00:00welcome to the a 16z podcast I'm Michael
00:02Copeland every organization these days
00:05is clear about the need to get its data
00:07act together but that doesn't mean the
00:10path toward data bliss is clear data has
00:13gravity it resides in different places
00:15at different organizations on premise in
00:18the cloud and flowing from external
00:20sources and the rate of change within
00:23organizations is always different so an
00:26approach towards handling data that
00:28works for one company may be the exact
00:30wrong thing for yours Steven Sinofsky
00:33leads a conversation with three founders
00:35Pratt Mogae from cocina Gaurav Dillon
00:38from snap logic and romance tonic from
00:41good data about the opportunity and
00:43variety of ways forward for companies
00:46looking to make the most of the data
00:48that matters Steven Sinofsky kicks
00:51things off this is they're gonna be
00:54super fascinating because the behind the
00:57scenes all of you share a very similar
01:00set of problems and challenges and
01:02opportunities when it comes to dealing
01:04with data often what differentiates you
01:06from your competitor is how do you get
01:09the data and what do you do with that
01:10data and what decisions do you make
01:12based on that data and it's a world
01:15that's just being completely inverted
01:16from what we used to think of data used
01:19to be the province of a very small
01:20number of people who would generate
01:22reports print them out move them up the
01:25chain and distill them down and has been
01:27into talked about in PowerPoint slides
01:29and now we have the opportunity if you
01:32build out the right infrastructure to
01:34access that data analyze it look at it
01:36and make choices all from a mobile
01:38device all using the cloud and so that
01:41is is the centerpiece of this section
01:44and what's really interesting is that we
01:46represented in you our CIOs and CMOS and
01:49so we have a sort of a supplier consumer
01:52relationship that we want to we want to
01:54navigate the desire for ubiquitous
01:57access the needs of security the
01:59challenge of on-prem cloud hybrid cloud
02:03private cloud and then just the desire
02:06for faster or more and better and so to
02:09explore this topic I'm super excited to
02:13executives and founders of portfolio
02:15companies that will introduce themselves
02:17as they they join us here okay
02:20founder and CEO of cocina I'm Gaurav
02:24Dillon I'm founder and CEO of snap logic
02:26I've been in the data business for
02:29formerly co-founded informatica and as
02:32chief executive built up built that up
02:35into a decent sized public company snap
02:37logic is version 2.0 a journey and I'm
02:39pleased to be here Steve and Almonte
02:41Stanek founder a CEO of good data so I
02:46kind of I trying to decide where to
02:48start with with this and I think I just
02:51want to start with a big question that I
02:53do want each of us to look at a little
02:55bit which is just demystifying the big
02:57part of data and and help people to
03:00understand that how as building out a
03:03new company in the new sort of cloud SAS
03:06you know mobile world what what is big
03:10about Big Data yeah yeah sure yeah so
03:14it's interesting you know on one hand I
03:17tend to cringe every time I hear the
03:19word big data but on the other hand you
03:23know if you just look at the world you
03:24know I was before cocina
03:26I ran the product line at nutiva which
03:28was a big data warehouse appliance and
03:31if I remember our customers back then
03:33anybody who had three or four petabytes
03:36of data was considered a huge customer
03:38now when I talk to people they talk
03:41about you know mobile data social data
03:43existing data and so petabytes is no
03:46longer you know like being up there on
03:49the other hand we talk to many customers
03:51where it's not about volume it's about
03:53you know having existing data but just
03:55being able to get it together to analyze
03:57it faster so a lot of it is about
03:59agility of data I sort of defined big
04:02data as it's a mindset
04:04it's about being really fast about using
04:07data to make decisions so it's not just
04:09about petabytes of data it's about you
04:11know how fast can you leverage data to
04:13to create business outcomes and and so
04:16that mindset is what is different now so
04:18just to help with a little house except
04:21for cuisine in particular where is
04:22cocina cocina on the stack of
04:24this this problem what is the Xena yeah
04:27but not not too much of a pitch no no
04:29 but really I'll go yeah
04:31so what could that's always a founder
04:33challenge you know you asked for the
04:34pitch and you know right away you get it
04:36yeah we solve world hunger problem now
04:39you know what cuisine is essentially is
04:41trying to move big enterprises to
04:44leverage the cloud for their big data
04:45processing and so what we're seeing is
04:47large enterprises CIOs CMOS they're at
04:51this crossroad where the stack is
04:53transforming and clouds coming along so
04:55there's an opportunity for a new
04:56platform but they're all wrestling with
04:58figuring out how to use the cloud how to
05:00make it easy and that's what we address
05:01sure so so taking a slightly different
05:04perspective you're coming at it from
05:06from above talk what yeah like what's
05:10big about your big data yeah well you
05:12know it's bigger than oh but actually
05:15you know the sort of breakthrough for me
05:19on all this is if you think about the
05:20data warehousing industry which is about
05:23a 10 billion dollar industry some local
05:25successes here in Europe Business
05:27Objects phenomenal success in the 90s
05:29various other things at the earthquake
05:32etcetera but but basically if you think
05:34of the 90s what we essentially had was
05:37the industry around data warehousing and
05:39analytics there was fundamentally about
05:41the barcode scam right here's a
05:44technology that was invented to help you
05:47get out of a supermarket faster you're
05:49standing in line you can checkout faster
05:50that begat an industry of analytics
05:53Nielsen and IRI would count how much
05:56beer was it more you know local brands
05:59people drink more Stella or something
06:00else you know and now we're going beyond
06:05that sort of comparing this verses that
06:07versus geography in big data to me the
06:10fundamental breakthrough is providing
06:12information from multiple places and
06:14producing insights where the data finds
06:18the data so for example a consumer
06:21packaged goods company traditionally if
06:23they were looking at the sale of
06:25lipsticks would be looking at in a
06:27classical business intelligence way
06:29price volume geography etc but when you
06:33bring in a social media stream
06:36and you see the discussion around that
06:39particular product you find that out of
06:43stock is like a big deal so what the big
06:46data provided is an insight that out of
06:50stock because colors of lipstick come
06:51and go being out of stock was a huge
06:54issue to people and more importantly
06:57that it was out of stock because people
06:59were ending the life of that product
07:01based on volume whereas this should be
07:04and now they are looking at the lifetime
07:06value of the product particularly
07:08lipstick shades that apply to minorities
07:11or someone who may be a lower volume
07:13purchaser but once they find the right
07:15shade they're gonna buy for life so so
07:17to me data finding the data is the magic
07:22of big data that is the promise that is
07:24being fulfilled in a predictive way that
07:27we never could do in the 90s it's not
07:29about volume it's the context it's not
07:30about volume but with with good data
07:34part of it is is actually putting that
07:36in front of your your your typical
07:39member of the marketing team the sales
07:40team the field how does that fit into
07:42the the big side of big data absolutely
07:45we believe that you know with all the
07:47investment in Hadoop and this and how do
07:49you know most companies are still data
07:51bankrupt you know Hadoop or data
07:53warehouse or whatever is a place where
07:56data goes to die and our goal is to
07:58actually change it and and we see good
08:00data as the last mile of analytics you
08:04know that's the last mile that connects
08:06connects your user as your business
08:08partners your business networks internal
08:10and external audiences with data in
08:13Hadoop and with data in data warehouses
08:16and so on and it's kind of non-trivial
08:19because we all kind of know how data
08:21works and so on but our customers are
08:24people literally in the field and and
08:27people who manage stores and people who
08:29manage you know sandwich shops and so on
08:32and we need to deliver data to them in a
08:34way they can actually understand and
08:36there's a big kind of impedance mismatch
08:38between the way the data is in data
08:40warehouse in Hadoop which is actually a
08:42huge advantage of a Duke that it can be
08:44stored in so many ways but it doesn't
08:45help somebody who manage assembly shop
08:48to do actually underst
08:51and so our goal is to be that kind of
08:54like a lot of smile of analytics and the
08:56way we actually do it is that you know
08:59we actually let our customers big banks
09:02big telcos big you know insurance
09:04companies and so on the white label good
09:06data and sell it under their name so so
09:09you know we have about half a million
09:10users and very few of them actually know
09:13they use good data because they see
09:14somebody else's logo but it's okay as
09:17long as they get access to the data
09:19which is kind of the biggest problem
09:21so but what I find fascinating about
09:25about trying to navigate this space is
09:28is that in most corporations finding the
09:30answer any question is is often
09:32incredibly difficult and yet I want to
09:35know like anything is there a movie
09:36ticket available how many cars are
09:38available to drive me somewhere can I
09:40get a plane ticket a hotel like as a
09:42consumer I have like this immense access
09:44to data and so I think what what is it
09:48like how do we break down that barrier
09:50because I think representing the CMOS of
09:53the audience like they they they know
09:56that all the bar codes are being scanned
09:58they know that you're using a great
10:00reporting they know it's there but there
10:02there's some impedance mismatch what
10:04what is it it's good I think I asked a
10:09good question if they're fighting over
10:10answering it so I think you know you can
10:12give your perspective but I think that I
10:17sort of take a contrarian point of view
10:18too I think it's not about it's not
10:22about technology first off so it's not
10:24fundamentally about saying I want to ask
10:25any question I want because the moment
10:28you take that approach it then becomes
10:30like you were saying it is a Hadoop
10:32store you know can I ask any question as
10:34opposed to sort of saying what are you
10:36really trying to get done what's the
10:38business outcome and so what we've seen
10:41then you've looked at many big data
10:43projects the ones that fail are once
10:45where people have taken this approach of
10:47saying I want to collect all the data
10:49and then I want to figure out what
10:50questions I can ask I want to look for
10:52hidden patterns as opposed to people who
10:55sort of looked at it and say I got a
10:56marketing problem I don't know how to
10:59you know how to track my existing
11:01customer so that I can upsell
11:03I want to convert an existing customer
11:05much better I want to give a better
11:07experience and and so whenever they've
11:09approached it at with a business problem
11:11and then to say what data do I need to
11:13bring together to answer that question
11:15you have lot better in terms of
11:18formulating a narrower scope of those
11:20projects and asking those questions any
11:21time it becomes like just collect the
11:23data and figure out what it is then you
11:25you start having those issues so it sort
11:27of the top-down approach was at the
11:29bottom of i'ma slightly different
11:30vantage point you know and my vid is
11:32agree no not yet oh by the way did you
11:35say something yeah we have a 5 pound bet
11:38on who says you disagree first he won
11:40see he said that so so you know the
11:43vantage point we have is and what I've
11:46come to believe in again looking at the
11:4790s and looking at this decade in the
11:50century is that it is it is not that we
11:53a priori know what the problems facing
11:56us are we don't because there's too much
11:59going on thus too much change you know
12:01all the way from a world economy
12:03recession entrant of competitors it's
12:06just the the intensity of change is too
12:08great and I think sort of the hangover
12:10that the data industries had is this
12:13whole data warehousing batch hangover
12:15and and the truth is we're living in a
12:17world of streams of information and
12:19consumers particularly you know people
12:22who come in came into the workforce in
12:23this century Millennials have an
12:26expectation of wanting stuff now right
12:29so the vantage point that we see is
12:31combining streams and in a sense be or
12:34use an overused word mashing things up
12:38and providing new insights is hugely
12:40important and that is dynamic and it is
12:43done in an interactive way and broadly
12:45speaking you'll have some kind of
12:46thought like are we trying to sell more
12:47widgets or are we trying to kill the
12:49competition or whatever but but you
12:51don't know exactly how do you engage
12:53well let me ask you now let me ask you
12:55this because I think part of it is is is
12:58how how do we get from a model of like
13:01every Friday I'm supposed to show up and
13:03find the products that aren't selling
13:04well or find you know where do I need to
13:07stock something - where does exploration
13:09how do you enable enable that that's
13:11what I was actually going I actually
13:13believe that the biggest problem of data
13:15not Big Data small data any data is the
13:18rate of change of business you know you
13:20you don't want to be doing the same
13:22search every Friday and and in a current
13:25set up ID is supposed to gather and
13:28curate data and business is supposed to
13:30do the exploration and it doesn't work
13:32you know for the the IT to really kind
13:35of be you know in charge for the tools
13:38and the time timeframes and so on a
13:41turnarounds they are way too long for
13:43business to actually kind of depend on
13:45it and so business then goes to excel
13:49and and other products that you might be
13:52familiar with and I'm gonna return to
13:55that and so that's the biggest problem
13:59and I've been in so many meetings with a
14:01CMO and CIO where there's like a zero
14:04cognitive overlap you know they have not
14:06the same kind of interest and so on and
14:09so I actually believe that you know your
14:12example that I is a consumer I have so
14:14much information and so on that's
14:16actually not a good example because the
14:18information I get is some sort of a
14:20curated by you know by Intuit and
14:22curated by Google and so on and business
14:25people don't have that kind of
14:26experience in any company and that's why
14:28they go to tableau and Excel and and
14:31click tech and so on because they
14:32essentially kind of resigned on i.t
14:34getting them kind of that information
14:36and it's it is in Hadoop and it is in in
14:39data warehouse and so on but there is
14:41this kind of you know again it's the
14:43last mile and I'm not saying that we are
14:45kind of able to solve it you know
14:48generically at good data but we are at
14:50least solving it for certain types of
14:53problems you know getting data to
14:54business networks getting data all well
14:57the boundaries so there will never be
14:58one solution for everything because the
15:00biggest problem and the business the the
15:02rate of change is way too high but I
15:04feel like there's no good you talk about
15:09this exploration idea right the and then
15:11coming back to Romans example so we have
15:13you're working with speaking about how
15:16data changes business business models
15:18there is a fast growing restaurant in
15:21the u.s. this is kind of like the next
15:22Chipotle and and the guys there they
15:25grew up in in Chipotle which again is a
15:28Chane but they basically decided to
15:30build ground up completely differently
15:32right and the way these guys are
15:34thinking about data and and people
15:36driving in to - you know these guys
15:38serve fresh salads but they look at you
15:42and say this is Steven Steven likes
15:44eggplant or Pratt's vegetarian or you
15:47know and also their whole idea is that
15:49if I could profile people coming in at
15:51noon as they come in
15:53I'll basically figure out how to build
15:55the right product for you
15:56right so it's fresh but it's it's
15:59customized for you but you still want to
16:01do it at scale so there's a whole new
16:04breed of we heard this this morning like
16:06the full stack you know sort of the full
16:08stack app thing like verticalized
16:10experiences everything that matters I
16:13think that's where it's going I think
16:15what it's going is all that data gets
16:16surfaced it's in a product it's in in
16:19some form there if it can be surfaced to
16:23you get magic so let me let me ask you
16:25that then is is this is what you
16:27described like so that's not like a
16:30person doing a report so help me to
16:33understand is there is there's some
16:36elements of a whole new style of
16:38analysis based on machine learning based
16:40on incorporating other data source where
16:42how does this because I think that this
16:44leap is super super critical to
16:46understanding but that where why the new
16:49tools have to be cloud-based why they're
16:51there they do what they do so so it's
16:54it's not about what's not selling on
16:57Friday but why it's more about what
17:01could we do based on the data that we
17:03have that would help us be more
17:05successful without doing a traditional
17:07business intelligence and it doesn't
17:09matter whether they do it on premise in
17:11Excel in the cloud the traditional
17:14rearview mirror view of business
17:16intelligence has some element of return
17:18but it's also at some point been well
17:20done there are ways to improve that but
17:22we're getting to a point of diminishing
17:24marginal returns on that where the
17:26returns are is the wealthy people in
17:29technology are doing predictive
17:31analytics and trying to figure out what
17:32their data is telling them and in that
17:34they're using machine learning
17:36algorithms and certain kinds of
17:37open-source algorithms many of which
17:39come from Berkeley's amp lab and they do
17:42categorization next corner grassroot
17:45university California Berkeley well has
17:47a whole variety of some of the leading
17:49technologies that like spark is clean
17:51item you do can download them and use
17:53them right so but you need the people
17:55behind it so what we have now is a new
17:58population of user who is using the data
18:01from a Hadoop or something and that
18:03person's a data scientist this is now
18:05widespread outside of financial services
18:07in a Goldman Sachs Morgan Stanley always
18:10had quant jocks now everybody has quant
18:12jocks and how can we obliterate the
18:14barrier of enabling that person with the
18:17information in near-real-time to do
18:20better job of predicting their company's
18:21future I think it has shifted the battle
18:24has shifted to that so okay well now I
18:28can't even talk about data scientists I
18:31absolutely believe in what you said this
18:33is kind of it you know machine learning
18:35and so on unfortunately some you know
18:38most companies are not big enough to
18:40have big enough sample for machine
18:42learning you know big banks what makes
18:43Google Google what makes Amazon Amazon
18:45is that the data sample is so big that
18:47it can actually really learn from it
18:49typical company would look at their if
18:51you know thousand invoices and there is
18:52nothing to learn from so I actually
18:54believe that that's why analytics is
18:56done in a cloud as actually a lot of
18:58value because we see data across tens of
19:01thousands of companies and we can
19:02actually do machine learning from you
19:05know massive data sets that individually
19:07don't actually mean anything but
19:08analytic that's reports no no it's not
19:11it's not predictive and right we're
19:13gonna take it back so let me let me let
19:16me let me let me turn it around and and
19:19ask it asked him in a little bit
19:21different way because I think so you
19:23there were two people mentioned Excel so
19:24that was that was my that was my cue to
19:26talk about Excel yes so so my what what
19:30I think is so interesting is that if I
19:32were to to query the room upon and we
19:34would find out that most people find
19:37Excel the most valuable analytical tool
19:39that that they're using and and there
19:41may be two reasons let's just touch on
19:43you know one of them is just that it's
19:45the one that they can use that does what
19:47they want but another one is that the
19:49gap between the CMO and and the IT
19:53organization is often there's data
19:56and that there is some source like it
19:58could be geographic data it could be
20:00like wow this report doesn't even list
20:02all the stores we have or all the
20:04outlets or it doesn't have our web logs
20:06or there's just another part of the data
20:09that isn't yet in the do isn't in some
20:11lake and so so much of the job is is
20:13just bringing together and then applying
20:15that knowledge because I do think that
20:17that's what differentiates you know like
20:20the difference between between
20:21Minneapolis and Bentonville in the u.s.
20:23is is not necessarily the product they
20:26sell but I started with this one and
20:28it's gonna be short I always believe
20:29that there are two types of people in
20:31the world people who can use Excel pivot
20:34tables and people who cannot use Excel
20:36pivot tables you know and we always
20:38belong in one of the categories and
20:40believe me the people who can we worked
20:42really hard to make pivot tables you
20:44know people later in the afternoon we're
20:50gonna hear from the actual person who
20:51made them easy to use in 1989 and so
20:55your example actually assumes that
20:57people actually know how to use to be
20:58able Excel they were tables and that's
21:00why some of the most frequently used
21:02kind of data analytics tools extremely
21:05basic because they actually look and
21:07feel like sheet of paper like
21:09two-dimensional sheet of paper and you
21:11know so that's that's the problem with
21:13analytics that on one hand we have you
21:14know very complex systems with you know
21:16SPARC and Hadoop and so on and yet you
21:19know most of the people actually using
21:20those tools you know don't like the
21:22abstractions they like sheet of paper or
21:24rows and columns and and so it's very
21:27difficult to bridge that you know that's
21:29that's why you know you actually see a
21:32lot of kind of analytics that's
21:33something that's successful it's either
21:35embedding so it actually already is kind
21:38of you know very you know vertical and
21:40industry-specific and so on
21:42or people are using you know I believe
21:44that we all of us compete with excel in
21:47bodies you know somebody looks at it put
21:49it Excel and and sends it over you know
21:54Microsoft's a good partner and investor
22:00I'm a very different point of view on
22:02this which is just a show of hands in
22:04this room and don't feel afraid how many
22:06of you use tableau in your shop are just
22:10two hands I find that hard to believe
22:12well the on Excel yeah so sort of my
22:17view is that tools are really hard to
22:20change because tools usually embody a
22:23business process right they embody so my
22:27views that like large companies
22:29particularly the ones where they're
22:30analytically driven I think the way this
22:33is for them to really leverage fast
22:36changing data fast changing the
22:39fascinating world you got to sort of
22:41figure out how it gets there like you
22:43were asking this question right so the
22:45legacy data flow is probably going to
22:47stay on premise for a while now the
22:50question becomes how do you leverage
22:51these new technologies how do you
22:53leverage the cloud and so it's going to
22:55be an augmentation strategy where there
22:59is this concept coming up which is
23:00called the pipeline right and the
23:02pipeline idea is that data is like a
23:04river it's it flows and so maybe some
23:07part of this data whether it's external
23:09or internal will flow into the cloud you
23:11know certain kinds of processing will
23:13happen and then over time what will
23:15happen is you'll take that data and then
23:17maybe land it in certain place where
23:19data scientists can analyze it some data
23:22will continue to go to excel some data
23:23will continue to go to tableau or the BI
23:25analysts so it's not going to be a a
23:27world where you know everything just
23:29disrupts right overnight people who do
23:32things will continue to do it but yeah
23:35so let me let me ask every tool but the
23:39point is every technology is good for
23:41doing something it doesn't subsume spart
23:44doesn't subsume data warehousing hadoop
23:46doesn't subsume you know streaming so
23:49there's just like different technologies
23:51so speaking of subsuming which is a
23:53great a great way to ask this because i
23:55I do want to recognize that the CIOs in
23:59the room are dealing with a very you
24:02know real challenge and real opportunity
24:03which is I'm guessing for most all the
24:06people in this room their system of
24:09from structure sequel Oracle based
24:11system and for all the CMOS that's the
24:15starting point for most of the data that
24:17they they need to get to how how do what
24:20message how do we help the customers in
24:23the room bridge that reality that they
24:25deal with or said another way like
24:28where's the opportunity how do they
24:30start a new project what do they do in
24:32order so maybe I think so so you know we
24:35get asked this all the time sometimes I
24:39those people use my former products and
24:41so on so so look here's what we
24:42recommend first of all I believe the CIO
24:47and CMO have kissed and made up
24:49in a big way in a big way well they're
24:51all here they're there they are together
24:53sometimes in the same table nobody's
24:55nobody's hit somebody on the head with
24:56anything yet but but beyond that what
24:59you have is this concept of a pipeline
25:01we used to call it an information
25:03it's a pipeline now because has
25:05real-time streaming attributes but
25:06people should be thinking about being
25:09able to use the new price-performance of
25:11Hadoop spark to obliterate their
25:13traditional data warehousing appliance
25:15not unplug it but the rising tide of the
25:18data Lake we think will drown out the
25:20data warehouse in the fullness of time
25:21so that is a like that's an interesting
25:25exponential change that's probably a
25:27huge opportunity that actually that
25:29might be worth taking thinking about for
25:30a second which is even if you have a
25:32petabyte in your your structured stores
25:35today if you turn on the right sources
25:38within the company you'll very quickly
25:40have lots more than that to potentially
25:42work with that might end up being even
25:45more valuable indeed and you know I
25:48actually want to answer your original
25:49question like how do we actually you
25:52know how do we deal with the fact that
25:54most data is on-premise anyways you know
25:56cloud-based company I need to go my
25:57business so so our focus is actually on
26:01data that goes across the firewall
26:04anyway you know so we really kind of
26:06help companies to monetize the data and
26:08and deliver analytics to their business
26:10networks our biggest customer is a one
26:12of the large credit card issuers and
26:14they have merchants they have you know
26:17issuing banks they have
26:19acquirers so they have most of their
26:21audience for data actually sits outside
26:24of the firewall so instead of emailing
26:26data in in CSV files or excels and so on
26:29we actually help them to really kind of
26:31build a kind of analytics distribution
26:33platform and I believe that that's you
26:36know we don't have time to wait for
26:37companies to move the data the primary
26:39data to the cloud that may not even
26:41happen but more and more kind of
26:44datasets are being used in this kind of
26:46across the firewall insult in mobile
26:48mobile scenarios and so on and that's
26:51where we see kind of 90% of our focus is
26:55is it really in case that I mean I I
26:57didn't get one of the things that's so
26:59interesting is is that in general
27:00there's just more data outside of your
27:02organization than there is inside our
27:04net a key part of the tools is just how
27:07you connect those I think there's a few
27:09things going on there's definitely a
27:11shift towards the cloud it all depends
27:13on the verticals some verticals are I'd
27:17say more skeptical there's regulations
27:19that basically say certain data cannot
27:21physically leave but in most companies
27:23even with financial services we've
27:24noticed that they're very eager to
27:27explore the cloud either for external
27:29data or to even look at all your
27:31internal data and not all data as equal
27:33some data is PII some data is not and
27:37and there are newer technologies that
27:39allow you to encrypt data in motion
27:40addressed the clouds mature this very
27:44sophisticated security control so you're
27:46seeing you know it's not as religious as
27:48I as it used to be before and that's one
27:51of the reasons why you're seeing CMOS
27:53and CIOs come together because there are
27:55platforms that where the CIOs can
27:57basically now stand up projects to move
27:59data provision things in AWS on redshift
28:03as you're and run these projects so that
28:06they feel like they're part of a private
28:08cloud right while while they're really
28:10running on public cloud and so there's a
28:13very quickly it's changing yeah it's
28:16always one of the challenges in these
28:17panels you want to we want to address
28:19the breadth of needs and and realize
28:22that that people are always gonna be at
28:24a different place and making this change
28:25at the same time you know beyond
28:27and what makes this kind of you know
28:29slow down is this kind of local
28:32regulations and so on instead of putting
28:34one data center for you know my audience
28:38you know user base we need to build
28:40multiple data centers and that kind of
28:42fragmentation and balkanization of data
28:44will continue and that's gonna be more
28:46and more difficult for cloud vendors to
28:48manage that and manage understand all
28:50the regulations and so on so I think you
28:52know look the the fact is data has
28:54gravity and you have to respect that if
28:56your data's on-premise you should
28:58probably put your Hadoop or other kinds
29:00of analytics on premise if your data is
29:02in the cloud you're using a website
29:04hosted in Amazon or a Deutsche Telekom
29:06or Swisscom in Switzerland then
29:08obviously it makes more sense to have
29:09analytics located there
29:11but I think in all cases what is
29:14blindingly clear from looking at the
29:17various people we talk to is that you
29:18have to have a predictive analytics
29:21capability however you do it on premise
29:23or in the cloud you're dead without it
29:24you know one do certain things make
29:28sense to migrate to that and certain
29:30things not you know if you're a credit
29:32card issuer conflict resolution how does
29:34a customer return a product there's no
29:36money in trying to put that into a
29:37modern system you've somehow got it
29:39going you worked out the disagreements
29:41leave it where it is but trying to
29:43understand how a social media stream
29:45interacts with your product how people
29:47are complaining to Twitter rather than
29:49to your call center is a great
29:50importance to you right so we think that
29:53that is probably what people should do
29:54is get a predictive analytic strategy in
29:56place and start to bring based on data
29:58gravity the right sort of technologies
30:01to solve that issue feel free
30:03I think you've talked about technologies
30:04and tools the other thing we've noticed
30:07it's all about people so the the shift
30:10it's a transformation and transformation
30:12always begins with leaders yeah and
30:15we've noticed that it's CIOs CMOS
30:17forward-thinking guys sometimes its CTOs
30:20that helped push that sometimes it's
30:22CEOs chief data officer as chief digital
30:25officer this is one great guy in the
30:29ossama fired from Barclays I don't know
30:31where you're sitting some of but you
30:32know he's he's had those experiences
30:34before and so when he's in a financial
30:37services company he's thinking about how
30:39dear sister so I think it's all about
30:41getting those people and then bridging
30:43basically the gap awesome well thanks
30:46everybody for a lively discussion and
30:48appreciate the insights on on data thank
30:51you very much everybody