Decision Intelligence and Data Storytelling with Ganes Kesari

Ganes Kesari: Data scientists explain what they have found, but they don't often provide what the users need. What do the users need to solve a problem? What will enable a particular decision they need to make in their workflow?

Clay Hausmann: I'm Clay Hausmann, CMO of Aktana, and host of Contextual Intelligence. Today, we're fortunate to be joined by Ganes Kesari, the co-founder and chief data scientist of Gramener, which uses the power of storytelling to help businesses make smarter, data-driven decisions. Ganes advises CIOs in ways to strengthen their data science programs and see ROI from their AI initiatives. He's contributed articles to Forbes and Entrepreneur Magazine, he's spoken at TEDx, and now he naturally continues his industry visibility, culminating with his appearance with us here on Contextual Intelligence. Ganes, welcome to the podcast and thank you so much for joining us today.

Ganes Kesari: Thanks for having me.

Clay Hausmann: So, let's start at the start. You co-founded Gramener about 11 years ago, I believe. Could you begin by telling us what were you seeking to solve? What motivated you to start the business? And what problem were you trying to solve?

Ganes Kesari: The biggest challenge when we started the company was data being disconnected from the decisions. When we looked at organizations, they had a lot of data, and some of them had great analytics as well, but it was not leading to actions on the ground; so that's what I mean by data being disconnected from decisions. We wanted to solve that; help organizations use their data better, and whether it is a strategic decision, or a tactical position, make it data-driven. That was the mission we started off with.

Clay Hausmann: And what created that gap? What was preventing data from connecting to the execution, or the decision that needed to be made?

Ganes Kesari: We call it the last mile challenge. Often, organizations that do analysis of data, they might have data scientists today. Data scientists explain what they have found, but they don't often provide what the users need. What do the users need to solve a problem? What will enable a particular decision they need to make in their workflow? So, that is the consumption challenge. The end-user needs to understand what the data is. And importantly, they need to be given a suggestion as to what action they should take. That's how then you bring in data science, analytics, and data storytelling, you're able to bridge that gap, and not just explain what you found from the data, but also provide what a user needs to make that decision.

Clay Hausmann: So, is that a matter or a challenge of sort of translation where the data is in a state where the non-technical user, or even the mildly technical user has a hard time interpreting exactly what they should take out of it, and how they should apply it? Is it a matter of context, which obviously we talk about a lot on this podcast where you need to take whatever the data is, and put it into the relevant context for the decisions that need to be made. Where did you find to be the biggest challenge, and you focus most of your energy on solving?

Ganes Kesari: There are three elements, which also happened to be the decision of data storytelling. So, if I were to first define the practice, it's the process of identifying insights, and then explaining them often using visuals in a way that promotes understanding and action. So, if we break that up, so there are three parts. So, the first one insights identification, that's where analytics models and all of these come in identify patterns from data. The second part: explanation, that's where data visualization, context narratives, all of those come in. The third part is where you connect it to a business workflow and action, and you make up a specific prescription or a recommendation to the user by understanding their challenge. And this brings in the social science aspects, understanding the people and helping them make a specific decision. So, when you have all these three combined that's data storytelling, and that solves the last mile challenge we earlier talked about.

Clay Hausmann: It makes a lot of sense, and it's very similar to our experiences. One of the interesting aspects of the success of an AI project is that although the temptation is to believe that it's a highly technical exercise, it is as much of an ethnographic or human exercise in terms of determining how it can be successful, because it's augmenting humans and teams and the way that they make decisions. And in many cases, it's not replacing them. In some cases, it is, but in most cases it's not. It's augmenting them and you have to figure out that coexistence. So, let's dig into the storytelling part a little bit. It's a personal passion of mine, one of the most formative parts of my career was I spent a couple of years pursuing screenwriting, and it really opened my eyes and changed my perspective on how to build a connection with an audience, how to be empathetic to what they're looking for, how to put it into the appropriate context and environment in a way that they can consume it as opposed to what you may want to communicate. So, I'm very interested in this aspect around your business and your approach, where you have a storytelling emphasis. Can you explain a little bit about what that means in context of the work that you do.

Ganes Kesari: Looking at your background, and the technology, and arts, I'm sure you must be a great storyteller. In fact, what we have seen, even in our organization, we have seen that when people come in from a non- technical specifically arts or social sciences background, the confluence of these skills actually makes a great recipe for storytelling. So, I'll give you an example. This is a telecom client we were working with, and the challenge they had was attrition. And we were planning to build a machine learning model that could detect churn, understand and identify the customers who are likely to leave. So, that was the initial objective we set up with, so now when you give it to data scientists, they can bring in the best of models and identify a list of customers who might churn next month, and they assume that their job is done there. Whereas when you talk about the users through the marketing team, and the people who are in charge of retention, just giving them the name of a customer and saying, go retain them is not sufficient. They need to be explained as to why these customers might churn, the explainability part of it. So even if you have a black box model, there has to be some element which says that we think these people would leave because for instance, the pattern of spending has been different, or that some of the behavior which is changing. So, the explanation is one part. And number two, be able to share a recommendation as to what they could possibly do, because the same action may not work for all users. So, if you can identify based on the user behavior, the demographics, what is likely a successful retention action, that is where the recommendation goes in. And then you combine that with visuals that the users can understand, doesn't have to be fancy visuals, but anything that the user can process and act upon. So, when you have these elements coming in, that's what resonates with a non- technical user. They are convinced that, yes, this makes sense, and they are clear what action to take. And there's a visual summary, which makes it easy to look at it every time they come back, and look at it on a weekly or a monthly basis, or it could be on a daily basis, they just get it. So, that is the power of data storytelling.

Clay Hausmann: Interesting. And how has that grown over time? Was that part of your initial mantra as a company? Did you have that emphasis on storytelling from day one or has that kind of evolved as you have gotten deeper into your work?

Ganes Kesari: Actually, we started off with data storytelling even before we got into machine learning. We saw storytelling as a big need in the market. So, the first year, most of the projects were on only data storytelling and then as we saw customers adopting it, they started asking for deeper insights. So, the second year of our operation, we got into machine learning and deeper insights, and since then we have been going with the practices hand-in-hand. Along with our technology platform which is used to build those custom solutions, and enable decision making.

Clay Hausmann: Have you along the way, tried any forms of storytelling or visualization that did not work? That surprised you because a lot of storytelling is experimentation, and trying different things and seeing what connects.

Ganes Kesari: Very true. Yeah. There are several engagements where we've come up with visualizations which the users may not understand, or the enthusiasm at times we come up with, for instance, a diagram like a Sankey chart or a chord diagram, which for a non-tech user, they may not understand how to interpret it. So as again, data scientists or information designers, we are very excited about it. But unless the user understands how we interpret the chart, it is difficult.

Clay Hausmann: Yeah.

Ganes Kesari: There are some cases where through training, and once people go through work for, say a week, they get it, but you won't have that patience from all the users. So, you'll have to really customize it, and along the way, we have strengthened our practice of user, persona building, studying what really works, the preferences, and all of that, and that's where the social sciences, and requirements, in this training, all of those things are very important, extremely crucial for storytelling.

Clay Hausmann: Interesting. There are parallels in a lot of ways to screenwriting, but it's changed in that industry. So, it used to be that when you made traditional studio films, they use the language around a four quadrant film. You're trying to hit the four quadrants of the segments of audiences. So, you would have as broad a reach as possible cause you wanted to have huge box office, but now with streaming services, you can go very hyper segmented. So, you could do a storytelling tool, like you mentioned, that just appeal to data scientists. Because now if you have an interest in, I don't know, Norwegian veterinarians, there is probably a whole thread of shows that you can watch on Netflix about Norwegian veterinarians, but for the work that you do, now you are still in that prior realm where you need to make it consumable for a variety of audiences. It can't just stay in a very specific lane for a very specific user. It has to be consumable by everyone.

Ganes Kesari: It depends again on the user segment we're targeting. There are cases where we've worked with media organizations. For example, we did some work for a large media house in Asia, and they were covering the general elections in India. And here you're talking about billions of users who are going to be watching it on national television. So, there you'll have to really make sure that it appeals to a wider audience, and the visuals... do they interact with the visuals we created? It had over 10 million views in about 12 hours on the day of the results. So, there was a lot of excitement and what ideally we were happy about was that moving on from numbers, and just take course, which I'm talking about more than five years back when national television is showing just the numbers and saying, this is a state of the constituencies and the candidates. So, from there we brought in visuals and the anchors at the media companies were able to do real time analysis saying, this is what happened last time on the last elections. And this is a watershed, how's it shifting dynamically. And they were able to talk a little more deeper insights. And the user reaction, the general public reaction was phenomenal. So, that is an example of casting the net really wide, and catering to everyone. But if there are some engagements, we do solutions which are specifically for decision- makers. For instance, there are a set of KPIs which say COO or CEO and small group of five or 10 people have to consume. There we go very targeted and look at what kind of insights resonate with this audience. What kind of visuals? What kind of color schemes resonate with executives with a certain age segment and so on? So you'd have to go really targeted, like the Norwegian example that you mentioned. So, there are enough cases where you need to do that as well.

Clay Hausmann: Again, we keep coming back, trying not to be too obvious about this notion of context, but understanding the context as you just described; the context of the mass market of the India election, versus going to a user group of 10 and what they're trying to accomplish-

Ganes Kesari: Absolutely.

Clay Hausmann: Is going to influence everything that you do, and if you apply the same methodology to both of those exercises, one of them is not going to be successful because they are so unique in their attributes. Ganes, how often do you get brought into do or apply AI and machine learning for the first time versus to improve what they already have in place? It's kind of a loaded question because often in life sciences, when we're working with a customer, it's the first time they've invested in AI and machine learning for commercial operations. So, there's a lot that we need to build from the start, which is a bit different than when you're coming in to refine or improve something. Where do you typically engage with customers?

Ganes Kesari: Over the last 10 years, we have a wide footprint that are existing customers, many of whom we are working with for more than seven, eight years. So, often I would say we have brought them to start their AI journey. That would be the majority of the cases. And there are some clients who are already doing, and they want specific support in certain say computer vision projects, but the majority would be fresh starts, and maturing the customers. The way we work with clients also, as opposed to a particular project or a machine learning initiative, we help them improve their data maturity, and how to build a data road map over time. And typically these are multi- year engagements. You start off for example, that there's a customer who has been doing BI, but say Power BI or Tableau dashboards, and they want to do something even deeper, and more advanced. And before getting them into AI, a natural next step would be to do diagnostic analytics because they are doing descriptive summaries. For example, find out if their market share changed. Why did it change? Which customers are leaving them? Diagnostic analytics. And then slowly get into statistics, machine learning, where you bring in a predictive aspect, and simple predictive aspects, for instance saying which products are likely to have more defects. And then you really get into those high-end computer vision, natural language processing, but the language models, all of those come in after that. So, it's important to ease the organization and the users through the different stages of maturity, and that's how we approach, and when we feel the users are ready for it, and the problems are big enough to bring in AI. That's when we suggest it.

Clay Hausmann: Do you find that there are commonalities in those situations where you are getting started for the first time that you have a bit of a starter kit that you apply to work with customers? Or is each situation different and unique?

Ganes Kesari: There are some common best practices that are four or five things, which we have seen always work. And they have to be obviously tailored to an individual organization's needs, but if I were to call out some common patterns, if let us say someone wants to get started with AI, they already have some maturity, some presence. The first suggestion would be to look at what are the outcomes? And who are the users they want enabled first? So, being very, very outcome-focused is important. And number two, based on that, pick up some small wins, what are those projects which can be executed fairly quickly, say a couple of months as opposed to one year or even six months? So, identifying those quick wins, small projects. And third, starting off with their data, and tools that you already have. Often an organization, they think, they try boiling the ocean, they try to set up elaborate data warehouse, which in itself is a two years project. And they say that we start, we will do machine learning after that. You really don't have to queue it like that. With the data that you have, as long as it can organize and ensure data quality, you can start with that tomorrow. And while you demonstrate the value from machine learning, then you can convince the users to invest or convince the executives to invest more in data warehousing, so that many of these initiatives can go hand-in-hand.

Clay Hausmann: I want to transition to a topic that is a challenge. It's a bit elusive for some people on the topic of AI, and that is ROI. Obviously that's something that many who are tasked with driving these initiatives internally at their companies are also being asked to show direct impact metrics. How have you seen some successful examples, and what do they share in common in terms of being able to demonstrate ROI that allows the initiative to move to the next step and beyond?

Ganes Kesari: That's a great question. ROI from AI or any data initiative, that's important because organizations, when they don't pay attention to this metric, if they just start a project, there's a high chance that they might get the solution, but they're a few months down the line because they don't see the benefits from it. And no one is talking about quantified outcomes, whether it is a cost saving or additional revenue generator. So, there are four steps for going about quantifying the ROI from AI or an initiative like that. Number one, we talked about outcomes, identify what are the outcomes and what are the indicators of success? Step two, define what are the metrics that can sufficiently represent those outcomes, those end states, and start collecting the data for that. And number three is to establish that the benefit we are getting is really from this project attribution, the third step. And fourth one is to factor in all the costs and add up all the costs, all the benefits, and then really work out the math to compute the ROI. So, these are the four steps that organizations should follow.

Clay Hausmann: ROI metrics typically fall around either a sales lift, or around efficiency metrics. And that's not to say... It can apply to many other things, but I know in a lot of commercial operations, in terms of the application of AI and machine learning, it drives one of those two quite often. How can I get more out of my existing resources? How can I make my team or my teams more efficient? Or how can we drive top-line results by being able to provide a personalized experience to our audience, so that they become then more likely to purchase and engage with us. Have you seen any particular area where it is easier to gain access to ROI metrics, and to draw that direct line that will allow a project to move from a pilot to a full-scale deployment, or from a full- scale deployment in one region to many regions? What have you seen be the most successful in those ways?

Ganes Kesari: It's not very easy. And in fact, the point you mentioned like top-line sales lift is a great measure to have, if we can establish that. But as efficiency improvement, it can be a little fuzzy. So, people... The productivity increases, are probably you're able to deliver things a little faster. What does that mean for a CEO who's funding this effort? Yes, people are able to get more done, but does it translate into more capacity based on which I'm able to deliver more business, or so more customers? So, we need to translate that efficiency gain into additional... Again, top line or bottom line impact. That's where the challenge comes in because you will have to make the attribution and the connection, whether it is within a project or across the organization. And that's also the reason why a lot of organizations struggle to measure ROI because outcomes are fuzzy, and often you may not have the data to measure it. Even getting the data to execute a project is difficult, but now to measure the outcomes, you need to collect a separate set of data. For instance, if it is making your customers happy, it may not show up in your regular voice of customer reviews, the data that you collect there. You might have to set up a new process to collect new data, which asks specific questions saying, have I improved in these aspects? Would you likely increase the business that you give me? So, we need to collect some new data, and we need to establish that linkage. That's where the challenge lies.

Clay Hausmann: Yeah, that makes sense, and we've seen the same. I find we're most successful when we're able to partner with an in- house impact analysis team, because then there are very complimentary skills in terms of the access to information and data, the historical perspective they have, and the cultural perspective they have in terms of the types of analysis that work most effectively. We can then partner and bring best practices from across all the different projects that we've done. Is that something that you find as well, that when you're able to work with an in- house team that is get a similar charter and you can partner with them that it enhances the work that you do on impact analysis?

Ganes Kesari: Yeah, that's a good point. Yes. Ultimately the buy- in has to come in from something internal and rather than say an external partner presenting that these are the benefits. If there's some internal impact assessment team or a business leader who comes forward and presenting, this is how we have benefited. And then we can help those business partners to measure and quantify, and then they can take the message ahead. That always works; otherwise, it's like as an external party coming in and sharing, is it really objective enough?

Clay Hausmann: Very true. Well, one of the other things we find in terms of making these programs successful internally is that you need to help coach your customers in terms of building alignment across C- suite peers or across departments, because AI and machine learning efforts are rarely confined to one department. That's part of what you were talking about earlier that data scientists may speak a certain language, but when you need to move outside of that team, and start to work with other business partners, they may speak a different language or have a different technical proficiency. So, how do you build that alignment or coach customers to achieve that alignment with their departmental head peers or even down into their teams, so that they are working together and are collaborating around what the AI and machine learning is producing or helping it to influence all of their decisions, as opposed to just one department over the other.

Ganes Kesari: Bringing in that broad-based support is extremely crucial for even execution, and eventually adoption. Ultimately the solutions, many of these AI solutions have to be used by people across departments. The best practices we have seen are one, there has to be a clear owner, why we can have many departments coming in. There has to be one owner who drives the initiative, and is responsible for success of the initiative, for success and ROI. And then onboard leaders across these other departments who could be broadly stakeholders, either users coming from those groups, or some of them have to provide the data, which will be used by the initiator. Let's take this example of customer experience improvement. So, while that can be a customer experience owner, but in the organization who has a horizontal role, they will have to work with the product teams, with the marketing teams, and sales teams. And there are several other stakeholders. So, some of them, for instance, the product team, they will have to give feedback as to how to improve the features. And I include the customer satisfaction, whereas with the sales and marketing, but the kind of intervention will be different. So, we need to have leadership buy- in across all of these teams. Think that yes, they will be involved and the specific defined role for them as part of the engagement, while it will be shepherded and owned by one particular leader and one team, which is building and owning it. So, that actually works well to bring about collaboration within an organization. And this has to be sustained throughout once you have project going live, once it's taken to production. You will have to periodically reach out, do road shows, and report the success stories across all of these teams. So that way this building, this initial synergy amongst the teams and ensuring collaboration is important.

Clay Hausmann: So yes, I'd like to... We could keep talking for quite a while here. It's fascinating to me, and the perspective that you have, especially around conversion of data and insights into storytelling. I just, I find very interesting, but I want to ask you since you've spent the last 11 years, very deep in this work, a lot of what I think holds up AI programs or adoption of it. Sometimes there're misconceptions that there is... it's not hard to develop a very superficial level of knowledge because there's so many articles being written, and there's so much discussion about it, but that becomes dangerous because it can mask a lack of awareness and understanding for all the details that are down below the surface, which ultimately make these programs successful. What are some of the misconceptions or pitfalls that you've seen over the last 11 years of working with customers that you helped them overcome so their programs can be much more successful and longstanding?

Ganes Kesari: Yeah, There's no paucity of buzzwords in this discipline when it comes to data, we've had big data, AI, and a lot of these big words, which probably mean a lot of different things to different people. So, it's easy to get confused. If we were to pick three and demystify, I would talk about business intelligence, and second, say, data science, and decision intelligence as three broad heads. They are important because most organizations, when it comes to business intelligence, how do we define it as simple reporting and descriptive summaries from data, before little business intelligence, ad hoc reports, which tell what happened in your business yesterday. So, that is business intelligence, and data science goes a notch higher. It brings in deeper techniques and is able to answer questions using, say, diagnostic techniques and predictive techniques. That's where machine learning, AI, and all of these other subsets of techniques come in. All of these fall into data science. Decision intelligence is where it's actually by definition it is the practice of ensuring you convert information into organization-wide decisions. And that's a very important discipline, Gartner says that in the next 10 years, that's going to get a lot of attention, decision intelligence specifically. And there are several roles which will be created around that. And data science, there's a gap, again, we spoke about data science versus decisions. There are gaps, and decision intelligence brings in two other practices. One is social sciences, and second is managerial sciences, but we talked about social sciences, the people aspect of it. As part of managerial science, that's where change management and running projects to ensure that they're assimilated, and they are absorbed by people across the organization. All of those are part of decision intelligence. So, for an organization to be successful, but using data, they need to have all these three working hand- in- hand even when they are doing machine learning, AI, there is a place for BI and similarly data science is not sufficient. They need to think decisions and decision intelligence is the discipline, which can help them with that. So, those are the three I would call out.

Clay Hausmann: Wonderful. Well, I'd like to transition into what I think will also be a pretty interesting section of this. And this is where we talk about Ganes in context. This is about you and some of your influences. So, if you're game for that, we'll transition into that. And I'm going to start by asking you who has been an influence on your career that might surprise us

Ganes Kesari: I'm hugely inspired by Bill Gates, for obvious reasons, the foresight execution, how he built... But the surprising part is I'm more inspired by his journey post Microsoft, how he has used his influence, the money, and has skills to create impact, social impact in the world is just mind- boggling. So, that's something which I would want to emulate and do more.

Clay Hausmann: No, I agree with you completely. And I think not only that, but there are many technology founders who establish a foundation like that, but they don't pursue it as exclusively and rigorously as I think Bill Gates has. So, I agree with you. I find a lot of inspiration in that as well. If money was not a factor, what career would you most like to pursue? And it can't be what you're doing.

Oh, okay. I mean, I was about to say what I'm doing is actually the right sweet spot. So, any role that brings them a combination of big picture thinking, I like learning and keep reading and different forms of learning. So, anything that helps me understand the big picture trends, how are things moving and where are we headed, in the way the world headed? So that kind of ideation and big picture thinking is something which I would definitely want to have in a dream job. And the second aspect is I don't want to be away from execution and guiding teams to deliver value. So, how can I combine this big picture thinking, but advisory and execution so that we solve some real-world problems. That would be my dream role. And that's why I mentioned it is closer to what I'm doing.

Clay Hausmann: I was going to say, I think you might be in it. I think you might be in it. I hope your co- founder doesn't redraw the revenue sharing part of your agreement now that he knows that you enjoy this work, and would do it for free. And what profession would you most not want to pursue? No matter what it paid.

Ganes Kesari: I think the opposite of this often I've... Before Gramener, I've worked in large organizations. So, I dread if I have to get into a large organization, playing a narrow role with less flexibility, because what you can do in your own startup, and that the kind of flexibility, and impact you can create, and how dynamic you can be in terms of redefining your responsibilities. So, any role which is large organization, narrow role is something which I would hate.

Clay Hausmann: Makes sense, makes sense. What is the best book or film or show that you've enjoyed recently, and why?

Ganes Kesari: A recent book, The Three-Box Solution. This is by a professor, his name is Vijay Govindarajan, and he's at the Tuck School, Dartmouth. This is a great strategy book that shows leaders how to innovate and execute. So, that's what he calls the three boxes. Box one is to do the execution of your core business. And box two is forgetting the past selectively, those practices which will hold you back. And box three is innovating for the future; experiments and innovation, incrementally doing it so that you stay relevant in the future. How you balance these three boxes is what the book talks about because organizations often get stuck on box one, to run their day- to- day business. And they don't think enough about two and three, or organizations get too obsessed with the future. And they don't the capacity by letting go of things, which is what box two is about. So, I think this is a very simple and fascinating explanation and a lot of great examples in the book.

Clay Hausmann: That sounds very relevant. This is the best part of that podcast for me because it's my fastest way to build a great reading list. So, that's helpful for me, I will find that. You are at a family gathering, and your eight- year- old niece asks what you do for a living. What do you tell her?

Ganes Kesari: Okay. I've often thought about this, and at times when people ask me in gatherings, so I've tried different questions. My most recent one, I've tried this out with my kids as well. So, this is how I would explain it. Let's say you have$ 50 in your piggy bank, and my niece, so if she's looking to buy a toy that has fun, safe, and also has some learning element, how would she go about making that decision? So, she has to look at all the toys available, find out what is within her budget, and which of these tick the boxes of fun element, safety learning, and you'll have to balance all of it and make the final choice. So, that is what I do for a living, helping organizations do that with whatever money they have, and help them make their choices, and get workers, deliver the business, and what they will like doing.

Clay Hausmann: And how has that gone over with your kids?

Ganes Kesari: It's actually good. In fact, recently they were trying to buy a Lego set. So, a similar process I laid out and I asked them to make the choices in a little part, so it seemed to work.

Clay Hausmann: Good. Well, that's always the key. All right. Our last question, your ultimate dinner party for four, who's there and what is served?

Ganes Kesari: Given the current times, I would love to have a larger family gathering. All of us would like to get together, and after one and a half years of this pandemic, and isolation. So, that would be good, but since you're mentioning four people, so it has to be wife and my two kids and they... Like I mentioned, Lego is one, and then they are a lot into history and nature, which I love too. So, a themed party on any of these themes, and you see Mexican we all like, so that would be our favorite.

Clay Hausmann: Very nice. Well, I thought for sure Bill Gates was going to get a spot at the dinner, but I think he loses appropriately to the wife and two children. That's great.

Ganes Kesari: Right, yeah.

Clay Hausmann: Ganes, it was so great to have you on today, really enjoyed the conversation. I thought your insights were really helpful, so thank you for joining us today.

Ganes Kesari: Loved the conversation. Thanks for having me.

Clay Hausmann: That's it. For this episode of Contextual Intelligence, I'm your host Clay Hausmann. You can find all our episodes on Apple Podcasts, Stitcher, Spotify, or wherever you get your podcasts. And please leave us a review, or a comment, or a question, or all of the above so we can make sure that this podcast brings the proper context to your work. Thanks everybody for joining us.