USF Business Analytics Forum – Panel Discussion

USF Business Analytics Forum – Panel Discussion


Welcome to the Inaugural Forum on business
analytics in Florida. We’re really delighted to have you here. We’re so excited. Thank you so much for being here and we’ll
get to the panel right now and I’ll briefly introduce all for of them and then subsequently
I’ll introduce them again in a little more detail but you know we have four brilliant
people from our area; Tracy Bell from Bank of America, Joe Kwo from Fintech, Jim Stikeleather
who is a USF trustee and has had many hats in his career for many organizations, and
Brian Fuhrer from Nielsen. And we’ll be talking to them about supply
chain analytics, media analytics, analytics and innovation, and the future in social media
analytics too. So let me very quickly get into our first
panelist here. So Joe Kwo whom some of you know is a USF
alumnus and he is CIO at Fintech and the company itself processes alcohol invoices for more
than 400,000 clients, 25 billion in payments annually. And so this was a very interesting, this is
a very interesting company that went from a payment processing firm in the alcohol industry
delivering alcohol to consumers to becoming an analytics company and Joe essentially along
with the CEO Scott Riley led that transformation of an invoice processing, supply chain company
into a very powerful analytics company today and a very niche market so please join me
in welcoming Joe and the thing that I want Joe to talk to us about is you know for him
to take us through this transition from going from invoice processing to analytics and share
with us some examples of what Fintech is doing in that industry today so Joe… Thank you very much Balaji. So Fintech has been around for 25 years. The word Fintech nowadays is industry but
we were around before that maybe before that people think we were training dolphin but
we never train dolphin we are in public service to make sure that you have your preferred
alcohol when you go to the location that you want to buy so and thank you very much for
your contribution to our costs. So lets go through this one then. Fintech as Balaji mentioned Fintech is the
leader is processing regulated invoices. We were U.S. Chamber of Commerce for the whole
nation. We serve all 50 states. More than one million lines of transactions
every day but 30 to 36 billion dollar a year of our transactions. That’s a lot of alcohol obviously. We provide more to over more than 500,000
nationwide relationships. So let’s try it again. Still not? Ok. So what, when we were processing payments
we have a lot of data that is very dirty in terms of raw data. So what you call UPC in the world of alcohol
beverage is almost nonexistent because in the alcohol world there are three tier systems
it’s not like suppliers you cannot like suppliers cannot sell directly to retailer so they have
to go through distributors. There are about a 3,000 over 3,000 distributors
in the whole nation and Fintech has already connected with about 95 to 98 percent of them. The reason is that every distributor send
differently in terms of data. For example, Budweiser six-packs there may
be about 600, 700 ways of the data being received and still counting. Everyday there is these changes that happens
on the distributor side that is going to make data is very hard to recognize for retailer. So the way I give example to everyone is that
nowadays everybody is being asked to drive a car 70 miles per hour and change the tires
at the same time. Imagine that we have all this dirtiness in
the data, basically you don’t have proper gages you don’t even have windshield wiper
to clean your windshield so you are driving your car with very dirty windshield so that’s
the way this data this retailer trying to drive the beverage alcohol industry in terms
of hard to see the transparency in terms of supply chain in terms of managing pricing
in terms of knowing for example like one retailer that I’m not going to name the name they have
about 6,500 locations around the nation. Prior to Fintech they were able to only roll
out to maybe about 10 to 15 percent of their locations to sell alcohol beverage because
it’s not scalable in terms of the main power in terms of how to do the consolation how
to manage inventory and everything else because it’s just the data is just that dirty every
state has their own distributor even to the county level. So you multiply that by 50 you multiply that
by hundreds of thousands of products it’s impossible to do that. With Fintech we basically partners in the
project that now they can roll out to those 6,500 locations. So Fintech provides in terms of alcohol pricing
in terms of whether price discrepancy exists or a purchase habit and you know in terms
of the product. Some of this may seem so trivial to you but
imagine that when you are managing all these locations with the number of skews that they
are to manage it’s quite a challenge to do that. Let me just play this video because even without
sound but it really tells you the extent of Fintech business nowadays. So at the beginning basically right delivery
in terms of made to stores to retailer’s locations is only cash checked or money orders. So to retailers it’s very very hard in terms
if trying to keep up with this because like if the manager is away then the delivery person
needs to wait and everything else. So we started with EFT payment over two decades
right so it saved the distributor in terms of the costs of running those cold trucks
especially for beers you have to run those running those cold trucks are hundreds of
thousands of dollars. For distributor that enroll in Fintech they
can save those money and they don’t have to wait 45 minutes for manager to issue money
order or anything like that. So that’s what they saying is that with all
the data historical data that Fintech has been processing we have been able to analyze
to do analytics on those data all those impurities from the data we can filter them out and remove
them so that it become data that become real information. And just like mention earlier it become the
right information and become the right it become actionable knowledge. And not only that in terms of timeliness of
the data right so we have one big retail chain that told us before Fintech it would take
them about four man hours four days of man hours to get the data while right now with
Fintech they can get it in 30 seconds. So right now we are becoming a comprehensive
platform it’s not just payment it’s data analytics and also we are basically partners to our
clients in terms of compliance because for those of you who do not know how the alcohol
regulation work is that if you say you have 60 locations in Florida if one location didn’t
pay according to regulation then all 60 locations in Florida may not be able to sell alcohol
until it gets straightened out or worse you could lose all licenses. So that’s where in terms of regulation and
everything else. This video is also on YouTube if you just
Google Fintech Tampa because there is another Fintech company in South Africa I believe
but you know it’s not related to us. So but a with the sound it’s kind of cool
in terms of because in less than three minutes it basically tells you the whole span of our
business in terms of payment to all the way to data analytics. In terms of what Balaji mentioned in terms
of the impact of data analytics like I mentioned earlier about those giant retailer that couldn’t
roll out without our capability to clean the data and everything else but in terms of Fintech
itself in terms of the growth of business those companies that were not really interested
in payment just payment now become really interested in us because we are more than
just payment we provide them analytic and data allow them in terms of optimization of
business operation and everything else. So I want to mention in terms how invaluable
USF has been to us. Fintech has been USF practice center since
2009 specifically for ISDS department. USF has been really a partner to us right
so for businesses around this community around here projects that we did were projects that
we would have not been able to do with internal resources but USF has been there and it’s
very very flexible in terms of accommodating our needs we have been working with (unintelligible),
Balaji, Warner, and (unintelligible) and I really highly recommend it in terms of like
use USF as a resource here because it can and it will definitely help you. Thank you. Thank you so much Joe. Can we take any questions for Joe on alcohol
analytics, supply chain analytics for him? So let’s move on and then we can have questions
again for the panel. So I’m going to turn it over to Tracy Bell
now Tracy is senior vice president of Bank of America doing social media analytics and
Tracy has been doing it since 2007 very very early days in social media when very few companies
were doing it and if you look at the companies now ranked on social media analytics Bank
of America is actually very high in that group and they have also worked with us for many
years and the thing that we learned from Tracy which I keep telling all the students when
they do these social media analytics projects for the bank it’s very business problem focused
it’s not open ended fishing and that has led them to do amazing amount of creative things
but the methodology that they use is so good and she keeps telling people that it takes
as much time to find nothing as it takes to find something. Right and so from a bank patient point of
view I think that’s a big lesson for the big stakeholders and the company so thank you
so much Tracy and I’d like Tracy to basically tell us about her journey and what are the
kinds of use cases in social media analytics that she has worked with Bank of America so
thank you Tracy. As Balaji mentioned we started this journey
in 2007 it was I had other responsibilities at the time but I felt it was something that
we should explore that you know I was interested in and I love working in unchartered territory
I’m not one of those people that likes to color outside the lines, I don’t want any
lines at all you know I want to start with something that hasn’t you know doesn’t have
a trail to follow. The biggest challenge back in 2007 is this
data is very different than most of us had worked with in the past. So when you think about typically when you’re
doing analytics and you’re working with numbers I mean the nice thing is a three is always
a three you know it doesn’t matter who wrote it down it doesn’t matter where it was written
down a three is a three but in the world of social data we’re dealing with human expressions
with thoughts with emotions and virtually every word that’s said has multiple meanings. Another layer in social is that it’s not just
the words, punctuation changes the meaning, capitalization or not changes the meaning,
who you said it to, what was happening in the world when it was going on. So you can imagine you know starting with
what appears to be just a glob of text and the metadata that has to be wrapped around
it to bring meaning to it and there weren’t any standards for this at that time. There was no one with experience to hire other
companies were doing this so but they were where we were, we were all muddling our way
through but there wasn’t a lot of talent on the market looking for a job in this phase
but that turned out to be just a huge blessing for the entire project because I ended up
with this collection of extremely diverse thinkers that were open-minded, enthusiastic,
and just driven and focused on bringing structure to what is a very unstructured and messy world. Public data there was hardly any of it and
we’re just talking about public data here tweets and so forth comments on websites,
ratings on products and so forth any place the consumer can post content. There was very little out there. Twitter was producing 20,000 tweets a day
now in comparison today it’s half a billion a day. So and of those 20,000 they were not talking
about brands they were talking about pop culture and brands didn’t have much of a presence
there so you know instead of having big data to work with we had the data petri dish which
was fine for where we were. The tools that we had at he time were immature
they weren’t built for scale but you know they did what we needed at the time but the
good news is because interest was increasing social platforms began advancing very rapidly
and just you know fortunately instead of getting tied to a tool we knew the tools were not
where they needed to be our demands were far beyond that so we really focused on our business
processes that could plug in to any tool that again turned out to be of tremendous value
to us because the processes themselves are still in tact today with a lot of evolution
there but it starts with a business process and we don’t care where the data comes from
we can switch at any time. No standards there were no standards at the
time so when you think about in our discussions earlier this morning a lot of people talked
about selecting the right metric. Well we didn’t have any metrics back then
we had define what is a meaningful metric about human emotion or statements or feelings
or opinions and we were evolving well how do you measure sentiment how do you measure
impact how did that move the dial what do you call you know how do we measure visibility
is it actual views is it potential views? So the industry all of us that were in this
space at the time we were struggling with what we call these things, how we define these
things, we were learning as we went and comparing notes with everybody who would talk to us
about them. And lastly businesses themselves didn’t understand
the data it was messy didn’t yet trust it because there was no business application
there was no business value this was a big experiment at the time. So I did spend a lot of the first several
years educating people about what it is what we can do with it and so forth. The good news is by 2008 we had a very mature
capability and those of you in the finance industry know what happened in the fall of
2008. At that time when the financial crisis hit
we were armed with some amazing information that that we were able to get real time rapidly
and feed that back into our decision making processes and became an invaluable resource. So oh now let me see if I move the slide correctly. Ok so as we move forward to today it’s still
data with inherently human qualities. It means we do have to spend a lot of time
it’s massive, massive amounts of data. We talked about Twitter having a half a billion
posts a day there’s also and nobody knows the exact number but lets estimate about 150
million site generating content from consumers that is public and available. So massive amounts of automation require to
collect data extract it and then on top of that again turning that mass into meaningful
information really takes the pairing of human ingenuity and basic data science algorithms
and so forth to continue to tune what we’re looking at everyday. Everyday we’re looking at what has changed
in the data, what new words have been invented, what happened in pop culture today that spun
off a new word, what did a reporter say that is suddenly now an acronym or something like
that. So we have resources focused just on understanding
the changes in the data and accounting for it in our logic. So and again we’re applying this at scale
adding that to adding that context into every post. The metrics are finally consistent, reliable,
and widely understood and I’m I say that from our front line to our c-suite to our board
of directors this is data that they look at they understand. My CEO can tell you what our sentiment scores
are and he can also tell you what it means and that is very rare. Another thing that has changed a lot is social
media is now image based so this is a new challenge that we’re working on if you see
the the graphic over there on the side we sponsor a lot of major league baseball events
and easy to pick up our images. Image recognition itself is something it solves
but it’s very difficult when you have no text translate that into so what do you do with
that? So for us we say you know image recognition
says yes there’s our brand yes that’s a major league baseball ok that goes towards sponsorship
nobody’s saying anything about the bank but that goes through viewership and then we have
engines that parse all of that. So whether if someone comments about an atm
but they don’t mention which atm if they take a picture of it, we have it and so if someone
says this is the worst atm ever this is the best atm ever somethings broken somethings
you know anything about one of our branches one of our facilities if they happen to get
even a fraction of our logo in it we can capture it, we can use it and get it to the right
place. But again those are some of the new challenges. And also what makes the data more interesting
when we first started data was something that was read in a linear fashion it was you know
look at the word cloud it’s flat it’s meaningless now that we’ve enriched the data so we understand
the context and who is speaking and who is driving the conversation who people are talking
to who’s listening we’re looking at data in a relational format where we can literally
walk through the conversation and what you’re looking at right there is an actual conversation
I think that one might of been from a sporting event I’m not sure. This isn’t bank data specifically but it’s
the tool that we use to map out what, who, how is starting the conversation and sometimes
you’ll see like a bubble outside of the conversation you’ll see the big map and it looks like there’s
a satellite and as a group of people talking to themselves and that in itself has value
as well you know whether or not they’re influencing the global conversation or if it’s a group
just off to the side. And lastly what we apply and what we apply
this to is virtually anything that you can imagine this is an input yo our decision making
processes across the business it’s still one of many inputs social data alone doesn’t make
or break any decision it’s just it’s an important piece of the overall puzzle but it’s one that
is genuine it’s coming from consumers in totally unedited form unprompted and it happens real
time and so much information can come in so quickly it does become an invaluable source
whether we’re analyzing a marketing campaign our product developers I mean the best thing
you can do in cellular mobile banking is look for people using the words I wish in front
of something related to mobile banking and that goes straight to product development
and we don’t care if they’re talking about our brand we don’t care if they’re are talking
about any brand. If it’s a product we’re working on it’s a
way that we can see immediately what are some unmet needs. We look at the entire industry our competitors
we trend over time we have kept all of the data from day one so we have this opportunity
to go back and learn to understand you know compare current situations to prior situations
what worked last time what didn’t work last time becomes a invaluable resource for one
comparing you know not just what actions but was the performance of this campaign good
compared to what how did it really move the dial and these are things we literally around
the company watch real time we have what we call visia screens where if sentiment changes
from a headline the entire company can see it see the needle move at one time. Messaging improvement feeding back so if we
put out a press release if someone tweets I wonder what they meant by so and so we know
we missed our mark there and we have some clarification to do and the nice thing is
we can immediately go out and do that. So if we see questions coming up out about
a new product it informs our marketing department what to talk about next and then we listen
again to tell them if we solved the problem. So our conversation with our consumers is
two-way they tell us, we correct, we listen again to see if we adjusted it and that is
it in a nut shell. Thank you Tracy. Any questions for Tracy on social media from
the group here? But you know I had one very brief observation
and then a couple of years back there was research in academia that showed in social
media that if something stays on social media for eight hours it becomes true and so when
we spoke to Tracy about it and that time interval is shrinking rapidly and so when companies
are faced with very noisy messy social media the speed of response now is becoming so critical
and you know I think there Tracy you know have you seen specific applications now I
know Bank of America has a lot of people who delicately respond here but have you seen
customers now expecting that kind of response now? Yes I do I mean we’re in a real time world
in terms of listening but absolutely when it comes to responding directly to customers
the issue is also staffing for that in a way that meets their expectations so that is a
continually challenge of finding the balance between the resources and the costs to have
you know enough staff on hand. We do have service windows in which we commit
to respond by but the customer would like that to be shrunk they literally want somebody
watching their individual twitter stream 24/7. Yes. The question was monitoring third party products
or in a capabilities in house. We do use we do partner with some amazing
companies for the base infrastructure things that we do not want to don’t need to be in
the business of and there is a collection of like I said certain of that solves niches
we glue those together with our processes but our secret sauce is the rules engine that
sits on top of their data and that’s what we do in house so we give them what they do
best we keep what we do best. In one of your earlier slides you mentioned
that the business really didn’t trust the data initially or know what to do with it. They didn’t know what social media was either
so it was fair. Talk about innovation and a little bit about
this common issue. What would you say was the flex point that
changed and what do you contribute to the champion of the acceptance. The question was about Tracy can you go ahead. Oh yes so I apologize I should’ve repeated
the question for the audience. She was asking about what are what makes a
difference in getting new companies or companies to accept new processes new concepts I mean
how do you sell that. One, you’ve got to put a good sales pitch
together to begin with and then you go out and you find those sponsors and you keep knocking
on the door and you don’t give up and then you tie it to business cases and like I said
I literally spent three years educating going up and down the entire company explaining
what it was why we were doing it and what the payoff was. I actually had a pretty easy time I think
in the scheme of things but it did take a lot of active outreach. Alright so I want to move to Brian next because
you know we’ll move from social media analytics to media and measurement and you know I first
want to thank Brian a lot for being here I mean today is an extremely busy day for him
at Nielsen they’re hosting a massive number of people clients and Brian was insistent
that he wanted to be here for this panel so he is a senior vice president at Nielsen in
Tampa he been he’s a USF alumni been with Nielsen for a very very very long time he’s
led tremendous initiatives in media measurement and analytics at Nielsen so you know he is
here today and you know he can tell us about so many different things but mainly I wanted
to ask him about in his role at Nielsen and where Nielsen is this company that measures
TV and so many other things, what do you see at the forefront in media analytics and measurement
and you know tell me some of the interesting things you are working on today. Sure maybe I better stand here so I don’t
get feedback but I have really good news before I start I don’t have any slides so that’s
a your welcome. I sent over 70 slide deck and Lorie kept saying
oh I can’t load it I can’t load it and I got the hint. Before I start I just wanted to make one quick
comment Moez opened up today’s session and he talked about USF and the college of business
being the greatest college of business in the world and President Genshaft being the
greatest president I think we’ve got the greatest college of business dean right here so let’s
have a hand for Moez. I don’t know how many of you have dealt with
Moez but if he asks you for something you have to understand he doesn’t know the word
no back he hears maybe or not yet but you might as well just say yes if he ever asks
you for anything but we love yo Moez. A couple things I wanted to talk about really
briefly that we have going on that are I think really important. First off, a number of years back we started
a strategic alliance with USF and it was like nothing we’ve really ever done before with
any university anywhere and it’s a data sharing alliance that really costs Nielsen nothing
but a lot of work to put the agreement together and to share the data. So one of the things that’s really intriguing
is we share basically anything that’s requested by USF of the vast quantities of data that
we produce with USF for their research purposes and I encourage you if you have data if you
have access to data you think about that because it’s amazing the benefits that we reap back
and the types of studies there’s no way we would’ve been able to execute as a result
of USF and what they’e been able to do. They’ve been great partners and of across
the whole world with Nielsen’s 45,000 employees we hire more USF grads than any other college
in the world. So that gives you an example of the benefits
that have accrued to Nielsen. From the standpoint of things that we’re doing,
two things from a applied analytics perspective. The first one sounds a little bit unsexy when
I but it’s had a huge impact on the television industry as you know viewing and fragmentation
of viewing and sample size is a really big deal. At Nielsen we measure what people watch and
what they buy I work on the television the national television side and we use we have
a sample we had a sample of about 25,000 homes about 60,000 people we had under measurement
for the national overnight ratings that we deliver. In those homes we have what we call people
meters which are little boxes that people log in that identify themselves as to who’s
viewing. We also had other panelists and other homes
for other purposes in the United States we had about 20,000 homes that didn’t have any
boxes so what we did over a period of about 18 months and working with the networks and
all of our clients and advertising agencies we started a new approach where we combine
both panels and we modeled on the viewing using a technique called viewer assignment
on to the homes that don’t tell us who’s viewing. So it was a major initiative that really took
us from 22,000, 23,000 homes to about a 45,000 home panel now we have about 120,000 people
in our measurement. That wouldn’t have happened unless we had
a lot of modeling techniques and data scientists and a lot of analysis that went into making
the industry comfortable that we could make that kind of a change. It’s a very from the standpoint of motion
and changes. We don’t move quickly in our industry because
we support about 70,000 or 70 billion dollars worth of advertising expenditures are transacted
on our data so everything we do is scrutinized very closely but this was something that really
changed the industry and made a big difference. That’s the one that you know pure sample size
increase was a big to our business. But from the I’d say the sexier side is what
are we doing beyond TV? And we’re doing something that we call digital
ad ratings and measuring all kinds of mobile computer and digital measurement as well and
we’re doing that in a completely different way instead of having a panel and measuring
what’s happening in a panel we do a couple things. One we have a software meter that is integrated
into every app and player and different ways of distributing video for example and we measure
every single on a census basis every single exposure to content in that way. But the interesting thing is we bring that
back and then we partner with a little company in California called Facebook maybe you’ve
hear of them that a and we get all of our demographics we match what happens on the
different devices with the Facebook ID, Facebook has age and gender. So that’s a big data approach and we use a
lot of big data in a lot of the things that we do but we also use a lot of little data
and when we talk about our panel really think that the panel is probably the most well curated
media research panel that exists based on the amount of advertising expenditure that
transacts on it. But what we do with our panel is we get all
the data back from Facebook and then we calibrate it. We adjust it based on the true set. So as you work through analytics and you apply
things that you’re learning what you’ll find is big data and little data I think are going
to become more and more important over time as more and more people use this. It’s not just big data it’s big data that’s
calibrated to a true set so you have a higher degree of confidence with the inputs. So for example, on Facebook older people say
they’re younger I’m 26 younger people say they’re older so they can get on and actually
use Facebook so we have to take the data in from Facebook then compare it against data
we have in our panels to calibrate those data sets. So those are just different things that we’re
doing you know from a standpoint of analytics. A lot of work happening from the panel side
to understand what’s happening with Netflix and other things that the new ways that people
are consuming video but just a couple of examples. Last thing i just want to close with is saying
I’ve had the opportunity to work with Balaji and the whole analytics group for some time
now and it’s just so exciting to see this group in an inaugural session and how this
whole thing is really coming together and blooming and I just want to say thank you
to Balaji for all of your work. I’m muted so any questions from the audience
for Brian. Yes. You know I should know that but I don’t know
off the top of my head but I know that it was from what I understand at least as good
as last year that was their prediction. I think the question was about the (unintelligible)
and Brian can explain that as well what Nielsen is doing about it. Yeah the question was what’s happening with
cord cutting what’s happening with people getting content in different ways than traditional
and thanks for asking that because it gives me the opportunity to point out it doesn’t
matter what people are doing on their TV we measure 100 percent of the time on TV so whether
it’s coming from Netflix or whether it’s coming from your local cable company whether it’s
an over the air antenna or whether it’s coming from one of the newer services like Sling
TV and we call them virtual multi-channel video providers. We’re working closely with those because we
have to cover everything we have to be able to measure everything so our clients can make
the decisions that they need to do so exciting time to be in this business exciting time
based on all the changes that are happening to be able to measure it. Thank you and I think you know one question
for you Brian on the kinds of things you measure you know beyond people are aware of TV for
instance can you share a few non-obvious ones that people don’t know Nielsen is measuring
things that you can talk about. Yeah so from a standpoint of I started off
with what we measure what people watch and what they buy we also measure all retail sales
we’ll partner with all the big measure we do this globally and we do it everything from
Walmart what happens at Walmart all the way down to small villages that are kind of cash
transactions in Africa so there’s a lot of different types of measurement and we have
to keep pushing the envelope whether it’s social whether it’s the internet you know
and every market is different and that’s why we have to kind of tailor it to exactly what’s
happening in the market and what they can support because the concept of you know big
data doesn’t hold in places where it you know it might be a bizarre or village where things
are transacted so we measure that as well as out of home measurements and you know we
equip people with little devices if they’re in bars and in Balaji’s example in health
clubs and you know what they’re exposed to so there’s a lot of different ways that we
measure to make sure that we track what media is being exposed. Thank you Brian. So let me a let’s thank Brian one more time
and I will move to our final panelist of the day Jim Stikeleather and you know he is right
now he is a USF trustee and also very interestingly a student in our DBA program Doctor of Business
Administration. I’m guaranteed to graduate. He writes his own certificate so we have to
sign. Now the really interesting part about Jim
is throughout his career he’s been an innovation guy. Creativity innovation and he has done a lot
he’s brought together analytics and creativity and innovation the kinds of things that we
really look for in a center and the kinds of things that are so rare. His most recent job before he came here was
chief innovation officer at Dell and he lived for many years here and he kept flying out
there and back here so he is going to talk to us about analytics and innovation and very
interestingly it kind of comes full circle because Ronny Kohavi’s started the day with
how analytics drives innovation right through experimentation and it’s very neat that we’re
going to finish the structured version of today back to the link between analytics and
innovation so Jim it’s a pleasure to have you here please join me in welcoming. Thank you. I should start by saying that one of the reasons
that Balaji asked me to come here is I’m going to be the curmudgeon today. I’m going to tell you all the bad things about
analytics and then I’ll come back around and tell you why it’s important. And the reason that I have that perspective
is as he said I have been doing and involved in innovation since the late (unintelligible)
so I’ve got a lot of experience and one of the things I can tell you when you start thinking
about innovation is innovation that we did back in the 70’s oh I gave it away anyway
innovation back in the 70’s and innovation today overtime the only real change has been
which was the department of no. Back when I began when you had an innovation
you wanted to try something new and different the department of no was a legal department. No no no no way too risky you can’t do that
our brand will be at risk. And then a little bit later on it evolved
and it became the finance department. Sorry this isn’t big enough to be interesting
to us. I mean how many people have heard that one. And then a little later on it turns out that
the IT department became department no. No no no no our systems won’t support that
you can’t do that. In the end my last job the department of no
was the analytics department. The numbers don’t support this. So what I’m going to share with you is how
I got around that and how analytics actually can help innovation. How many people know the story about for one
of the nail the shoe was lost for one of the shoe the horse was lost for one of the horse
the knight was lost for one of the knight the war was lost for one of the war the kingdom
was lost. Well the same thing holds true with analytics. If you don’t have the end data you can’t do
the analysis and if you don’t do the analysis you can’t create a story and that’s what I’m
going to focus on because the key to tying analytics to innovation is the ability to
write and tell stories. Now we’re going to have two seconds of totally
shameless self promotion for a second and about late last year HBR brought together
about 12 of us to put together a book to facilitate communication between the decision makers
and peoples who are doing analysis and it was basically to be a gentle introduction
to decision makers and what the analysts were trying to tell them. Now interestingly some of the feedback we’ve
already gotten from it is the fact that it’s actually very useful for the analysts to be
able to phrase their communications appropriately for managers to understand them and I only
put it up there because the only reason that you are doing analysis is understand the numbers,
make better decisions and very importantly present and persuade. And this is where a lot of analytics groups
fail miserably is the present and persuade. So let’s talk a little bit about the forms
of analytics and what they’re useful for. There’s obviously the script of analytics. How many people know who Florence Nightingale
is? Everybody knows she’s famous for saving lives
in the Civil War right? Actually what she’s famous for is she invented
that pie chart in the upper right very few people recognize that she was actually the
person who invented representing information that way. Which by the way basically proved that people
were dying in the hospitals not on the battlefield, if you just left them on the battlefield they
survived at a higher rate. And we won’t talk about where medicine has
progressed since then. The other thing is exploratory analytics being
able to look at the data get an understanding and feel for the data which is one of the
things if you’re going to write a story we’ll talk a little bit more about that later and
then there’s predictive analytics and then there’s the mother of all predictive analytics,
technical analysis. And they are still debating whether technical
analysis works or not. And lastly there is simulation which is the
area that I’m particularly interested in as we are starting to deal with analysis of more
and more complex systems and in fact if you are doing analytics you are doing systems
engineering of a sort you’re understanding a system you’re trying to get a feel for it
you’re trying to model it and what’s really become interesting is as systems have become
more complex we’ve had to move into this area of simulation or agent based modeling and
that becomes critically important a little bit later on when I start talking about the
fallacies of analytics. The other thing you have to know about analytics
is who are you presenting to and why are you why are you doing it and who are you giving
the information to. Several people have talked about pick the
right analytics and do the right processes. And really when you start looking at analytics
at least in the business situation what you’re thinking about is I’m doing analytics to confirm
something I have a belief about the way my business operates I’m going to collect the
data I’m going to pull up a dashboard and I’m going to confirm the business is operating
appropriately. Or education I’m going to teach people about
what’s going on I want to give them a feel for this system that we’re talking about. Exploration, which is discovery. Execution is you see a lot in the area of
prescriptive analytics if the data is telling you this have the system or the machines or
your sales or reduce the price the airlines have gotten really good at this in the area
of load balancing they change the prices based upon the dynamics of whats going on in the
sales. And lastly innovation. Now there’s a video up there you can watch
it but it’s only there because I want you to see a great example of somebody who is
taking analytics and turning it into a story. Known by the name of Hans Rosling which anybody
who’s in analytics is familiar with Google him he has a whole bunch of Ted videos what
you’re going to watch is him taking data from all kinds of sources, compressing it, and
then telling a story about economic development and life expectancies. Below that now that’s a video and it’s telling
a story and it’s kind of a little bit real time but below that is a very very famous
infographic from the 1800’s. You know history doesn’t repeat it rhymes
right? So what you see there is Minard’s representation
of Napoleon’s march to Moscow and it’s return. How many people have seen that? It is an incredible infographic because there
is not a single bar chart there is not a single spreadsheet there is not…but it can tell
you everything you need to know about that entire process when he left France, how he
got to Moscow, what happened to his troops, what he encountered at what locations. That’s a story. That’s a story on a single infographic. So let me reinforce that the value of analytics
comes into play when the person who is doing the analysis or the team that’s doing it can
take that analysis and put it into the form of a story because you can get up in front
of a group of senior executives and you can present all these charts and all these bar
diagrams and all of these spreadsheets and all of these numbers and 35 seconds after
they’re out of the room they have forgotten every last thing you said. But tell them a story and they will remember
that story, they will remember it in the middle of the night, they’ll remember it when they’re
having a shower, and they’ll be able to make decisions with it because it turns out our
brain is designed to take stories and retain stories. How many have heard the term seven plus or
minus two? Basically it’s how many individual bits of
information we can keep in working storage to make a decision about. That’s absolutely true when you’re dealing
with data allowance. Finite things. When you take those and put them together
in a story it becomes increasingly unlimited the amount of information that can be retained
to make a decision with. So let’s talk a little bit about the obligations
of the storyteller to the audience. Needs to be a compelling narrative. You know the nice thing about telling a story
is what you are doing is you are crafting not just the data but what the data is implying
you have protagonists you have antagonists you have a theme you have a plot line all
of which can tie a whole bunch of information together for someone to be able to make a
decision. You also have to be aware of who your audience
is. You have novices. You don’t want to over simplify but you want
to get them interested you want to wet their appetite. A generalist generally knows what you’re talking
about but is really looking for the major themes. Managerial in depth action I’ve got to make
some decisions need to be aware of the intricacies and the interrelationships. It’s really, I don’t know about you but I’ve
never been successful at drawing a five dimensional chart but I can tell a story with a plot that
goes through five different areas and communicate that information to them. There’s also an obligation on the part of
the person whose doing the analysis to be objective and offer balance. You may have your prejudice, you may be trying
to persuade but when you put the story together you want to out it together in as objective
way as possible because later on stories are being very useful for collaboration. People can share a story and add to it and
subtract from it can see nuances and see things that you’re not going to see if you were just
telling it with charts. Now a little nit earlier you heard the term
minimum viable product which is one of my favorite terms of all time. One of the things that you run into when you
do heavy analytics especially in new product development is you come up with what is it
437 different types of features that need to be added in. What happens when you talk about a story is
you start narrowing in on what’s important you start seeing the interrelationships that
generally don’t come out of the data when you’re just looking on the data. You’re also connecting that data to human
experiences you’re putting it in to a form of that’s more meaningful to putting a subjective
perspective in the data that says hmm this is how people will probably really want to
do this. You begin to see subtleties out of the data
because if you’re telling a story you’ve got to start…wait a minute there’s a gap here
how do I fill in the gap. It forces whoever is presenting the story
to be concrete. Not an abstract we saw this we saw this, turning
it in to something real in terms of the real world that the decision maker can make. And as I said earlier the brain basically
enjoys stories so that you’re going to get more attention paid to it. It’s also our brain is wired to convinced
by stories which s how we get in trouble a lot in politics because we tend to narrow
things down to very short sound bite stories but the fact is your brain’s designed to do
that. Lastly, stories are very very sticky. And what i mean by sticky is they stick with
people once they’ve heard them. You can tell a story to somebody, they can
generally tell that story to somebody else maybe little errors little bit later. Give them a bar chart they probably forgot
it two seconds after you gave it to them. The last thing I want to talk about is why
it is so important the key to innovation stirred by these stories. The first thing that has to happen is trust
in the analysts. You’ve heard it discussed several… that’s
probably fundamentally the thing you have to accomplish first. The other thing that’s interesting about analytics
is in a lot of companies it’s become very narrow and very specified. Finance will have its analytics people, marketing
will its analytics people. One of the great things about story writing
is those people have to get together and they have to write the story together or you have
a group that can actually do the story writing based upon that. The other thing that’s important is to avoid
the analytic traps. As you heard I’m a student here I’m in the
process of getting my doctorate so in a typical professorial mode you have homework. Feel free to look up Goodhart’s Law, Murphy’s
Law, Campbell’s Law, McNamara’s Fallacy, and the Lucas Critique. You will find the very inner…how many people
know what any of those are? Murphy’s Law which is the least interesting one. My favorite is Goodhart’s but also the Lucas
critique and I won’t spoil it for the analytics people but basically Lucas won a Nobel prize
for this which basically says you can’t predict the future with analytics but that’s another
story. Ok so feel free to look those up. Stories also facilitate collaboration and
most innovation is not the product of one. How many people believe in the lone wolf innovator? Doesn’t happen, never has happened. The reality is is what the innovation research
tells you. There’s always at least three people or three
personas involved: the idea generator, the person who comes up with the idea, the idea
manifestor, the person who actually takes the idea and makes it real, and the idea communicator,
the person who can tell people why it’s a valuable idea. That’s a collaborative activity. All innovation is collaboration. Also all innovation is based upon failure
but that’s another speech. The reality is is the story supports collaboration. Putting a chart up on the wall is not going
to get a whole lot of collaboration you may get a lot of people questioning the data or
questioning how you drew the chart or how you chose to do that. A story will cause collaboration. It’ll also initiate imagination as opposed
to arguing over facts you’re actually taking it a little further and you’re doing a little
bit of imagination so you’re encouraging other parts in your brain to get engaged. It tends to be big picture oriented versus
the details and what you do is the collaborative activity the details start to emerge as people
fill in the gaps in the story. And one of the, one of the things I have found
that’s very very useful is the idea that it encourages disassociation. If you’re dealing with a story and the protagonist
in the story you’re not talking about well your department screwed up last month. It’s a totally different environment when
people are trying to make a decision. And the last thing is, actually two last things
is one of the big complaints I have with a lot of analytics groups that I’ve worked with
is the ease with which they dismiss the outliers. I will tell you most innovation is found in
those outliers. So it’s the ability to incorporate the outliers
into the story that becomes incredibly powerful. And how many of you have ever engaged in a
discussion with a five year old? You know that if you dare answer a question
that you don’t know about you’re doomed because he’s going to ask why and why and why and
why. And that’s the nice thing about a story because
a story let’s anybody in the room contribute a why question without sounding ignorant. They may not know the ratio between metric
x and data point y but they feel perfectly comfortable asking why a certain thing was
happening in this story. So with that, and I hope you enjoy the cartoons;
I am the last speaker for the day so we’re almost done so thank you very much. Thank you Jim. Questions? From this group. Moez. Can you give us an example of a story? Ok this is…yeah I can tell this one because
it’s in the public and now you know. One of the projects we had at Dell had to
do with we would take a lot, we did a lot analysis of future trends and signals coming
from the marketplace and we had to then turn and craft those into a story. So I’ll tell you what I’ll tell you two stories. One is what we call the frictionless economy
story. The frictionless economy story talks about
two good ole boys Floridians obviously get in their boat and row out to the middle of
the lake and they’re doing some fishing you know they’re casting they’re reeling they’re
casting they’re reeling, and suddenly they realize that just a minor modification of
this reel they can catch a whole lot more fish. Or being good ole boys from Florida drink
more beer and catch the same amount of fish but you get the general idea. So they row their boat back they get on their
PC they immediately get onto a cad cam and I don’t know if people are familiar with the
makers movement but the makers movement is where people…if you’re familiar with open-source
makers movement is the real world version of open-source. People get together they build new things. So they get on they find the a cad cam version
of a reel they make their modification they share it with their friends they get some
suggestions they go out they open up they post it up they open up a kick starter campaign
oh they print it out on a 3D printer, I haven’t told the story in a while I left…ages since
I left Dell, print it out on a 3D printer show it to their friends they get feedback
they make modifications. They open the kick starter campaign, you know
raise a couple of hundred thousand dollars, send it for offshore job shop manufacturer,
open up an Amazon store front and the next time 30 days later they go fishing in their
67 foot yacht. That’s a story there were all kinds of analytics
that were underneath every step of that story which then people came back and wanted to
know the numbers but any executive at Dell can tell you that story now and remember it. And that’s an example of telling a story with
analytics underneath it. I won’t tell you the serendipity economy one
because it embarrasses me. Jim I had a question for you on that right
so the thing about stories and data is that a lot of the stories can turn out to be false. Right? And it’s very easy to do that now Ronny before
he left took us through the testing process in experimentation. Right so how do you see those two coming down
I mean are there dangers of telling stories that turn out to be untrue because it hasn’t
gone through the validation and how do you see that? Yeah and absolutely a great question. What we found was you go from the data the
analysis the analytics you create the story you engage people in the story that will then…which
actually results in new questions to go back and do more analytic research on. And so as you’re getting into what I would
call a virtuous circle of that then what starts appearing is ok the data is not right or we
didn’t do it correctly. You get testing because you’ll have somebody
will say ok well the person could’ve done this or could’ve done that then you can go
back and test against the data. So you have to think of it not as a deliverable
but so much as a collaborative process where you’re writing a story and then you know kind
of the decision makers or the editors that are going back to the story writers and say
ok you need to clean this up and your plot line is not clear here which is really the
analyst going back and doing another cycle through the data. Thanks Jim.

Danny Hutson

Leave a Reply

Your email address will not be published. Required fields are marked *