA:  But I just clicked on one of our manufacturers, our predominant manufacturer.  And it’s telling us about the problems that we’ve seen for these service calls.  And I can compare that to potentially another manufacturer from whom I’m buying.  And in some cases, I may want to go back to that manufacturer and work with them, which we do.  Or I may think about a certain geography and think through, OK, I’ve got to work on a user education for this new type of equipment that we just installed.


JOHN BUCCOLA:  So this is going to give me all that.

This is the other cool thing about Power BI.  I can make any of these metrics that are shown here, I can surface those on my main dashboard.

JAMES PHILLIPS:  So you pin them back just like I was showing.

JOHN BUCCOLA:  I pin them back.  And so we’ve taken the data and we’ve essentially said, you know what, IT is getting out of the business of cosmetics and how data is presented.  IT is in the business of being a steward of that data, but a broker of services, not a builder anymore, but a broker of these services.

JAMES PHILLIPS:  That was hard to do in the past.

JOHN BUCCOLA:  It was very hard to do.  And Power BI is a platform that just allows us to take this and essentially give it to the business and let the business decide what they want to see.  It’s no longer sort of the stone tablets of reporting.


JOHN BUCCOLA:  It’s all live data, it’s all surfaced right here.  And now I took that one metric we saw earlier and I’ve pinned that to my dashboard.  I can take that and can be alerted for any kind of an exception that’s flagged, or if it trips over a certain threshold.


JOHN BUCCOLA:  So that’s service.  I’m going to quickly go through these last two.

We can grow our business in one of two ways.  We’re a billion-dollar business today, we’re going to quickly become a $3-billion business by penetrating new markets, by opening new verticals, and also by managing and optimizing pricing.

So as I think about new offices, we have a concept we call the micro branch.  Micro branch is basically a small implementation of our service, but has all of our technology.

So I look at Miami.  So we’ve determined through the use of data and the studies that we’ve done, through econometric modeling, that gas prices and certain weekly wages, unemployment rates, these are external factors that create an optimal environment for our service.

We know that when gas prices fall to a certain level and unemployment falls to a certain level and weekly wages rise to a certain level that our product, our service is going to go up, the usage of that service.

And so what this is showing me is I can say, “Which markets do I want to look at potentially entering?”  With this micro branch concept, we decided that Miami was the right market for us.  So we just opened the doors in Miami about three and a half weeks ago.

JAMES PHILLIPS:  Congratulations.

JOHN BUCCOLA:  Really excited about that.  And we’ve got a couple more on the way.  But that kind of insight that I’m able to determine which market to enter into, I can surface.

JAMES PHILLIPS:  And once you’ve entered that market, can you use the platform to understand if it was a good decision?

JOHN BUCCOLA:  That’s exactly right.  So we went into the Phoenix market, just completely green field.  Planted our flag, set up shop there, and we’ve been very pleased with our performance there.  We can show all that performance and how it’s been tracking versus our plan all through these systems.


JOHN BUCCOLA:  So last point I want to show on Power BI is pricing.  Pricing is huge for us.  We have 70,000 locations.  What should the price of laundry be at each of our locations?  I don’t know.  And a human really can’t know.  That’s what we discovered over many, many years of trying to sort of solve this riddle.

JAMES PHILLIPS:  Throwing darts?

JOHN BUCCOLA:  It’s throwing darts.  And so we spent a long time trying to figure out, what is the best way for us to determine what the price of laundry should be for that location?  We don’t want to overprice.  If we overvend, then our customers are going to go elsewhere.  And then if we undervend, we’re missing out on an opportunity for our customers because we split revenue with our customers.

JAMES PHILLIPS:  Interesting.

JOHN BUCCOLA:  So we want to get that just right.  And so what we found was these factors, along with a few others, by putting this together with our data and our enterprise data, we’re able to get a better flow-through rate of usage by optimizing these metrics.  So I can take this information, Power BI does this for me.  In the old days, crawl through spreadsheets.  Crawl through government spreadsheets, put all this census data in, gas prices, all this stuff, and it needs to get updated all the time.

With Power BI, all this information gets funneled right here.  And it’s in terms I understand.  So now I can see the gas prices and the unemployment rate in Hayward, as an example, that’s our East Bay market just outside of San Francisco.  I can see exactly what weekly wages are doing there, I can see what’s happening with unemployment, I can see what’s happening with gas prices.  These are favorable trends for us and for our business.

So Power BI has been very powerful for us to take all this information, put it in one place.  Not only our internal data, but to surface insights about our business even through the lens, potentially, of macro-economic data.

JAMES PHILLIPS:  Beautiful.  Thank you, sir.

JOHN BUCCOLA:  Thanks so much.

JAMES PHILLIPS:  Appreciate it.

JOHN BUCCOLA:  Thank you very much.  (Applause.)

JAMES PHILLIPS:  So John showed you that by taking various kinds of data, he had service data, he had revenue data, he had macro-economic data, by pulling all that together, using Power BI, he had the opportunity to learn things that move the needle for his business.  They’ve seen a 3X increase in their revenue as a result of the ability to price based on this data.

And we’ve seen similar results across a very large base of our customers, and it’s the power of both the totality of the data that you can bring to bear and the simplicity of Power BI that unlocks it.

So let’s move on to the next stage of our equation here, our value flow, and talk about intelligence.

Now, data, as I said, comes together, but then you need to look at that data through various analytical approaches in order to pull the value out.

I was trying to think about how can one visualize that?  And I landed on this picture.  So this is the Crab Nebula, the M1.  And this is a visible spectrum view of the Crab Nebula.  So you can think of the light, the electromagnetic photons that are flowing into the telescope as the data.

Now, the exact same data, analyzed differently, a different spectrum of the data, processed differently, that’s the same exact view of this nebula, but in the X-ray spectrum and processed differently.  And it allowed scientists to understand that there is a neutron star orbiting at a very high rate in the center of this nebula.

And as you apply all of these various forms of analytics, you get different information.  Same data, different outcomes, different insights.

Similarly, with our data platform, if we bring Power BI back in, there’s a hidden layer here that I haven’t talked about before, and that is the intelligence platform that sits between the data and Power BI.  You can view Power BI as an insight delivery or an intelligence delivery vehicle, but that intelligence layer is what really provides the value.  Taking that data and converting it into intelligence.

Stream analytics, I talked about that before.  Machine learning, the ability to take that flow of data and to learn automatically and then be able to process and apply rules in the future based on that.  Statistical analysis.  Building custom intelligent applications.  All of these forms of analytics are provided by the Microsoft Data Intelligence Platform.

And when you combine full coverage of data with this rich set of instruments that can process that data, you have an unbeatable platform to go impact the business.

JJ Food Service is the largest independent food delivery service in the United Kingdom.  They’ve been around for many, many years.  They just celebrated their 25th anniversary, actually.  And I’m going to have Mushtaq Ahmed come out and tell us a bit about their business and demonstrate for us how they use the Microsoft Intelligence Platform — there you are.

MUSHTAQ AHMED:  Good morning.

JAMES PHILLIPS:  Welcome.  Thank you for coming out.  (Applause.)

MUSHTAQ AHMED:  My pleasure.

JAMES PHILLIPS:  So, Mushtaq, tell us about JJ.

MUSHTAQ AHMED:  Well, as you mentioned, we’re one of the largest independent food service providers within the U.K.  We deliver thousands of products to thousands of customers all over the U.K. from eight different branches.

And we have been around for 26 years now.  And we have been on Dynamics AX platform for the last 10 years.

JAMES PHILLIPS:  Got it.  So your customers are restaurants, caterers?

MUSHTAQ AHMED:  It’s all sorts of customer.  We have school canteen, office canteen, any kinds of caterers, management companies, chains, restaurants.

JAMES PHILLIPS:  Got it.  And they order from you ingredients and then they make the food?

MUSHTAQ AHMED:  Absolutely.  And we are an omni-channel type of service provider, so customers can order online, by the call center, email, fax, any other methodology as well.  So we take orders in every single channel.

JAMES PHILLIPS:  Got it.  And so how is data, how has the Microsoft Data Platform helped you transform?

MUSHTAQ AHMED:  Well, as I mentioned, from our sales history, from our inventory management history.  What we recently noticed is that we were almost 100 percent a call center company, but over the last five years, 60 percent of that business has gone into the e-commerce portal, which was pretty much a success story.

But the down side of that was we’re losing the human touch.


MUSHTAQ AHMED:  So in the call center, agents were talking to the customer, they could upsell, cross-sell, introduce new products verbally, explain things.  But as the customers moved to the online portal, we just lost that ability to convince the customer about the introduction of products and so on.

So what we found out, we’re sitting on a gold mine.  Ten years’ worth of transactional data.  So rather than going outside, we looked inward.  Like, OK, let’s do something about this.

So then we started looking at various tools and platforms.  We realized that Microsoft has got something called Azure ML.

JAMES PHILLIPS:  That’s machine learning, Azure Machine Learning.

MUSHTAQ AHMED:  Absolutely, yeah, machine learning.  And we took advantage of that in the sense that we didn’t have any data scientists in house.  We didn’t have the expertise.

So what we did was very simple.  Pull out three years’ worth of sales transaction data, get some clickstream data from the website, and put it into the machine learning model.

And we created a recommendation service.  And one of the services we created was predict what the customer is going to buy today and that is mimicking the scenario where the customers do not have to look for the products when they come online onto our portal.

So the concept of the traditional way of selling something on the e-commerce portal is keep the customers glued to your website as long as possible.  We did completely the opposite.  Give the customer as little as possible, but sell as much as possible.

JAMES PHILLIPS:  We had breakfast this morning, you were telling me that we went from call center employees high touch, but they couldn’t always remember — they didn’t know.

MUSHTAQ AHMED:  Absolutely.  Thousands of products.  We’re introducing new products every month.  How can you remember all of those?  So the framing burden was humongous.

So now using this toolkit, the call center agents do not have to remember anything.  They can use the same features.  What we deliver online, we deliver back into the call center AX portal as well.

JAMES PHILLIPS:  Got it.  Let’s take a look at it.

MUSHTAQ AHMED:  Sure.  Let’s have a look at what we have got.

JAMES PHILLIPS:  So what do we see here?

MUSHTAQ AHMED:  So this is sort of like — I have here a Delve dashboard.


MUSHTAQ AHMED:  This morning, right?  OK?


MUSHTAQ AHMED:  Delve for our customers, right?


MUSHTAQ AHMED:  So what matters for the customers if you look at it.  What is relevant to the customer at that point of time when they logged in?

JAMES PHILLIPS:  How do you do that?

MUSHTAQ AHMED:  As I said, we look at customers’ buying patterns, the frequency, the quantity used per month, per week or per day.  If you think about it, some of the long-shelf-life products which also come in bigger packs last longer on the customers’ shelves.  But because they can actually buy on a promotional offer, they’re not buying as frequently as they would buy fresh produce like mushrooms and tomatoes or milk or cheese.


MUSHTAQ AHMED:  So we have got like a mixed variety of frozen, chilled and ambient items.


MUSHTAQ AHMED:  So it’s very difficult to remember all of that.

JAMES PHILLIPS:  Oh, you learn not only what they buy, but you learn how often they buy.

MUSHTAQ AHMED:  How often they buy.

JAMES PHILLIPS:  And only represent to them what you think they’re going to buy today?

MUSHTAQ AHMED:  Yes.  Trying to mimic the shopping list which they have created before they came online.


MUSHTAQ AHMED:  That’s what it is, really.


MUSHTAQ AHMED:  So let’s look at some of the features we are processing off of the machine learning toolkit here.

So this is just an ordinary website, but what is external is a sequence of products actually displayed, and also the relevance of the product.

So if I start now adding products, let’s say I want some milk.  I’m just going to try to select a channel.  We offer also multi-channel, so we deliver and customers can come and collect.  So I want to collect from one of the JJ branches.  And I’m selecting a time clock for the customers to figure out.

So let’s say we want some milk.  And as you see, I take milk, but this is actually a runtime call to Azure ML tools which suggests that usually skim and whole milk are bought together.

JAMES PHILLIPS:  So I don’t have to go searching for it, it kind of knows I’m probably going to want it.

MUSHTAQ AHMED:  We don’t want the customers to search.  We want the product to come what is relevant.

Now, being in Atlanta.  Now, let’s see if I can do something with Coke.  There it goes.


MUSHTAQ AHMED:  So Coke and Diet Coke go together hand in hand.  Let’s click on some of the other products.  Again, these things are usually bought together.

JAMES PHILLIPS:  And this is all happening in real time?

MUSHTAQ AHMED:  That’s real time, basically a runtime call to the Azure ML toolkit.  Right?

JAMES PHILLIPS:  That’s powerful.

MUSHTAQ AHMED:  So that’s a very fast feature we are delivering on Azure ML.  We call it frequently bought together or usually bought together.

Now, let’s look into some of the other products, right?  We can also look at the product ingredients.  So let’s take this milk again.  And I want to look at the product page.  I’m looking at milk, but some of the other products, and this has nothing to do with dairy.  So if you look at the eggs up there, mango juice, some cleaning materials, sugar.  Right?

JAMES PHILLIPS:  And how did you choose that?

MUSHTAQ AHMED:  This is, again, nothing to do with the individual customer.  It’s actually all about the products, all about the market sector.  All of the industry.  So the machine learning toolkit looks at the relevance for these products.

JAMES PHILLIPS:  You were telling me that based on the kinds of products that I’m buying, you can infer what kind of restaurant I am?

MUSHTAQ AHMED:  Absolutely.  Now, that’s going to be the next feature I’m going to show you now.  So let’s assume that my shopping is done and I’ve got four products in my cart on the top of the screen.  So I’m going to check out now.

And as I’m checking out, we’re making a call to the Azure ML, and by looking at the products in the basket, it is now recommending at the last minute — it’s really the last minute if we can push another item into the basket, you are infusing yourself.

Now, the thing about this, every 100 items we present in this way as a recommendation, five of them actually end up in the basket.  So that’s a sale, right?  Customer probably didn’t have it in a shopping list, he had his list, he did his shopping, and he’s going to check out.  We’re putting another five extra items.


MUSHTAQ AHMED:  So this is the power of machine learning.  We would never do that with human beings, absolutely never.

JAMES PHILLIPS:  So let’s talk about the sensors that you’ve got in your trucks.


JAMES PHILLIPS:  So you’ve got customers that come and buy from you, and then you’ve got to deliver these products.

MUSHTAQ AHMED:  We’ll have to, right?  So there are thousands of customers, thousands of products.  Half of it is collection, half of it is delivery.  We have to deliver to the customer’s door.

And what we do, we do total paperless delivery.  So our drivers have got a Windows app running on a Lumia phone, right?  And this is the beauty.  Again, we used to be traditionally looking at very ruggedized, expensive devices.  More like a $1,000 investment per driver, right?  I think it’s coming to $100, just a consumer device.  But with a $5 case around it, it becomes very, very rugged.  Very, very rugged, and quite tough.

Now, the goods I mentioned early on, we do chilled, frozen, ambient.  Now, in order to provide the best service in the industry, we need to make sure that the frozen goods arrive really frozen, chilled goods maintain their temperature, because food safety authority is very strict in the U.K.

And having said that, lots of these companies are doing the same job, you know, buying and selling.  But what is keeping us ahead of the competition are all of these ancillary services.  That way, we make our customers’ journey seamless on the portal.  The way we monitor the temperature, the goods in motion.  Let’s call it goods in motion, right?

So we are putting some of these temperature sensors — let me take one of these out so that I can actually show the audience what we’re doing.  By the way, these are the sensors.  We actually picked it up from the medical industry.  So these are designed to monitor patient body temperature and doctors can actually stick it onto the patient’s body and then monitor it remotely via their various apps like Windows, iPhone, or Android apps.  So we’re trying to bring it back into the food industry.  Right?  That’s the idea.

So what I’m going to do now is basically — by the way, the temperature is on the screen like they’re all sort of plus, but the reason they’re plus is not coming from a vehicle, because the sensors are here.  So you’re looking at the room temperature on that app screen.

But now, if my Wi-Fi signal doesn’t drop, you’ll be able to see the temperature has gone up because I’m holding it in my hand.  So it is now reading my body temperature, and is real time putting it into the Power BI dashboard here.

JAMES PHILLIPS:  So let’s land that because it’s incredibly powerful.

MUSHTAQ AHMED:  Absolutely.

JAMES PHILLIPS:  So from a wireless sensor to a Lumia phone to Azure Event Hub into Stream Analytics into Machine Learning and Power BI.  And so you’ve closed this loop.

MUSHTAQ AHMED:  Absolutely.  And you mentioned this is like our transport department or the customer service department looking at this screen all day long.  We believe that drivers, actually, because they’re very busy people, they have to drive long days, they have to deliver to customers, all the goods on time.  But then somebody can actually look at this and pick up any kind of anomalies.  If a door has been left open, if the freezer has broken down.  Right?

So the temperature is actually being monitored remotely directly from within the vehicle using these little devices and using all this Microsoft end-to-end technology.

JAMES PHILLIPS:  Incredible.  And was it hard?

MUSHTAQ AHMED:  Well, not really.  Not really because we didn’t have to do much because a lot of these things are pretty much out-of-the-box solutions and they’ve been tried and tested.

So what actually also can happen here if customer service agents or transport agents they actually want to look at some of the queries here, and you can see the natural language query, we don’t really have to go develop it to run some analytics out of it.  And what it can also do, with one click, you’re going to put it into the dashboard.

JAMES PHILLIPS:  So showing the pinning.


JAMES PHILLIPS:  So this data streaming in real time.


JAMES PHILLIPS:  You’re seeing it on the dashboard, but it’s also being collected in a different kind of data technology.

MUSHTAQ AHMED:  Absolutely.

JAMES PHILLIPS:  That allows you to also do analytics.

MUSHTAQ AHMED:  And you can actually publish it into a dashboard.

JAMES PHILLIPS:  Incredible, close the loop.

MUSHTAQ AHMED:  Yes.  Let’s run another one and we can see some numbers here as well coming up.  You can actually go to the dashboard.

JAMES PHILLIPS:  And keep building that up.

MUSHTAQ AHMED:  Yes.  Absolutely.  I can do another one.

JAMES PHILLIPS:  What are you looking at now?

MUSHTAQ AHMED:  So we can actually go back to the dashboard and we’ll see some of those things.  That’s right.

What we also have got, we’re remotely monitoring the battery life of the sensors.

JAMES PHILLIPS:  And what are you going to do with that?

MUSHTAQ AHMED:  OK, so putting all this technology in the truck and then you don’t want to let the customers down, not let the drivers down, right?


MUSHTAQ AHMED:  So we would like to hook this data back into the machine learning engine again.

JAMES PHILLIPS:  The battery life?

MUSHTAQ AHMED:  Battery life.  And we would like to prevent the drivers or the sensors to go out of any connectivity or battery life.  So once we collect enough data, we’ll actually put into the machine learning algorithm, and we’ll predict when the battery is going to run out so that it never runs out.

JAMES PHILLIPS:  Interesting.  And then you’ll just do maintenance on the schedule.

MUSHTAQ AHMED:  Absolutely.  We’ll hook it up to the vehicle maintenance schedule.  And then, in time before they actually totally run out, the batteries will be replaced.  And they’re actually quite long.  I mean, I’ve been running it the last three months, they’re still going well, 90 percent.

JAMES PHILLIPS:  90 percent.

MUSHTAQ AHMED:  Absolutely.

JAMES PHILLIPS:  Thank you very much.

MUSHTAQ AHMED:  Right.  OK, due to the restrictions of time, we obviously couldn’t show all the details here.  But I would like to invite all the people in the audience to come to the expo area because we have got all the sensors and the apps running.

JAMES PHILLIPS:  You’ve got a truck out there I think.

MUSHTAQ AHMED:  Absolutely.  There’s a vehicle there, the sensors are there, and you’ll be able to see it, touch it, feel it real time.

JAMES PHILLIPS:  Thank you very much.

MUSHTAQ AHMED:  Thank you.  (Applause.)

JAMES PHILLIPS:  OK.  So we’ve kind of run the gamut here.  We’ve talked about Power BI making it super simple and you’ve seen it in motion multiple times.  The completeness of our data platform, the power of the intelligence layer, the analytics systems.  And we showed how flowing to people, eventually leading to action, really does unlock the value.

One of the things that we just saw, and in fact, we saw it in both demos, is not only do we provide an arc that allows you to get intelligence to people who can take action, but we also allow you to create roles and to automate so that action is taken automatically.  Machine learning, the ability to learn behavior and to alert when things were going wrong, the ability to change pricing based on learned information and even doing that automatically.  Wash plans in the future to adjust pricing in real time based on current conditions, much like you see on freeway tolls.

Power BI provides that human connection.  The automated connection, the ability to automate is provided by Microsoft Azure.

So that brings us to the end of the presentation.  Microsoft provides the most complete data platform, the most powerful analytics layer, and perhaps most importantly, we make it super easy.  Connecting people, connecting your machines to this platform and unlocking business value.

Super easy to get started.  Power BI.com, Satya announced earlier today Power BI was announced to be available in 140 countries, so this is now available for preview globally.  I’d love to hear from you.  I’m [email protected] if you have any questions following this.  I hope you give it a try.  Thank you very much.  (Applause.)