Industry sign banking massachusetts presentation fast
so how is the session at the Deaf stage nice right fantastic thank you for coming so you know this today we want to talk about a very journey that we started sometimes back around three years and the journey was to provide a platform to help business users to see and understand data better right it was a very small experiment and which grew into a revolution and like every revolution you need a revolutionary there and we have our own in Lockport Walla his senior vice-president finance at HDFC Bank and today we are going to share our journey of you know our three years at with tableau of what we did and how we have helped people get information better and at the right time thank you bowing over to you thanks Abbie shake so I'm just gonna keep moving around I don't like standing and I know they've made this thing so I'm just going to keep moving around right quick introduction right I'm Pavan and I guess as I go through the presentation you realize that almost everything that I've have put on the slides somebody or the other has covered in right so you've been through various you know the keynote yesterday there's two slides gone from my slides right if you've been to the Bank of America Merrill Lynch presentations as to mow slides gone so I can just actually go okay know what that's Jays that's Merrill that's JPMorgan and I'm done but I'm just gonna run you through these Thanks jambavan I I part of the finance function I look after the business intelligence unit which takes care of almost everything that is relating to performance management I also have your shashi with me who manages our IT side of things so he's my IT partner I don't understand ID so please could be scanned in touch with him in case you have questions around IT the presentation is much more functional driven I have some snapshots of what we do where we do this and you do that quickly run you through where we are and I just a show of hands how many are from financial sectors so that I can get a sense of I don't need to explain some of the terms right again how many of you have heard of HDFC Bank that's good no customers I hope so just tryin to go run through it 2015 11 we started off the journey itself and I have Deepak sitting here used to hate Tablo in India at that point of time who came to me and started talking to me about what data visualization is all about and gonna run through I've skipped a year because sort of didn't make sense doing 17 and 18 it's just merge it together since there are very few people no let me just run you through what HDFC Bank is I also look after Investor Relations for the bank so I'm gonna quickly run you through I wouldn't do justice to my role as IR if I didn't tell you anything about the bank so we are a large bank right in an Indian context about 97 percent of our branches 97 percent of customers and our assets are in India for those of you who are already on tableau will understand the fact that I've made the graph using tableau using the pin codes the dot represent actually locations so as you can see we are across the length and breadth of the country India is a large country we have 43 million customer we continue to acquire about 2 to 3 million customers every single year as we speak the branches number is a little bit dated we just release our half year results we are about four thousand eight hundred twenty five branches now the unique thing about that one and the challenge which came in a different kind of a challenge when we started doing tableau or was it the ruler in semi-urban if you see about 50 percent of a branches are in the deeper geographies now that sort of creates a different set of challenge because connectivity in India is not necessarily as you would have expected everywhere everywhere else in the country oh by the way we still a very young bank you know just having a chat with the gentleman here he is a hundred and fifty million five hundred and fifty year old can Bank we got created in 1995 so we had a twenty year old company actually be just about twenty twenty-one year old company so we're pretty young as far as banks are concerned this is a slide from my investor presentation it's a bit busy but to give you a sense of we do everything right the businesses are broken up into retail wholesale and Treasury within the retail banking we do loan products and we do everything from microfinance which would be a hundred and fifty dollar loan all the way going up to an SME very giving funding to the local small businesses to the large government corporates in the country and to the government as well so we are the second largest collector of taxes for government of India on the wholesale side of things as well we covered everything from project finance to doing merchants equation and advisory as well both these segments actually use tableau as we speak today so Treasury is the only segment at this point of time where as I have not yet given tableau in terms of what what data we present to them but for wholesale and retail today both of them are on cello how did we get there and just two before that to give you a sense of where HDFC Bank is in terms of number of branches you know they are there between Wells Fargo you know JP Morgan we're there in there about there five six thousand branches we have five thousand branches but if you see in terms of assets under management or total asset size we are actually very very small right a dollar is 74 today so that sort of explains that but if you look at the market cap right we are actually more or less where Goldman Sachs and Morgan Stanley is so we pretty large in terms of value that we create for the investors and shareholders we are listed on the new york stock exchange by the way so if you guys want to buy my stock go ahead write to it write in numbers number of employees we have about 95,000 employees so you know you're a large bank in terms of customers but we also have a very large employee base in terms of India where we are because you know while I can keep discussing about going to from a global market but as I said 97 percent of my branches and customers are in India from a market care perspective we are by far the largest bank in the country we are the third largest company on the stock exchange SBI is the largest bank and SBI would rank among the top 50 banks in the world in terms of assets that's been around like forever so if you can see from a total asset size they're there and there abouts in almost three times our size as we speak but that sort of gives you we are a large bank as far as India is concerned okay that's my I are bit done I'm gonna dive into the other role that I play which is the business intelligence banks and you know I sell a lot of hands from a regulatory science in banks and financial institution a highly regulated country you know HDFC Bank is no different in fact Reserve Bank of India which is our regulator is even more conservative than some of the other regulators in terms of freedom so what we've done is we've created the business intelligence unit to mirrors that of the business itself so you know what you saw earlier there was a wholesale unit so I have a business intelligence team for wholesale the same thing but it's half retail and a separate team for Treasury within retail I have different segments creating for assets or liabilities so that's how we've thought through and created the business intelligence unit which is the single central repository for all performance management data and I think that's where we are slightly different from some of the other presentation that I sit in wherein you have somebody else managing the data and somebody else managing tableau this is where we all put it in one single unit so we are the only people responsible for doing performance management there's no other report which goes out from performance metal things as you can imagine you know ninety five thousand people the amount of attrition that we have the risk of somebody leaving a branch and taking with him customer data or you know host cough host profitable and not was one of the key reasons or one of the key risk that we were running on an ongoing basis so data governance and what can be seen at the right people was very very very important to us why did we choose to have low the simple answer was to replace Excel at that point of time right so I'm back in history 2015 the only thing I wanted to do get rid of Excel right that was the wondrous data this and that's where again you know point to deeper because he ended up spending a lot of time with us as the tableau champion at that point of time ahead of tableau in India and actually explained to us how some of the other banks have done this right so that helped a lot that gave us the confidence to at least try to proof of concept at tableau as you can imagine you know we have the articles and the IBM's and the Microsoft this world already incumbent partners we have tried a few things with them and they obviously have a similar offering or offering which they can compare but we went ahead and did the POC with tableau I actually remember conversation that he had with my boss the CFO and he's promised my boss that he could do presentations from tableau and I just saw the devil on stage so say well you know what you can now do the presentation three years late that's okay what we did is we didn't really take tableau to the masses straightaway right so for those of you who know starting on the journey right start with a smaller group we took it to be the wholesale group small group small set of people data itself is very small but it's quite deep so the amount the things that you could do with the wholesale customer is very different from things that you can do with a reader customer you can sell many more products from across set is much higher it's more complex so we chose a complex data set or a complex report and we took those set of customers and saying we're gonna start with the B or C with the whole sales Jim just the beginning was just to getting to know tableau right so yeah we use tableau the local offices and local representative to help us getting to do the first dashboard that we did what we did is we took the monthly account profitability report so every every wholesale customer we create whether what what's going on in the last one month we took that as a base and we did a three-month pattern or a run quick advice to those of you want to start doing this don't stop the existing report so if it's going in Excel let it go in Excel start doing something together so we did a three dimensional run and at the end of the third dream and I just switched off the Excel I interesting insight by the way is people still came back and said you know what I need the Excel and I said look it's here right why do you need something and they were what we thought was an end state report they actually took and created databases out of that to make further repos for them right so I'm sure all the first signal I can see some faces noting and I say yeah yeah you know I know what's going on but because of that what we ended up doing and that was the first time that we opened we created the you know while we're doing this right it's it's we rolled out the first desktop license so we identified those people saying hey what are you gonna do with this and we rolled out the first list of license importantly when we're doing all of these things we didn't go to business users to say give me a you give me an ROI right I bought everything the server the desktop in fact in a few places I actually installed physical desktop to say hey you know what just use this forget it I don't want you to upgrade your laptops just use this right going back to the need the need was to remove Excel and if I needed to do something of that sort forget DRI go ahead with it so this is what we made right this is 2015 so you know you can clearly see we took contracts actual contracts and we rolled it up from there so what you see there right at the top is all the different times that kind of things that you could do with them you got business break ups so you got break up of each business heads how they're doing what they're doing and where we're going into it and a very simple you basic visualization gives you the fact that the big red you know what everything else is doing it does going to make a difference a big red is a big issue there right very simple very basic a lot of numbers not enough graphs and it's it's quite rudimentary we also what we gave them and what was I think the selling point was it because of rolling it up from contract level we could actually provide to them the drill through so what was earlier going to them through two different reports started going to them in one so they could actually run through the whole scream seamlessly to see which are em has contributed or not contributed or you can start off with a customer level so that to my mind again was very very important from a perspective of selling the concept to them and I still send gave them what was going this is what was going to them earlier right and I build it on tableau and I said look take it don't worry about it you have what you have right now so at the end of 15 right we had 230 users we had one tableau server and an 8 4 server we had two hundred and thirty-odd uses you're using sequel database behind that one what it allowed me what is let me crunch my time turnaround times by two days because what used to happen earlier was that all of these things you know these multiple you know cuts in terms of these needs to go to this gentleman this need to go to that gentleman the roll ups and everything else was all done in a excel sheet so you know we had a program running which through our tech serves when you to email it out so that was a big win actually for me for in terms of just just getting the tourist conscient rundown of the turnaround time so having done that we say look it's interesting we now have the benefit of speed because we can a do something on it very quickly we can send something out to the users in a very secure manner the account profitability is a monthly thickness snapshot because you're not going to create profitability on a daily basis what used to happen is a lot of these wholesale are ends used to come back to the business intelligence the my team and say can you tell me what's gonna happen to my customer over the next 15 days so if he has a you know somebody has a meeting with Reliance and somebody's gonna come in and say can you please tell me what's gonna happen in reliance can you tell me what's gonna what is maturing what's not maturing so a lot of ad hoc data points for coming in versus what we started doing from then is saying that we're gonna go and do a digital wholesale bang so we said no more ad hoc we're gonna figure out what everybody else is asking us all the time we're going to put them into a requirement because there's no real requirement coming from the business users so we created the requirement we went through our ahead rock needs and saying okay you're gonna ask me for maturity profile you're asking me it was going to happen to my LC you gonna ask me what's happening to my time deposits for who's what showing and not and that's what we created internally rather than going out to the users the idea again is to saying I'm gonna reduce my ad hoc workload we invested in tableau so the other thing that we did is we started going to the SME segment as well so up till now we had only restricted to the wholesale the link the large corporates but we said the emerging covers and the SME which is a second segment of the whole set side of things I think this was straight after TC 16 that we ended up buying 10 times so we had an 8 core and we bought another 80 core right so we just multiplied the whole thing by 10 times my intention was at that point of time to also start rolling out to retail but in 2016 we didn't really do too much on retail that'll be in the 17 piece that we do for again guys who are doing it for the first time or in the first year of your journey get the single sign-on right one of the things that we did in this one when we were in the previous screens and 15 you had to login to the windows you had to lo
into you email but you had to separately log into tableau so we used again using the help of the tableau team with you not be shaken deeper we got the single sign-on done that reduced a lot of friction it typically would happen in a second or third year of your journey pick it up up front you'll find a lot of people using it getting to use it much faster because in AK one click you're there the interesting thing was that once we started doing a talk all of these things and we created 4 5 6 7 10 different triggers these actually move back into just two single reports so what we sent today are just three reports the report one is the API which still goes out which is a monthly bit we did two more one is the ALM which is sort of a business position report of where who's this gets refresh daily every single day we also do a majority side of things which you know it helps people let me just show you it's so much easier to do it that way I'm just gonna take so that's the daily a limb that we did again its ruling up from a contract level we have a hierarchy already set up we get feed from the Cole banking system on a daily basis we get that we leave it out and all the different products are together so you get the assets you get sub products within assets so whether it's a loan or a bill or an overdraft what's going on you also get customer level company color-coded company level things to identify whether it's coming up which customers picking it up which customer is not he also gave movement so the other thing that we did is we said ok you not only want to know where your customers are but you want to know what's moved so we gave a radio button order drill down there when we can give you the outstanding so where a customer is perfect anding or the movements which have happened so what's happened over the period of time and you know they're a bunch of filters right at the top which allows custom I know the the Supervisory layer to be so much more effective right you could drill it down from the same dashboard all the way to a single arm the other thing is the maturity profile so the previous slide - mode gave you where you're standing today this one gives you what you do if you don't do anything was going to happen right same concept you have different assets liabilities and off-balance sheet items by the way the $9.99 there is actual data inside my technical team figured out a way to mask all the numbers and hence I can show you this without the risk guys came breathing down my neck so there's actual numbers behind them but there's a way to you know take a snapshot it with all of the numbers mass to 999 so it makes it so much more interesting to actually show you guys something rather than this talk about it same color coded by company details - saying this is what's going to happen if you don't do anything this is what's going to mature over the next 30 days 90 days 120 days this this is helping the our teams today plan what to do who to meet when to meet them right we also developed a very interesting you know we could customize something having had the data there we started customizing them for a particular business set so there's a new guy who joined our company and he said look I don't I don't like the design that you guys are made why don't you make something new for me and voila the same data points we gave them both right so once you have the data set ready once you have something which is working you could always go back and ask people saying that is it working for you the design working for you because finally design is something that the tableau guys are going to do so end of 2000 of phase 2 we got 10,000 users on tableau server right oh sorry I've got 10,000 users on tableau server we got 88 course at that point and we might be upgraded to 10.1 what I started doing is and I put I think that was the other thing that I ended up doing is I got the data resting on a p7 Oracle box so it's different from the data which is manufactured it's completely in a different environment and I'm extracting everything and putting it here to ensure that the speed is not compromised right small very small snippet here but this is sort of again help the business users start doing an experiment see I did so the question and I'm going to come back to that because I'm such as centralized and regulated environment the the actual analysis not done by the guys at the ground I think at this point of time I would not have more than 25 desktop licenses right so the way we've organized as an organization is that the business guys will come to us right there's an analyst team within finance those guys will first figure out what they need to do they requirement then comes to us so we don't really at this point of time use that many desktop so my ratio of core to desktop is actually very very poor I've been actually trying to over the last year I realized that because I do governance together right I don't let data be outside of BIU it's completely with me there's no data which is floating around and if you don't have data floating around they're not going to have that many people analyzing it as well right but I have something else on alka I'll circle back on that one right and it stands 2017 and reach here right so everything that we do for holes it is so very different from that of retail right it's a complete different scale we so we did start working in 2016 a little bit on retail but we didn't really have anything which was which I could which I can claim as a victory at that point of time the different other other challenges that retail gave is the deep jabber deep Gio that I have so rural branches 50% of the rural branches we don't have great connectivity there right and whatever connectivity is there right now there's a main link and there's a backup link but both of those are used mainly for core banking so I was never gonna get bandwidth to say hey you know what pushed a blow in there so that's the other challenge that we faced on retail and I'll figure out and explain to you how did we go around managing that one right unlike in wholesale where is a single hierarchy because there's no different product role and there's no different line role is the same RM who does both in retail because you have a branch channel and that's the guys are selling the products and you have different product channels so you have somebody who's managing auto loans somebody who's managing credit cards there are different hierarchies to manage as well so retail just brings in a different challenge of scale challenge of different hierarchies and for us challenge of connectivity yeah so I can keep quiet and I can just sort of tell you guys go back to Charles Forbes and representation so this is something similar to that we we most of the product guys used to push data to the branch's performance data at that point of time to say what's happening what's not happening and we actually did this journey it took us almost eight months to do this right but if you had to build the data as well right so the way earlier it used to happen is that the producting used to take data from the business intelligence unit and then used to take that data and present it in a different manner to the branches so there was a different journey which is going to the branch and as you can imagine if different products are sending different data points to the branches the formats are different the tats are different the auto guy is sending on the fourth credit card guy he's sending on the 12th the lab guys and a girl on the 25th and if there's a review you're certainly seeing something oh this this data is of last month but this data is actually of two months old because they've not got the refresh data so that's exactly what we tried to kind of go back into and so this is something like what child swap told you is my branch or my sales or whatever they had this is what he ended up doing this is where also we found the challenge of deep geography so deep geo could not start tab though right we send them the link they opened it it just went round and round we broke our heads to saying okay what's going on and we realized it's just a bandwidth issue we solved that by enabling subscriptions so the entire branch one view today goes to deeper geographies through an email channel right in the morning before the branch starts so you use a less bandwidth hopefully not all of them have to log into tableau obviously that also means a lot of other people wanted the same thing saying I need I need access from email so this is our branch fund view these are the things that the branches have to look after we tell our central lab is a digital hnw resource credit cards and others there's a dashboard behind offer each and every one of them there's a data set behind for each and every one of them the one in red the marks whatever is marked in red is a monthly refresh whatever is marked in green is actually a daily refresh right so we actually take a lot of information on a daily basis send it across to the branches so that they have a view and this is just a clean very very high-level view of each each of the parameters of the branches have to figure out branches have both targets in terms of number of accounts that new accounts that in the count and in terms of value both of these are always this way and if you know you don't like many or one of these things you can always go back into a deeper dashboard so for each and every line there there's a deeper - moon right so you get in in that one you eventually get a 12-month trend so you know what's happening over a period of time you get both the month to date position okay doing well this month but you still have a catch up on the year plans you get resource level information so this is a for a supervisor so you're getting percentage of branches meeting plan if you're a branch manager you'll get a percentage of resources meeting plan right retail franchise and you know for those of you are running a retail franchise in under sense all about supervisory level so this is becomes very very important from the perspective right at the bottom you'll find you'll get more insights and we still give this data in a tabular manner we don't let that go right we also give them same period last year to see if there's a seasonality it's it's not something which bits gets defeated to give you a sense of how large this piece is right the red dots there represent branches so this is where it reaches you're looking at 372 different data sheets behind a 50 GB of data which is summarized on an ongoing basis 7200 emails which goes out every single day so this is a very different scale than what you would otherwise be used to in terms of some of the you know smaller banks or some of the retail products or a single company product right so we did some more stuff right and I'm just going to run you through quickly through them as well within the brand Channel right the entire retail channel we have a very profitable niche which is the manage customer about 10 million customers in all which are managed three and a half million of them are actually managed by the physical RM so there's an RM associated with him the balance 7,000 or 8,000 customers are through a virtual RMS now this is a profitable segment and we want to look them look at them a bit differently we create scorecards for them we create scars for the customer the scorecards anyway just go to the RM Zanna on the table but we created insights into the scorecards so this is the first sort of dashboard that we created from which was not existing till we got tableau so gives you a summary position there are different programs which are running you get a sense of that you get a period this how which business heads are contributing so you see business one is is at the top as I said - where is happening you get a month on month summary of what's happening which is going up what's not going up and you know we'd start drilling it down to an hour and level so this gives you a sort of sense of exception reporting you can say that interestingly what we've been telling the HR guys to say guys of about five years they contribute the maximum and you know what the data is showing that and this you can see four different geographies and it's just same you know the whole right so what you saw till now was all what the position was and all over the supervisory level and how do you get things moving much faster having seen that the branches actually came back to us and said why don't they give us something that we can act very fast and so one of the most used dashboard that we provide today is what we would call as a casa attrition casa is checking accounts and savings account so in India you have you know individuals can have either a checking account which is a current account or a savings account and this has actually become one of the most used dashboards in the in the bank right it gives them the ability to see right what happened yesterday whether money flowed out of the branches or no correct in this day and age of internet banking right in the middle of the night two o'clock you move money outside of the bank and the branch manager doesn't even aware of it by the time he gets to know it you're done in dusted so this is very very important for them we use multiple hierarchies so you see same data points we grouping them differently right and it has the ability to go down all the way to a customer so this this is this is large sets of data we are taking 44 million records processing them putting them ready by 8:30 in the morning before the branch is open wouldn't have been possible the other thing that is this is very interesting is a lot of things that we did till now the computation was done in the underlying data piece here we let tableau do the computation so we just take the data extract from the cold bank we laid out that table do the computation so we've shifted the computation they have we lived into tableau so again we did some more supervisory stuff we did a lot more on the some of the operational stuff we now have enough data in tableau to start on the analytical journey I would say we were just started it's not something that we have I don't think we've yet reached any scale in this one but we did show this to the managing director right so we've taken some snapshots from what is already being made and we build something which which we thought was confident enough to show it to the MD and you know we've been now tasked we'll do a lot for a lot more on the analytical side of things we'll come in let me just run you through the dashboard itself I think that's the easiest thing to do it's a very simple dashboard it shows which customers checking account customer also has a fixed deposit or a time deposit with us it's as simple as that but what it what we give right at the bank level what's going on what's going on to managed and unmanaged customers right at the bottom you'll get Metro urban semi-urban rural because different customers behave differently and products so people who have a current account or a savings account or a salary relationship with so in one snapshot you actually get to see everything else what most important that we gave here is the ability to change the target so you've given an input field there right it's a 35% right now and that's what you're seeing is a 35% do you have the ability to change that and the dashboard changes right so I think that made of what-if analysis very very important very critical and I think that changed the way we start looking that so we used everything so for example there's a tool tip that we started using so we use everything that tableau had to offer to us at this point of time and this this is what I would call the first step in our analytical journey we build a few more similar since then we also give in using the same things the top and the bottom performers at the branch level and you can use the same hierarchy as normal it's a cross hierarchy and that's again very important in my view you know we put roll radio buttons there so very quickly you can tap through and that dashboard changes to understand what it is because you know you're never gonna have answers to you can define a dashboard which com
ares to whatever two combinations you may think through so we have actually done this is that by doing this we quickly run through products if you have all we shifted to savings account we want to savings account in us in rural we take the snapshot we actually take the snapshot out put it out whatever else he wants and we can compare the two very very easily right so where we are in 2018 yeah sorry we had twenty six thousand five hundred plus users on the tablet server I think we have two hundred desktops now they think about two hundred desktops at this point of time we have 96 core on tableau server we are on ten point five we have multiple data sources and and by the way we've also have deployed it on Linux so the large portions of this is right now still on Windows but we have I think couple of them right now on Linux which are trying to figure out what we do and we have done daily subscription so if greater daily subscriptions I don't think what's not worked right so everything's hunky-dory - now I've been fighting with the tableau team now for almost three years our bank is on a Citrix so we use Citrix as a sort of anything which is mobile has to go to Citrix MDM and I cannot get the tableau app to work on Citrix so that's not worked I think they've just recently given me something hopefully we get to the stage to say that but I cannot use tableau and I mean for example right now I can see tallow I can see it only through an excel sheet or throw an email or it's a browser-based experience which is nowhere close to I cannot give that to my management I've not yet done enough so we do have a couple of big data platforms I haven't done enough on them and love to hear if anybody else you know offline we can chat on if you guys use big data put tableau into an Hadoop lat cholesterol or a cloudy replaces we'll love to hear that replacing powerpoints and deep pocket promised my boss that he can do his quarterly presentation to the board on tableau we have not yet reached there so that's something which you need to do and I guess the last point is no longer valid because apparently they're gonna come in 2018 one the oscillator I should be able to do that right because one of the feedbacks that I keep getting from my branch manager especially the new guys and that there is so much data available ok this is I mean suddenly they have like 28 different sheets they don't even know which to go to to get what so what I want is a very simple natural language first says okay tell me everything which I which you have for auto loans and it tells you 1 2 3 4 5 that works I hopefully that can happen as well quickly and this is my version of tech scale-out so you can have a chat with him later on right we started off with an 8 core obviously in 2015 we write as we moved into 16 right we sort of quickly scaled out and we added a lot more core that was the simplest way to do that the thing was we still made it as a van big box so we didn't really change the architecture itself we just multiplied it right and from these sort of couple of them I took them out for subscription so we sort of earmark two of them for subscription but large portions of everything was together and which meant that people coming through the gate depends on who comes in concurrency became an issue for me because whenever it's one of those things right when your results come out suddenly the website is slow the same issue here if I release the H&W score cards I have 18,000 people hitting the server the wholesale guy started crying what the hell I can't login so as we move to 18 I want to first thing I want to do I want to break this out and so I think we have already done that we him a couple of them for wholesale casters so we played a different different set up for wholesale we picked a bunch of them we mowed it out for each here right the sort of couple of them we did core and for those of you who have shown anything to your managing director keep one separately for your managing director right you don't want to sit in a room trying to him something and the whole thing is just going round and round and round you can have a very nasty conversation but yeah so this this is how we're doing right now we're actually at we still at this point of time we're at 96 core and just last week we signed up for another 84 so we'll be at 176 by the end of this year so thinking back to slides morning I'm done right we have managed to element Excel right Excel is still being used unfortunately I still still play Microsoft some license for Excel but we don't not use it for reporting anymore so that I think is it's a sort of a tick off what we started doing in 2017 it's actually changed the way we do reporting right so what used to be earlier the very used to build up data was to create a report we're now building data for tableau it's it's it's ingrained in the organization it's ingrained with everybody who's there to saying this is what we're gonna do this is how we're gonna do these things governance has increased right so we do row-level governance they do you know creates a little bit of a challenge in a different manner but governs clearly as increased duplicate datasets are deleted so I'm the only data set which exists in the organization right if the data set is not from me it doesn't work it's simple as that for the branches and for the audience in fact I'm I'm getting more and more requests in why can't we input a CRM data point as well through this so having seen the dashboard I know now what to do and I have to send it back to my supervisor explaining to him why I have not done this and I don't have that ability obviously tableau doesn't have the ability it's not a collaboration tool so I'm trying to figure out something else sometime today so 2015 the question was why tableau whenever I used to go to everybody else now genuine question to me is why not tableau so I have a lot of business guys who come to me and say why am I still getting something static on an email why can't I get it in tableau 2015 we actually presented what was already being sent to the branches right so that was the journey that we did now we're creating purely for tableau and we started doing a lot of heavy lifting in tableau today that's that's been that's that's sort of been our journey I was told to keep this slide so please do that thank you so you're between me and Shashi open to questions if anybody has absolutely so lady gives you the technical side of things that make spectrum functional side of things so tooltips right they became very important for me from a functional standpoint because it reduces the navigation right unfortunately some of the older stuff that we have done doesn't have told so although we have migrated on your server the underlying dashboard has not yet been done so that's a journey that I am taking up as we speak right some of the older see the APR thing doesn't have the tooltip so that's something which I've given my team to say hey you know what we need to restart redesigning so some of these dashboards which are used extensively we actually review every six months and if there's a new capability which is coming in we pre design the whole thing I there's no solution and I come in good feedback to them to say they need to figure out how do we fix this what you want to take data from a technical side of things we found yes it's matching with our criteria success criteria then we go with the small set of probably one or two server upgrade and see the performance if didn't face an issue then we go for the remaining one so that's the process we follow same way we do the rollback testing also so in the UAT itself we upgrade do the testing rollback the product see if it is coming back to the previous version or not without any issues and it screws about the decision to do the heavy lifting on the tableau side rather than on your relational database management system led to that decision that do you find it's actually more efficient so I think if I better think through that one whatever was on a daily basis correct so the extraction from core banking trying to compute something and then presenting it to tableau just took me longer so if I found that for example the CAHSEE attrition right that's a daily compute and the computers and change too much so that I shift her to tableau but a monthly piece I still do it underline so what I'm the the I shifted it for the stuff which I time-critical and I let tableau do the computation because it means that when you log in it computes it rather than computing for everybody so that that's sort of where in my mind I you know not everybody's gonna log in every single point of time but if I did it in the base data I'll have to do it for every record convert me so that sort of in my mind efficiency between just affected do you send to blow fuse or reports or PDFs to your regulator and how does your regulator look at that because you know sometimes I was in no regulator I can actually speak freely I'm sure RBI is not here but we don't so this is a lot more than the performance data side of things so once a year the regulator comes to our place and they wants to review our processes as well so during that point of time they see this but I do not send anything to our to the regulator from tableau we do use it for compliance compliance team actually uses is quite efficiently because one of the things that we've identified or realize the tableau is very very good at is identifying outliers so compliance team actually uses tableau quite a fish the risk and compliance team both of them use this but they don't publish too much so they they are much more than the desktop side of things they consume data among themselves I don't get too much to publish it on my site so both risk and compliance uses it quite a lot actually do you feel like subscriptions burn up a lot of power yes yes yes absolutely so subscription do take up a lot of processing go and so I'm gonna keep a couple of them separately but I'm sort of compensating for the connect lack of connectivity because trying to get increase the bandwidth into the deeper geographies is actually forget costly but much more challenging this is really nothing there so I'm using subscription to substitute for the cost and that I would have applied at eyes however I'm happy with that I'm not changing I have two boxes right now doing the subscription so I'm buying more I'm not gonna move anything there those two are enough for me so I'm buying more right now because I'm actually hoping that some buying another hundred desktop as well and I'm going to distribute the desktop and well let people play on it so I'm gonna start the journey for the analytical side of things right so that's the reason I bought the more stuff I'm a little I'm I'm curious as to your your core banking platform is it just one one centralized banking platform are you looking at Stickley together multiple systems into your reporting platform yeah I fed point and I should have actually cleared that first we we are an open standard right we have actually five different core banking platforms running in the system we have the wholesale guys separately we have an EMI separate completely an EMI product separate we have credit cards which is very very separate from a checking account so the liability side of things and a full CR and between all of these things where the data is generated so yes the heavy lifting or the BI units not one we pick data from all of these guys first get it together so when I when I did the branch one view the real challenge was getting the data right the wrapper around that was not that hard in that branch dashboard that you had how many banking systems were you just playing I think it at least eight but I have for example third-party products that I sell I don't have it in my system so I do get extracts so in all if my memory serves me right there are 28 different sort of extracts or sources which I get wow that's awesome man hey thank you for coming here and talking I was wondering do you have any limitations or friction from your licensing model or are you all for all the time I'm all for all the time that's the advantage of being their biggest customer come to India and then you can talk to them sorry guys I'm all caught all the time because I find that's that's more efficient for me it's just I see that you know I don't think I've yet to come across anybody who's kind I don't have that problem I have a question more from the user perspective so given your rolling out to the masses how often did you introduce change to your dashboard to the front line because it's easy to make on the fly changes to the look and the metrics right but the front line has to continuously continue so how often did you have releases so I think that's ingrained in what we do right in fact even it's not tableau but the scorecard undergoes changes every couple of months right you're changing you know you want to push a product so you change the matrix you say okay we're gonna introduce this and you have something regulator which is coming in so you put really I see as one of the parameters that they have to do so the front line actually is pretty used to the idea that what their KPIs will change I said that becomes easier for me to then just quickly introduce it in here it's a it's a mixture of how we manage businesses in the first place - it's a a lot of things from the front front-end side right now is what is currently going on with them it's just making life easier for them I think the next journey hopefully is that they come back and start talking to me it's not yet reach there so do you have any operational data when I say operational data like your concentrate on any of tableau yet are you planning to get that I have a lot of CRM data in it okay but not directly from a call center via the call center puts data into the CRM and we have a lot of data or CRM in this okay next question are you do you have anything an omni-channel where your customers interact you know with different channels like phone web chat text etc do you have any not yet not yet that's our sort of hopefully the next six months we should have an omni-channel experience between the customers one last question so how do you address um I heard that you have the drill-down ability to the customer level which is 44 million records so what is it that enables to get to the drill-down level without tableau clocking is it the backend that is so strongly the process of power that's so strong or is that the ability of tableau so it's a mixer and I'll let you do that but one of the things that I have done to do that is the extract so I'm actually taking an extract and keeping it completely separate from my the engine which is creating it so that's the p7 box that I did so wherever there is such a thing I keep an extract out of it but much she would be a better person to explain so I didn't - that what we resolved this slowness issue distributing the all those background processor across multiple tableau servers and that's the way we have distributed the load as we have shown that in the presentation we have kept GroupWise some course that's a way we are balancing it as a bi unit is it that y'all have got multiple systems and you all have created a data warehouse or something of that sort and tableau feeds on the data warehouse or is it that you have a data warehouse and you've got multiple tableau extracts so just on the technical level so so I do have a data warehouse you separate and sort of performance management it's a little bit slow and I'm going to look at him because he's my data warehouse guy as well data widows takes a little bit time to be up and running plus the day you know one of the challenges that are facing under the arrow side of things is you know they need the data warehouse guys need everything to be perfect before they introduce a new element the little arrows my business team is not gonna wait till then and that is where the be are you unit came into the picture in the first place right so viu is actually a complete different data set different from the warehouse right I don't know I think I have about 24 25 terabytes of data sitting there right now resource data from all the core ba
king systems support systems subsystems third party products create the database here so and tableau sits on top of my mod it's it's not yet on the Netezza so we have a warehouse we have entity IBM that he's our warehouse platform tablet doesn't sit on that right now at least as of now it doesn't say I'm Shaw's sort of I can see a lot of people nodding their heads and where else has a challenge that it does not keep pace with the business guys I've got a question around training with this kind of roll out to thousands of users clearly you've got a lot of different tiered user levels here how did you approach that and kind of what's the measure of success for those things excellent question so in fact this the co e team right part of the center the co e team actually is now very closely engaged with the training team from HR so right from an RM onboarding we have a two-hour session that we do with them right for the power users in fact tableau team does the training for them so they end up coming and doing it I don't know once two months or something they do a batch and do this one but end user training because there are so many different people there right so it's a co a job that center of excellence unit who actually goes out and trains we have connectivity we have VC so in Metro hubs install we use that up to push to the training side of things but that's a excellent question and I think I'll make a mental note of it on that one I need to take care of that Thanks how do you address the data stewardship do we have people for that role I wish no so I do have a so the business unit the be on your unit itself has a full technical team as well right so we unfortunately keep adding more and more layers in this one we redone a lot of stuff because you know you publish a scorecard and then somebody comes back you know what I didn't screw up ops team screwed up we still have to figure out a way to do those things right and and so it's an ongoing challenge it's an ongoing way of doing things finally data is data right at the end of the day so if some if if the data is wrongly captured I still need to figure out how the mechanism to do that and so even that the whole server administration server management team is part of my team itself so the very independent viu is a very independent entity within the organization it's not so the IT team I keep them at arm's and Entering the only thing that I could I really do things with them is the tableau because that eventually goes out and feeds into the same bandwidth so that's how we sort of it's been that way over the years all right thank you very much guys we're just about out of time I didn't think it'll take me an hour [Applause]
