 Hello and welcome. My name is Shannon Kemp and I'm the Chief Digital Officer of Data Diversity. We'd like to thank you for joining the most recent webinar in the Data Versus Monthly series, Elevating Enterprise Data Literacy with Dr. Wendy Lynch. This series is held the first Thursday of every month and today, Wendy will discuss taking off the blinders, how experience limits our perspective on literacy. Just a couple of points to get us started. Thank you so much for joining us today. I'm so excited due to the large number of people that attend these sessions, you will be muted during the webinar. If you'd like to chat with us or with each other, we certainly encourage you to do so. And just to know Zoom defaults the chat to send you just the panelists, but you may absolutely switch that to network with everyone. For questions, we will be collecting them by the Q&A section. And to find the chat and the Q&A panels, you can click those icons in the bottom of the screen to activate those features. Thank you so much for joining the session and any additional information requested throughout the webinar. Now, let me introduce to you the speaker for our series, Dr. Wendy Lynch. For over 35 years, Wendy has converted complex analytics to business value as a sense maker and analytic translator. A talented researcher and consultant to numerous Fortune 100 companies, startups and healthcare giants. Her own work is focused on application of big data solutions in health and human capital management. As an author of books on effective communication and analytics, Dr. Lynch has pioneered the only structure system to empower a new generation of professionals who will revolutionize the successful application of data to solve business challenges. These trained and analytic translators allow companies to convert advanced analytics into actionable solutions, building a sustainable alliance between analytic and business professionals. And with that, I'll get the floor to Wendy to start the presentation. Hello and welcome. Thank you so much. Nice to be here, Shannon. Nice to join all of you. And so good morning to the West Coast and good afternoon to the East Coast. Welcome to everyone. Welcome back to those of you who have joined us before today is going to be one of my favorite kinds of talks because it's about putting a lot of different things together. I like to think about this topic as a way that we can broaden our perspective. And the way that we're going to start here is to think about what we're going to cover in the big picture. And what that is is seeing connections between things that we don't often connect. It's about busting through silos that still exist and continue to hamper the way that we do things in most businesses. And also a little bit on understanding how to make data matter more to a variety of different stakeholders. So let's get started. I want to remind us that last time we talked about how data literacy is only one direction. So too often when we think about data literacy, we're thinking about going toward literacy because if everyone becomes data literate, we will achieve business success. And there seems to be this feeling that if we just get everyone there, all will be well. And it may take us a little while when people start to understand specific reports, then they start to really gain insights. And then finally they reach a level of understanding that is in one specific direction. And last time we talked about how data folks need to also start to be more literate in business. And so it's a two-way street. We also talked about how all of this is quite interactive and that all of these data issues, whether it be the folks who need to be more literate or whether it's data people who need to understand what's happening in the business, it's never just one direction. It's trying to use data that's dynamic and changing to get us insights in all different directions. And so because it is moving us in an ongoing and evolving way, we can't just stand still and say, oh my gosh, gosh, we have to get everyone trained in X, Y or Z because everything changes essentially all the time. And what I find in almost every project that I do in any organization where I'm asking them to try and think bigger, we find that all of us like to stay in our own lane. We all like to be comfortable in our own pocket. So we love to tell other people that they need to be data literate, but we're less inclined to say, gosh, we have to learn all of the business language and the business priorities, which they live in. We want them to come and understand where we live. So today is about how to help other people care about different topics, specifically different data. So the way that I think about it is that we want people to care. We're not going to care if we simply say, all right, this is how you measure central tendency. And this is the difference between a median and a mode and an average. They don't really want to learn that way. People care if they see something that they didn't know, if they see something that changes their mind. If they see something that makes them feel smarter and have a eureka moment. If they see that things are dynamic and changing so that the decisions they made yesterday might not be the right decision tomorrow. People care if they find one of these things that's new and surprising and making them feel smarter and also if they can understand it and apply it. If we're thinking about what we're going to help somebody understand. It's just as important to figure out how to entice them with it how to get them involved in it. As it is to put together a curriculum of these things that we just really want them to know. So in my day job, I spend time with companies, trying to get them to think bigger about their data. And as many of you know, because you're in this field, silos are still the norm. And so when I go into a company, the data sets look a little like this. They've got people data. They've got medical and pharmacy data from their health plans they've got absence sort of disability and medical leave data. They have compensation data they have performance and productivity data. They have business outcomes and they have safety and workers comp data. All of these are in separate little silos. Maybe they've put a few of them together like people and rewards or maybe they've put a couple things together like productivity and business. But for the most part, big companies still have some of these pieces separate in a different place. Now, this partly happens because they've all evolved over time, and each department is separate and they've purchased different types of tools. But there's also other things that get in the way. So for example, if I'm working with an organization and I work with their benefits people who look every day at health care costs and pharmacy costs. They want their own type of dashboard. They want to look at health care utilization they want to look at how much they're spending they want to look at the trends over time check against benchmarks, and they want to really pay attention to what's happening in their lane. They're staying in their area. And if you talk to them about why they might want to integrate with other departments, they will instantly say well we can't share this it's personal health information and there's laws against that and we can't do that. And so they are comfortable staying where they are. So then down the hall from them. There are a group of people who are focused on how they compensate people. And what they spend their time on, they want in their own dashboard. And they want to look at salaries compare them to their competitors they want to see whether the right people are being promoted whether there's equity, they want to look at bonuses and what those earnings will be. They want to tally how many benefits people get so that they understand whether or not their total compensation package is equitable and fair and competitive against their competitive companies. And when you ask them whether they would think about integrating their information. They say well no we can't share that the leaders don't want other people or anybody to know how much they make. No, you know, we're fine. We're absolutely fine. So then you might look at another department that's in charge of risk and liability, and they're looking at safety and workers come. They want to look at what the risk level is because they're managing not only the risk and the health of people who might get injured. But they're trying to look at what their liability might be what kinds of injuries and safety things are happening. And so they have their own dashboards. And when you ask them, well, wouldn't it be great if that was all integrated, and they will say no we can't do that some of the sites look bad and they don't want their colleagues to know that. So even though the company owns the data. You still have different silos, making arguments for why their department needs to stay isolated, and they don't really want to share. And the leaders may not understand why they might want to put it together. Because what are we look determines exactly what we see. So I've run into this for decades now. I think the first time I tried to convince a company to integrate their data was 1989 which tells you a lot about why I have gray hair. But I've also learned how to start to get people interested. So what motivates a person what motivates a company what motivates leaders to look at things differently. It's not because we inform them necessarily. But a lot of times it's because we surprise them. So what I'm going to do today is I'm going to go through a version of a presentation that I did a couple of years ago that was intended to help a group understand why that integration might be necessary. I want to give a shout out to a client of mine, because they do this data integration and they're allowing me to use their examples. But what I want to do is set the stage that I gave a presentation to the health and benefits groups within a large health organization. They were siloed. And I wanted to have them think differently about what it is that might be required. If they took a chance and did things differently. And as you listen to this also think about literacy, because each group is literate in its own area. So it's not just are you familiar with data. It's are you familiar with data in these different areas. And so I started out very simply like this. So what we know in healthcare is that there is a straightforward relationship between health factors and how much we spend on medical costs. The more conditions a person has the more risk factors like smoking or obesity or lack of exercise. The more on average they're going to need treatments or procedures. We also know that it's standard to actually correct for certain things that we know are important like demographics or location. So risk adjustment is the typical way that we control for demographics and where the people live. So our hypothesis that we all believe in is that when you control for age and gender and location because a heart procedure costs some costs a lot different in New York than it does in rural Arkansas, for example, that better health is going to lead to lower costs. So that's how we look at the world. And let me give you a few examples of that. So for example if you look at costs over a two year period in the same population, you will see that people with chronic disease are going to have more expensive months because they need more care. People with moderate risk are going to be in general more expensive than people at lower risk. And so as people incur different kinds of risks and illnesses, they are going to cost more over time. We also know that we can do interventions. For example, if we looked at sick absence, so sick leave kinds of days, we worked with one group where they had a lot of absences in a call center in that one department. And over a second year, they went down by two thirds in absences after we implemented flu shots. So you can do an intervention that will improve health, and that improves outcomes not just costs but also absences. We also have great examples of what happens when you do wellness kinds of interventions. So this company I actually worked with starting in 2005. And that was when they went all in on exercise and healthy foods in the cafeteria and smoking cessation and stress reduction. And their leadership went to conference after conference in 2006 and early 2007, showing how their health improvement had kept healthcare costs almost completely flat for a two year period and they were thrilled to talk about their success. However, something happened, something really happened. And if you think of this as health improvement, did every single people get a terminal disease in the third quarter of 2007? Did they all start smoking and stop exercising and sit on the couch through 2008? If they were going to credit health improvement for why things got better, then why if their health factors didn't change, did their costs go up so dramatically? If we believe in our hypothesis, this should not happen. So we have to start to wonder if we are looking at this whole phenomenon in the right way. So today, Mr. and Mrs. Health Leaders in this group of 100, we're going to understand health and business outcomes more holistically rather than isolating health on its own. So I want to start by broadening the way we think about your workers, because every day employees make hundreds of choices that affect your business. And yes, some of those choices are choices like this. Will they make a healthy choice before they come to work and choose the door on the right? Now clearly, this company believes that well-being has a big impact. So that is one choice that they will make. But they will also make choices like this one after they go to an external meeting. Which door are they going to go into? The door on the left or the door on the right? That's another choice on whether they are going to work hard that day. And so all of these choices start to add up. So one choice can be do I show up at work? That has to do with absenteeism. One choice is do I give my full effort, which is productivity, and whether I go to the bar or back to the office. One is do I work hard to acquire new skills? Do I become even better at what I do? One choice is whether I'll make a healthy choice at lunch or at dinner. And I'll also make a choice on whether I'll even stay at this job or whether I will look for a different position. When we isolate health, we pretend that all of these things are not connected when actually they are. All of these things add up to create all performance, all costs, all absence, all turnover. These things come together. So I like to laugh at this. My job is giving me migraines, high blood pressure, chest pains and bleeding ulcers. I quit, but I like their health plan. There is a connection between how people feel about their work and how they feel physically. So if your focus is on health and health care costs or absence due to illness, then we have to think about this in a bigger way. We also have to realize that sometimes work is a place that we don't want to be. And there are reasons why people may want to escape. So if we start thinking about it this way, we realize that more is going on than simply this belief that health improvement was the sole reason why people didn't have more health care costs, didn't have more absences, didn't have more disability. Because what happened actually here, and those of you who recall what happened in the middle of 2007, there was an economic downturn. And that economic downturn, that's not good. Sorry. Let's hope that doesn't continue to happen. That economic downturn led to a hiring freeze. And that hiring freeze meant that people were afraid for their jobs. And as sales slowed, people thought, well, maybe I am not going to be able to stay in this position and stay employed. And what happens when people are afraid is they start making sure that they get every procedure and test done before they lose their benefits. So, I want to admit that those earlier graphs that I showed you were a lie. What I mean by that is, this graph that I showed you where I said that this top line was chronic disease, yellow was high risk and blue was low risk. Actually, this graph had nothing to do with health other than how much they cost. What it shows is those people who are getting ready to quit spend a lot more. Those people who are newly hired spend a lot less. And this is controlling for their age, controlling for their job, controlling for their health. What people use healthcare has a lot to do with their circumstances outside of the actual health arena. This example where I said that in the department, flu shots helped absences go down. It had nothing to do with health. What instead they did was they implemented performance bonuses. So that the more that you got done, the more you earned. The incentive there was to not be out from work, saving two thirds the number of absences, and they came out net way ahead. And so, when we think about human capital, how people were feeling about their job influenced, whether or not they spent health dollars, how they got paid influenced absence. The potential economic downturn and the threat about the company influenced combined health absence and disability costs. So the truth is that health care costs are not simply about health. They're also about how they feel about the company, how much stress they're under the demands of their job. How they're rewarded, how benefits are designed and many, many other circumstances. So you can't look at health in isolation. So if we think about it this way, it means that we're missing a lot. And it reminds me of a story, true story. During World War Two, a team of weather forecasters offered their resignations. Now they're scientists, and they looked at their track record, and it was proven that their predictions were no more accurate than random chance. And they were demoralized, they felt terrible, so they offered their own resignations. And they got an answer back. The general is well aware that the forecasts are no good. However, he needs them for planning purposes. How much are we using reports that we know are incomplete? Because that's what we're used to. How much are we looking at health care only based on health, looking at absence, only based on absence, looking at injuries only based on safety data. Rather than all of these things put together. It tells us again that where we look is going to determine what we see. And it's almost malpractice to ignore the things that really matter. So when we look at health this way, where we only include people data, because we're going to control for age and gender and location. That really ignores things that we know are also factors. So, instead, we need to include a whole variety of inputs to understand what is actually happening. We need to understand how long they've been there, how they're paid, what job they do. And how healthy they are, but also the community that they live in, whether they like their jobs, the attitudes and what type of industry they're in. And when we know those things, we can also start to understand some of these outcomes like paid time off, job performance, disability, how customers feel about what gets delivered, injuries, mental health. These things need to be together to understand them in total. So, when we think about an organization where all of these different supposed silos that operate individually. All of them really ought to be like this. We need to take all of them and put them together in a structure where we know how they go together. It's only when we have them together and have them together not just once to confirm the question, but so that on an ongoing basis, we understand how they all interplay with each other. So, we are now in a situation where the status quo doesn't work anymore. In order to have business intelligence at the level that you need to, to manage everything. Integration is mission critical. So I'm going to give you some examples of what happens when you integrate without integration. First of all, we're missing opportunities. We're making mistakes. We're losing money by solving the wrong problem. And so we have to think about how they are connected. Here's a really great example in an organization where they put people data and job data into the same integrated database as safety and accident and liability data for drivers in their organization. And what they found was there was a tipping point after five or 10 hours of overtime where between 10 hours and 20 hours there was a tripling of the likelihood of a safety event of an accident by their drivers. So they needed to not look at whether that driver was skilled, not to see whether they need more safety training, but to put some guidelines around how much overtime any individual driver could spend. They looked at turnover and turnover they often think of as something that happens when people aren't performing very well or they don't like their boss, but it is a really intricate interesting combination of factors about the type of job they're doing. There are different kinds of shifts there on how they've done in the past six to 12 months, whether they've been absent a lot, whether they're under a lot of stress, whether high performers have had a pay increase within a given amount of time. And there were others like what the dynamics are within and or within their department. But when you put together a model like this, we could predict turnover on a range from some employees who had a point 00001 chance of leaving to somebody who had a 94% chance of leaving. If turnover was an issue, which in this organization it is, they could spend time looking at specific departments or specific individuals to understand how they might be able to turn that around. We also looked on behalf of an organization who provided mental health care and handoffs from other types of programs. So for example, if somebody needed to go out on family medical leave, did they have a risk of needing care, mental health care or EAP services? And so we looked to see how well can we predict who might need that based on different data sets and incrementally we started with medical claims. And so if we think of 1000 people who were who needed mental health care over a certain period of time. Simply having their past medical claims and pharmacy data, they could only identify 12 and five of those were incorrect. They added a health risk assessment where people answer questions about stress and job satisfaction and it improved, but they still only correct two thirds of the time. They added changes in their reported stress and work satisfaction so that they could see over time whether they used to be have low stress and now they have high stress, and they got even better. They knew every single one of these incrementally, but when you put it all together, they were able to find a significant portion of the people who would need mental health support at an 80% accuracy. When you combine information, you understand what's happening much better. Just as they kept on going, they improved their accuracy to 90% so that they could identify 35 new potential cases who might need support and 30 of those were accurate. If we know the combination of outcomes we also can assess the impact of different types of interventions, much better than if we're only looking through one lens. This was a large national retailer. And they were trying to handle their sort of disability experience. And they said, what would be the benefit of reducing average disability duration by one day. So the average was about 36 days. Could they reduce that by one day. What they found was that the cost of disability itself for 100 employees that saved you $8,000 that's not really that big a deal. But because they weren't on disability anymore they sought less healthcare so that actually improved by $24,000. But it turned out that when somebody was missing on a team, it had an impact on their sales revenue. And so one day of decreased disability on average could save $122,000 for every 100 employees. And the largest portion of that 74% was because they could avoid this loss of sales revenue. Again, they can make a better decision than if they only were looking at one outcome. As we become smarter, as we get broader in the way we look at things, the more we can make effective decisions that are fully informed. So for the nerds in the group, I also want to share that we all know that using machine learning models lets us explain a lot more variability. And what that means is how well we predict things. And simple linear models that we all used in we were learning modeling will explain less and it is limited by certain types of relationships. So what we did was we compared using a simple model versus the fanciest machine learning models as we incrementally allowed the computer to consider more and more sources of data. And the outcome that we are predicting is how much lost time a person is going to have. That's their sick leave and disability and workers comp and all of the leaves and everything else. So the total lost time a person would have. And when we only include demographics as you might expect, we can barely predict anything. But as we add medical data, we see that yes machine learning is much better at predicting with just medical data. We are getting a little bit better and the linear models aren't very good. And then we add pharmacy data we go up a little bit. We add health management data which is programs that people have been enrolled in. And then we add what people self report about their risks that is helpful, their short term disability, their medical leave, then whether they've been injured, whether their total costs combined have a certain pattern. Their work data, what they do as a job, what kind of shifts there on then the communities that they live in. And then how they perform. So what we see is yes of course, those fancy models help us a lot and we hear people oh we use AI to do our prediction so we're great at everything. But when you think about it. If you don't have integrated data, you are not going to do as well as simple models that uses all the data. So if you are only using medical and pharmacy and demographics to predict who's going to have lost time. If you were to do worse even with the fanciest models, then a simple model that uses all of these. So, this is beyond what most people need to know, because it is really nerdy. The point is that from the standpoint of new discoveries from the standpoint of increased accuracy of decisions from the standpoint of prediction, whether that be of lost time, or turnover, or needing care. If you aren't putting all of your sources together, you are leaving value on the table. Our biggest challenge is the time zone difference in New York it's 245 but at our headquarters it's 1974. We all need to move forward. We need to make use of what we have. We have to think about these different areas collectively. And if you're in healthcare and you don't know anything about compensation or safety, it's time that there's an increase in literacy there. Those in safety need to understand some of these other areas. So if we want to take off the blinders. Whether we're talking about literacy or business intelligence, or any other way that we manage data. It's imperative that we begin to broaden our perspective. It's also imperative that we make a compelling case that creates urgency and buy in. After that presentation, there was tremendous increase in willingness for different groups to start to work together. And again, my thanks to work partners who allowed me to use these examples, because they are in the business of doing the integration. And unlike most organizations, they have access to it. So, again, they are kind of mine and I'm grateful. So I will finish up by saying that if you want to learn analytic translation skills, which include how to put together arguments like I made today. Please reach out to me and visit one of our two sites for organizations data into solutions and for individuals analytic dash translator. So, I will stop there and Shannon, if you can help me, let's see if we have any questions, any comments would love to hear from any of you. Wendy, thank you so much for another great presentation. If you have questions for Wendy, feel free to put them in the Q&A panel. And just as the most commonly asked questions, just a reminder, I will send a follow up email for this webinar by end of day Monday with links to the slides and links to the recording of this session. I don't see any questions coming in yet, but I will give everyone a quick moment. So, but there was a comment that came in in the chat when the correlation does not mean causation earlier in the conversation. Yes, it is absolutely true. It does not mean causation. And these relationships are helpful for them understanding. And it also helps us identify. So even if it's not a cause, when we find out that people who are more likely to quit have certain characteristics, it helps us understand who we might target if we have limited resources. So yes, it doesn't mean that we can fix it. But it does mean that we do have a better chance of identifying, but absolutely. Thank you. And how do you overcome the let's not boil the ocean thinking when you say let's integrate our data. Well, usually what I do is I use these types of examples to illustrate why it is that it would be helpful. And we can start incrementally. It doesn't mean that you have to do all of it all at once. But ignoring it and saying that's too complicated. It really is sticking your head in the sand, rather than using what's available to you. Nice. And what about privacy considerations? Are there any there are a lot. And so, if you integrate you need to make sure that it's an organization that has absolutely every protection possible. And make sure that the individualized data doesn't go back to the owners of the different groups but instead it aggregate so that you aren't violating any privacy concerns, but it can be done. It tends to be the first thing that people get worried about, but it can absolutely be protected in a way that individuals rights are not violated, and yet you're able to get the insights so absolutely yes there are, but you can overcome them. What about fears of data linkages and fear of loss of privacy. Yes, I think that what people get worried about is if somebody is at risk of a serious medical event or at risk. For example, people get really concerned about the mental health example that I gave, and the way that that is used is simply a flag on a record for individuals to provide mental health care. So, if somebody contacts them a flag says for whatever reason and it doesn't have to be risk of depression it could be family circumstances or some other kind of circumstance, so that that person needs extra care. But because those people already deal with mental health. They already deal with privacy. So, there is always a chance that somebody could misuse information, but we can put safeguards so that this is about supporting people, not about retaliating against somebody who has risk. I love all these questions coming in. So, is there a benefit to focusing on data literacy efforts on only the people in an organization that are more frequently using the company's data. Let me see. You know, my sense is, and it's my bias with analytic translation. I, I have concerns about what we are focusing on in literacy. So, to answer your question, we, we really need people who handle data to be literate in a variety of areas. So, for example, you have to understand workers comp data in order to interpret workers comp. I mean, that's one example. Medical claims data is immensely complicated and hard to interpret. So, we need in depth literacy in some areas, and then we need a decent level of literacy in other areas. And the more we have experts who are in between, like analytic translators, both who know how to deal with data but also understand different types of data, then the better we are but asking every single person to become literate at every single type of data is not going to work. So, and can you discuss how to establish causation, like one non spurious to time order three co variation. Yeah. So, we all know that the only way to officially establish causation is through very tightly designed interventions with control groups and random assignment and a variety of other things so officially the way clinical trials get done in order to establish that a medication or a treatment actually caused somebody to get better. That is officially the only way to establish causation. Now that said, that exact argument is what the tobacco industry used forever and ever to say that there's no way that you can say that smoking caused cancer, because we can't randomly assign people to smoke. So it's probably just people who like to smoke happened to be predisposed to cancer. So we can take that argument as far as we want, or we can decide that if we see patterns. It is worth considering depending on the cost and the investment, understanding how else we might intervene. For example, if we see that individuals who rate their boss poorly on engagement surveys are three times as likely to quit in the next six months. We can't say that it's because of their boss, it could be that people who are unhappy in their work rate their boss worse. We don't know which is the cause. But that doesn't prevent us from considering. I wonder if we ought to have some peer to peer training for the bosses that have low ratings, because it isn't harmful to have them get some kind of improvement in the way that they manage people. Or we could do focus groups and see whether or not people who intend to leave or who did leave say that they left because of their boss. So, I totally get where you're where you're coming from, because I was trained in experimental design. And when I first got out of school, I really had trouble with this idea of, we can't ever know anything, not really know it. But we have to understand now I wouldn't implement huge million dollar solutions on a hunch, for sure. But we can start to understand a little more, even though it looks like it is, as you might say spurious or whether or not it's just co variation. Now it also helps if things happen after the cause potential cause. And so time series is also another wonderful way to look that if it happens frequently after that happened. And yes, we can make some of those assumptions better. But it's, you know, it's a good question and one that we debate all the time. Thank you. And Wendy, can you start the data integration conversation without first talking about data literacy. Yes, I do all the time. I think that the big picture like I described today of why you might want to know about a and B, in addition to see can encourage people to make that move but I tend to be a big picture thinker. So that's probably my bias. I do think that people get interested in data when there is an interesting surprising finding, not the reverse. So they don't want to dive into understanding statistics because it's interesting to understand statistics and understand variation. And they get interested in that because they got surprised or amazed or enticed by an actual finding like the ones that I described today. Thank you so. I'm kind of thinking as a tool as well. When solving data challenges are these anonymous or that was a comment but then they also followed up with a question of for you are do you do anonymous or open surveys. Okay, so it depends. Open surveys are anonymous and they are aggregated at the department level know fewer than 10 people. So that one is always anonymous to get people to be honest in some of the health surveys. They are identified because they are enticed by getting a discount let's say on their. deductibles or some kind of discount on their premiums and then you have to wonder whether or not they are being completely forthright because a lot of people don't want to share that information. So there are pros and cons to both of those things. But we do tend to see that even if people hedge their bets on reporting, how much they weigh or how much they smoke or those kind of things. It does tend to still identify risk. And that makes sense. That is all the questions I see currently I do have. All right. Thank you so much Wendy for another great presentation lots of kudos going on here for you. So, including when this is excellent presentation one of the best I've had in the area of data integration. Yeah, yeah, yeah, so. Thank you everybody. Thank you Wendy and to answer the most commonly asked questions and I do see that in the chat to just let everybody know again reminder I will send a follow up email by end of day Monday for this webinar with links to the slides and links to the recording as well as Wendy's information. So thanks everyone for being so engaged in everything we do and Wendy thank you so much. And thanks everybody for attending and we will see you next month. Indeed. Thanks everybody. Have a great day. Thank you.