Enterprise Data Strategy

In this exclusive video, hear from Heather Wilson as she discusses the CDO role and how she Identifies how data can be used to support AIG’s most important priorities, ensures AIG is collecting the right data, and helps make data-driven decisions.

Video transcript:

We are going to talk about Babylon and that’s Babylon in Arabic, in Jewish and in old Greek and we are going to talk about the tower of Babel because that’s was what we are talking about and architecture in the Snellen eye chart and then we are going to talk about Sisyphus, anybody know who Sisyphus is? Excellent, excellent, good. I will be talking to you and after we get finished with Sisyphus and architecture development, how many people know, who Luca Pacioli is?, nobody, you are, how many people that took an accounting class here? And you don’t know who Luca Pacioli is?, oh you better stay for this presentation. Wow, how many people of you are business people, how many people are IT. Okay, I don’t talk to business people, I only talk to IT people because business people never want to talk to the IT side and IT never wants to talk to business, but this whole issue is about communications, that’s what data is all about, and we will talk about Luca and if you don’t know Luca, you all should be spanked if you are in finance.

That’s what I deal with primarily is global finance as well as the global management offices and then finally we are going to talk about speeding this whole thing up, because you guys are taking way too long, you take, cost too much money, and the quality isn’t there. You got to do this a whole lot faster and the reason you have to do it faster; one, is the internet, and two, is by iPhone. If you think you have problems with data, with the internet, you ain’t seen nothing yet. This is going to do things that we don’t even know yet what the impact is going to be and then we are going to talk about doing this stuff concurrently because what now, what’s being done in architecture now is done sequentially, we don’t have time for sequential. It’s got to be concurrent. Parallel processing you do, parallel architectures and then we are going to talk about the Boston Marathon because that’s what I do on this side is do marathons and I did Chicago marathon last year and anybody here from Boston, I hope somebody is here from Boston, anybody got any connections to get me into the Boston Marathon, this one, next year, I don’t care, I'm not fast, I'm slow, I'm a turtle, but I still do marathons anyway because I like bigger and bigger challenges, so if anybody is doing any data management for the Mars or beyond the globe, I like bigger challenges, you know, global as far as I can go right now and I need somebody is doing it on the moon or on Mars or something of that nature.

So, me, I started my career in Northern Trust Bank, is there Northern Trust people in here?, we should, too bad, you missed it. That was in 1974 and I don’t know if the gentleman was here who was complaining about people on third shift, that’s me, third shift, night shift, 12 o’clock, 8:30 in the morning, I had two decks of 80 column cards, which is what we did in the business that day in Chicago and I had to put them into a machine with a 300 byte modem on a analog line that wasn’t clean and get that through. My key performance indicator, we even had key performance indicators back then was to get done by 3:30, why 3:30, because if I didn’t get done by 3:30, I couldn’t get to the fish store by 4 o’clock that’s the only place you could get lunch at that time in Chicago in 1974. There was one fish store you could buy shrimp, so that was my key performance indicator. It wasn’t about the company, it was about me. It’s all about me, it’s all about you.

Then I went, Northern Trust wouldn’t allow me to be in operations and then do programming because you know I might have ________ [00:04:21] a lot of money, so I moved to Continental Bank. Continental Bank at that time was one of the largest banks in the world. I immediately moved into their international banking. I did funds exchange. I did international book keeping systems. I got phone calls and I was in their ski club, and every time we would go out and have a beverage, I would go home and about 2 o’clock in the morning, I get a call, and there would be this hissing sound on the phone, and it will be someone in Europe, “The program is not zerolizing.” I had the same reaction you did, what is zerolizing and the branches are down in Frankfurt and Antwerp and London, the programs are not zerolizing, and I would have to drive for the suburbs downtown and by the time I got there, I got five calls from five different branches, because they were all down and thought I died, because this call was initiated in Belgium. They could not believe someone lived over an hour from work. You know, he must have died, you know, is he dead, you know, he didn’t make it. You got to fix this program, all the branches are down and what zerolizing was, it wasn’t taking the ledgers back to zero so they could start the next day. They just kept, adding, adding and adding, so I had to fix that, find the program and fix it immediately. So I did that for a while then I got gravitated into data and we started getting to the idea of databases and was IMS and IDMS and then relational and Continental’s plan was, we are not doing international in the databases until the last business, that’s going to be the last function will be international when you do every place else, I wasn’t married to the banking, I was married to data at that point. So I moved to the Federal Reserve and I stepped right into regulations.

Anybody recall 1980, anybody old enough to remember 1980, the Monetary Control Act of 1980, Jimmy Carter’s fundamental changes to banking, biggest change in banking since the depression, guess who had to put those reds in, guess who was responsible, co-project lead for a number of bank accounting. When I walked out of my boss’ office after he congratulated me on having the responsibility, I said to my co-project lead, how long you are going to be here? And you know he said to me, “how long you are going to be here?”, I said, “I hope less time than you”, because I knew there was no way and when you are talking about this and you talk about regs, not only do you have the problem of regulations, the regulator has problems with the regulations. Our excuse was why we were–we were delaying it, because your banks couldn’t get ready in time. We couldn’t get ready in time, not even close, not even by 6 months, why, because we had data names in programs named Dorothy and Toto and the Tin man and I swear to God I had programs with those names and they are moving Dorothy to us and, and then try to figure out names and definitions in your data governance with those names in the programs and I would like to say those were the better names. They had worse, you know what worse than Dorothy and Toto is?, worse than Dorothy and Toto is move field A to field B, move filed B to field C, run out of 26 different names, field A to field AA, field BB, swear, did.

Data dictionaries, that’s where we started looking, somewhere there is a data dictionary for this, so we had a data dictionary to have look what B, oh, there is no data dictionary. Where is the guy who wrote this, I want his head. He is gone. Why did he do it in the first place? Job security. How can you fire me because you know nobody looked at the program until after he was long gone, nobody knows what are those names. I did that for a while and got out of that, and said, I'm done with banking. So I went out and got into patient accounting systems and installed those and then I had to deal with data names again, because I'm putting insistence for regulations again, Medicare, Medicaid; 251 names on a single form, 251. No definitions. I did that for a real short time because our company guaranteed we would meet all government requirements, guaranteed we meet all, we couldn’t do it. It was impossible. We had CIOs got fired because past the vendor couldn’t do it, I couldn’t have lived with that. I got off. I went and got back into options, Chicago Board Options Exchange and that was a wonderful time to be there in 84 because we were growing about 40% a year and money was no object. Well, almost no object, except if you were in the data space because they did know what that was and they kept ________ [00:10:02] get replaced to somebody else that whatever, but we did the first Teradata production system ________ [00:10:09] Teradata, first one in the country, in the world, CBOE, market surveillance system, ever heard of Ivan Boesky? Ivan Boesky trader, market surveillance system was the first one that picked up he was doing something shady. Teradata, parallel processing, big concept back then.

I did the first data dictionary, we didn’t have metadata repository because nobody knew what that was, it didn’t was in dictionary back then. They wouldn’t pay for it. What’s data management, we don’t care what, data management, we don’t care about that. So I wrote it in Datacom Ideal Fourth Generation Language and at that time there were a variety of different vendors around, some of them still around, some of them not around. They are Brownstone, Realtek, they got swallowed by Platinum, Platinum got swallowed by CA, part of things you talk about data lineage and process lineage. Some of these things you are going to be in the space, you need to know lineage of the products because when they talk about these things being integrated, are they really and how much. From CBOE, I did that until the crash and the way we measured how much volume we had in 87 at the crash, was how many carbons were on the floor. Black Friday, Black Monday, Black Tuesday, it was pretty black, ankle deep in paper. People walked in there with Rolex on their arms and Ferraris in the parking lot, they walked out, no watch on their arm, that was in the pawn shop, keys for the Ferrari were gone too. Boom, not a good time to be in there because we weren’t spending a whole lot of money, lot of people think it was the New York Stock Exchange that was risking failure in 87, it was not. It was the CBOE. Banks, Continental, Federal Reserve had to keep them from going under because this thread on the margins were so great, the traders wouldn’t do business period, even though they were required by regulations to do so. They simply walked out, went to lunch at pizza down the street, bought Dominos, pretty good pizza, it’s still there. Most of the traders have gone.

From their, I after 87-88, they weren’t doing data management too well, so I went to this company called Franklin Savings, anybody ever hear of those guys? One of the first guys into this product called mortgage-backed securities. So I had to do all the data modeling, data architecture on mortgage backs, prepayment, steep fall, etc., etc., etc., huge books this thick I figured if I can’t understand it how was the schmuck on the street going to figure this stuff out and we were about the third company into it. We were the third savings in loan that came under the resolution, TrustCrop, and the first one put out of business, 114-year-old firm. After that, I had a stint in another company, little company called Craft and we had, do you want to talk about data names, we had a little problem with sales, what is a sale. We had five different definitions of what a sale was and we were rolling them altogether and putting them on the annual report, and my boss asked me what does that mean, and I had one word for it, I didn’t need a data name for that or definition, I said prison. It needs prison. I said it’s fraud, what we have down there has a definition of sales, our total sales $6 billion division of Craft was a wild ass guess, that’s it, that was just as close as it gets because we had some regions putting in shipping charges. We had other ones putting in just the cost of the product. We had other ones putting in the sales commissions, the other ones putting in taxes, other ones putting in all of it in. How did–why did they do it because whatever motivated them the most was what they did, you know, but if they looked better by putting in the smaller number, they put in a smaller number. If they looked better by putting a larger number, they put in a larger number.

After Craft, I went to CNA insurance. I stepped in the regulations again. You have all heard of this one. I think you are familiar with it, it is called Sachs, Sachs, Sachs is the greatest thing this ever happened to data management because the CEO and CIO didn’t want horizontal stripes or vertical stripes going horizontal, you know, the motivation. Sachs’ has lost its motivation and I will tell, it is in Zachman column 6 and now you don’t know what I'm talking about because I have not defined Zachman column 6, but I will. They lost their motivation, I forgot where I was, I forgot where I am. CNA and Sachs, CNA, sorry about that. They weren’t, you know, they aren’t, keen in data management too much either and I did that for a number of years and decided, well I'm better suited for outside of the companies rather than inside the companies, so I went into consulting for Butler Technology as head of enterprise architecture, enterprise intelligence, then the Fujitsu which had bought Amdahl, I originally had been at Amdahl then the IBM and today I'm in HSBC, and what I do is strategy and governance for architecture for the data and information space that I have to bless everything from global finance and the smaller blessing for us the global management offices and if you don’t know HSBC, we have 2000 business units, we are in 80 different countries, we have five different regions and I could tell you exactly what we do today, but it would have changed about five minutes ago because with 2000 companies in 80 different countries, something changed somewhere, and if you are in a business of that same size, guess what, something changed again and all we are dealing with here is the only change.

I'm overeducated, I have a management degree in business administration, masters in business administration, masters in project management, telecommunications management and ISM. If you come to me with your data governance program or your data quality program and you want to talk about project management, if you do not have a management contingency, a management reserve, or what it’s going to take that you don’t know about, money for what you don’t know about and if it is not at least 25%, go away. You know what you are talking about. You are not prepared. Go back, come back, revisit the program, come back to me again when you know what you are talking about, minimum 25% of your, if you are trying to do this globally, you better be talking a 100 to 200% management reserve for what you don’t know, because if you can predict the future without accuracy, that you don’t have to have a management reserve, please give me the lottery ticket numbers for tonight. I'm not worried about what you are going to do for a year from now, tell me what the lottery numbers are tonight. If you can’t do that, you need a management reserve and contingency on your program for your data governance. Okay, I'm also TOGAF Certified Business Professional, blah, blah, blah, and if you don’t know about the DAMA Management Association, we started the chapter in Chicago and third one in the country, it is international, you need to know about the data management body of knowledge, a new release comes out at the end of this month at the enterprise data world conference on the 30th that will tell you all the functions that are in the data management space if you don’t know them already. There is a new one this year, data integration. There were 10 before, there are now 11, and that’s a draft, so it is subject to change and most of that activity is oriented out of Chicago because we are the ones that started. I was the original editor of the standards prior to the data management body of knowledge, but I couldn’t take doing standards on a Saturday five hours to get five lines of a standard done and it drove me crazy as you could see I'm, you know, hyper. I do a lot and I do it fast, and lastly I teach a class in Dubai, and that’s a subject for another time, on strategic business intelligence and the perspectives of Dubai compared to here, in regards to business intelligence is day and night difference.

We tend to be very internally focused, cutting cost, cutting cost. They are not worried about costs, part of the reason is that oil money. The other side of the coin is they are looking at the competition. They are looking at competitive intelligence, they are looking at markets, they are looking at industries and they are coming after ours, and to give you a perfect example while, one person was talking about the airlines in America and then this merger and that merger whatever and I work for United and did the models for them, they are coming after them. They got Emirates airline, they have Etihad Airline, they have Qatar Airlines. Qatar is now flying directly to Chicago. I don’t fly United to Dubai. I fly Etihad, better food, brand new planes, state-of-the-art video Microsoft, songs, movies, it’s great. You got to go to Dubai, there are no contests and they are going after our businesses. You think that these guys are sitting around on camels in the sand, forget about it. We are going to look at the, NTA here, go look at the Dubai. This thing looks like Stone Age, the Chicago one looks like Stone Age compared to Dubai. Enough about me.

We live in a complex world. It’s a big world. What I–and I have got, oh, boy, I'm way behind and I have to go really fast. Is, you are looking at data, data doesn’t come out of nowhere. You got to look at the business and you are business people, you should be looking at the business. It is a business issue, it’s not an IT issue. This is the current state of business. This is the tower of Babel in Babylon. We are talking in tongues, we are talking in multiple language. You say customer to somebody, everybody has got a different perspective. What we need to have is precision in our language in communication, so if I say a word, you know exactly what I mean. You don’t know exactly what I mean and the context of that word, you need to be able to look it up in either a data dictionary or metadata repository has that information and guess what, the people out there, the data governance think you are going to come down to one day and one definition, [Laughter]. Forget about it. There is no such thing. If you are a large company, it’s got to be federated, 80-20 rule, 80% you can standardize, 20% is going to be local, can’t do one. You are never going to get, how you are going to get, everybody in the company of a quarter of a million people which is what we have to agree on a definition of customer. Is there a different definition, take a dictionary, take the Webster’s dictionary, look up customer, there is more than one definition. Look up product, there is more than one definition. You think you are going to get down to one, forget about it. Also you are going to need is the source. This is all about language. It’s all about communication and the speed of the same.

Here is Hermann Snellen’s eye chart, how many people have seen an eye chart? Everybody. This thing has been around for a 100 years. A 100 years, this guy invented it, everybody has got an eye chart, everybody knows what this is. I will pick the top level 1, is where your exacts are. They mentioned customer, down here is your DBA or your data people, they are talking about customer. It’s the same customer, it’s a different level of granularity. You need to know them all and here, of course, you can’t read it, is 20/20, that is 20/20, you know, 20/20, do you need glasses, can you read that well. Perhaps, you need an eye exam. That’s what you need, you need an eye exam, get one right here. Everybody cover your right eye, remember when you went to the eye doctor, I'm the optometrist, cover your right eye. Can you see as well with your right, cover your left eye. Can you read down here this as 20/20 and you in the back, can you read down here, why not, maybe you need glasses, maybe one eye is better than the other, maybe you are looking at data and you need to look at something else and we will talk about architectures too. You can see that big tall building, can’t you? You know what architecture is don’t you, everybody thinks architecture is about tall buildings. Why do they do a lot of research on architecture and it is not about tall buildings. It’s actually about caves. Do you know where architecture came from? You know why we have architecture?

We have architecture because of people, because of utility, we needed to communicate across spaces. If you could see through this, there is a building out there that has more than one architecture, it’s not just the data architecture, you couldn’t build that building with just any data architecture, you have a plumbing architecture, an electrical architecture, heating and ventilation, cement, steel, etc., etc., and they all have to be integrated. You wouldn’t drive down a street without a car that doesn’t have multiple architectures. Why do you think business should be any different? It’s not. The idea of architecture is to provide a collective understanding so that all the parties from different spaces involved can come together and get something built, something designed, and if you do only data, you are at the bottom of the food chain. You got to answer 6 questions?, simple, 6 questions. Every language has those 6 questions, without exception, Greek, Hebrew, Old Greek, New Greek, Chinese, you have to answer whenever you are doing those six questions. What’s the most important one? Why, why are you doing it, what’s the motivation. Zachman column 6. Wow, anybody know who Zachman is? Nobody knows who Zachman, oh 10 minutes, okay. I got to speed up.

John Zachman invented enterprise architecture starting in 1987. In 1985 at the Chicago Board Options Exchange, he came in from IBM selling us business systems, planning with predated enterprise architecture. What I'm proposing that you talk about concepts, context, get into some antiques and some of these things, you go down this food level of chain, it is going to have all the things that you are talking about. What I'm saying is you need to start at the top and work your way down just like on the eye chart and why do you want to do that, because there is another view of architecture, you want to start with the business architecture. You want to go down to your information architecture and information architecture is not the same as the data architecture, not the same. A lot of people here are talking about FIMA, financial information management, but then you talk about reference data architecture. Two different spaces. This is the raw material going into your systems that’s the finished goods coming out. IT guys are using TOGAF as a methodology, anybody. TOGAF does not have an information architecture. I think it is deficient. You don’t define your output, what is the data good for. You are going to do reference data, who is using it, what are they using it for. If you have to change it, how are you going to know the impact. How are you going to do the impact analysis. You need both.

This is Sisyphus. You know who Sisyphus is? You were saying that you knew Sisyphus was? Sisyphus is pretty much 80% of the people in this room. You are starting at the bottom and you are working your way up and you will get so far in data management or data governance and the weights on your shoulders will push you back down again and you will resign to pursue other opportunities as many CEOs do, as our first one, when I started at HSBC two years ago resigned to pursue other opportunities after a year of doing, zip. Now we have a lot of CEOs, we don’t have one, we have many. I'm speaking here in lieu of one of our CEOs. There is nobody here from HSBC by the way is there? Good, good, good. So this is how typically most people do data governance, we are going to get one data name and definition and we are going to get it all done with, that’s a perfect world. We don’t live in a perfect world. You can’t start down there. The reason we start down there particularly in IT because this is what they know and one of the guys said you have to do data management and you not only have to do data management, you need to do information management, you are going to need to do knowledge management and you are going to need as Mike said, we are going to need to do, Mark, I'm sorry Mark. Who is the management and if you don’t know what those are, we could talk about it over a drink in less than 10 minutes. You really need a business architecture, because otherwise you are flying with one eye. Okay, you need both eyes, both, you don’t wear a monocular vision, you don’t wear eyeglass, you got glasses with both eyes, yeah, we need two views. This is what typical value chains are. Value Chains, Capability Models, Functional Decompositions, this is a traditional business architecture. I'm not going to go into strategic business architecture or anything else, I don’t have time. We could talk about that over a drink. Process Flows, Workflow Breakdowns, Activity Models.

What I'm proposing is you start at the top instead of being at the bottom. It’s a lot faster, what would you need is the top execs to pull this off. I have only had this once in 30 years work and that was at Franklin Savings. They said vice-chairman and chairman, this is what we are going to do. Anybody got in our way throw them out of the bus. CFO didn’t get along, wouldn’t agree with, couldn’t figure out why. So we circled him, figured out everything that went in, everything that went out and found out what he had done, why he didn’t want it. He had completely automated his entire effort. Everything was in Lotus 123. You got floppy disk ________ [00:31:47] that’s shoved him in there, read Fortune Business Week, Wall Street Journal at his desk everyday completely had automated his job. They fired him. They fired him, why didn’t you promote him. Maybe he could have automated everybody’s job. Start at the top, it’s a whole lot easier if you could get sea-level understanding. If you have to do the next level down, do the next level down, understand the information, what you are delivering before you start dealing with the raw materials.

Okay. Data is the raw material, information is the finished goods, finished good is consumed by what the business. I have got to go faster, it’s a marathon. No, I go slower in a marathon. It’s not a sprint. You really need a business architecture and an information architecture, okay.

This is how you put the two together. Semantic models directly will correlate to your value chains. Core processes in the airlines, customer shows up on core processes and the value chain 6 times out of 12, probably when you want to do 6 times out of 12, but I recommend that you do not start master data management with customer, even though it’s probably the one with most value, because you have to do ________ [00:33:15] assessments on everything you do in the data space. Where are you in the space, 0 to 5. If you are starting out and you think you are going to do master in data management with customer, have a nice day. Let me know when you, when you are replaced or you talking to the employment agency, when you resign to pursue other opportunities, because you will never get to the top, if you have never done it before. If you are going to start, start at something where you control all the data, you control all the information, you control all the changes, that’s with locations. Then move to product, why product next, because once you get a customer done, what are you going to relate it to. You got to relate it to the products. You control the product information completely so do locations, do products, then do customers.

Conceptual models, ontologies, taxonomies can directly relate across these, directly relate. So when you need to change those, you can change them based on the business. If the business changes, then you can change the business activities and relate it to the other side. We got to get to Luca Pacioli. Luca taught mathematics to Leonardo da Vinci. Perspective, you have seen the last supper, Mona Lisa, the perspectives, dimensions, depth, you had no depth if you only have one eye. Depth has to have two eyes, two views. He developed double book, double entry book keeping, ________ [00:34:53]. This can operate in the same manner, business side – verbs, data side – nouns. In architecture they say form follows function, function, form. Form follows function. This is not rocket science. This is double entry accounting system. This is basic accounting 101. You missed that class, you would now have a drink instead. That’s Luca. That’s why, that’s why you want to look at it, its basic accounting. Business information, data side.

What happens if you only have the information side? When you want to go somewhere, you can’t without understanding the business side. What is it being used for, how are you going to deal with data reference if you don’t know where it’s going, what the regs are going do to you, okay. A bicycle is not particularly good to get you anywhere if you only have one wheel. Architecture is the same way. It is a two-sided coin. You need both sides to have value, okay. You need the business side, you need the information side. They go together. Double entry accounting, if one side changes, you got to see what’s on the other side. If the other side changes, you check the other side, what happened. You got nouns showing up on business architecture, what’s changed. If you have something change on the business architecture, it should show up on your information side. All of this relates all the way down the eye chart. You start at the top, you roll it down, the names will add, customer here, guess what it is telling, potential customer. Adjectives add, next level down you will have two adjectives. You continue down. It’s language. It’s communications. Both sides. Customer on one side what the business is doing, if that changes, you are going to have to change the information on the other side and vice versa. What that drives the information architecture and it changes. I don’t deal with the construction. I don’t care all the time, okay. It’s my last slide.

You need business architecture not just your information architecture. If you want to be a winner, if you want to be successful in doing this, the sustaining time. I don’t care what you do in construction, if you want this the last, perhaps you dealt with data warehouse before. How many times did you data warehouse before you got it right. Do you want to do this same thing with data management and data governance? I don’t think so because you will be resigning to pursue other opportunities. Anybody get me in Boston Marathon, this is a little prompt here, you know, you can be a winner with the architecture, but I need some connections, so somebody got to be sponsoring something at the Boston Marathon, so that’s it. I hope it was helpful.