Video: The CoE’s Role in Onboarding, Community, and Long-Term Engagement | Duration: 3524s | Summary: The CoE’s Role in Onboarding, Community, and Long-Term Engagement | Chapters: COE Engagement Introduction (0s), Introduction and Speakers (68.30090392561985s), Building Ambassador Relationships (103.61091392561983s), Customizing COE Support (274.7959039256198s), Documentation Best Practices (341.6158839256198s), Documentation and Customization (492.3659039256198s), Defining User Journey (781.7209039256198s), Communication and Engagement (1167.0259039256198s), Communication and Reusability (1683.7209039256197s), Scaling CoE Support (2136.61590392562s), Train the Trainer (2449.05590392562s), Encouraging Data Reuse (2728.39610392562s), Gamification in Analytics (2890.67090392562s), Conclusion: COE Support (3249.3609039256203s)
Transcript for "The CoE’s Role in Onboarding, Community, and Long-Term Engagement": focused on governance, but governance is just one aspect of the COE mission. Today, we're switching gears. We're diving into another critical part of the COE's role, and that is engagement. So this session is called the COE's role in onboarding community and long term engagement. Because here's the truth, getting people on to an AI platform is a major milestone, but, you know, how do you keep people to stay active on that platform? How do you help them build confidence in their skills and make AI a part of their everyday work? It's what happens after that, so how you keep people engaged that drives long term engagement, a long term impact, and unlocks real scalability. And that's where the COEs come in. In today's session, we'll take a look at how teams are designing for engagement from day one through intention onboarding, evolving user journeys, and strong internal communities, and we'll show how that's coming to life in Dataiku, the universal AI platform. So without further ado, I will pass it to our two experts to talk about this topic in greater detail. So I'll pass it over to you, John and Christy. Hi. Good morning, good evening, good afternoon, everyone. If you're joining us here for our fifth webinar, this is your first one. Welcome. As Renata mentioned, we'll be focusing on talking about community today. My name is John Tudor. I'm a director of business architecture here at Dataiku, leading our self-service data enablement program. Christie, if that's the interest yeah. Thank you. And I'm Christie Hampton, and I have led, several community of excellence, center of excellence, community of practice initiatives at several organizations. So happy to be here to share my knowledge with you all. Great. Thanks, Christy. So with that, let's, dive right in. So first and foremost today, I'm a few of you in part of our other sessions. We've talked about the importance about bidding operating models, teams, and working closely with the businesses of CUA. The next extension of that is really this idea of how do you build more evangelism and more support for your self-service program and the use of Dataiku within the organization. And in many companies, there's this idea of power users or what I call passionate users. Those types of individuals who really care about the mission idea of democratizing this capability to the organization. And we tend to talk to these, you know, type of users as ambassadors or ambassadors for the program itself. What you'll see on the slide is the idea of how do you start to build stronger relationships with these ambassadors so you can build a better relationship across the business. The foundation of community is having that active activity between the COE and the community organization. So how do we think about doing that? On the left side, here you'll see a number of things that this partnership helps increase. It'll help increase your overall ROI, return on investment as you have more and more data products being built by the community. It'll also further, you know, enhance the career development of those individuals as they become part of this community. You also increase that level of trust. Additional things will happen such as your feature adoption, the number of active users, etcetera, even your cycle time. But, ultimately, there's many, many benefits to that partnership increasing that really helps the entire program. Now importantly, how do we start to think about really improving these relationships? So first and foremost is that as a COE team, as you look to make decisions, you wanna really focus on that design thinking paradigm that we covered in the third webinar. How do we think about really focusing on that enablement journey and doing the things that are gonna make the most sense for those users? How do we bring them into the conversation and build things directly towards what will make them successful? Another option is to do surveys and interviews so that you can actively get feedback. Right? As a human, we wanna provide feedback. We wanna feel like our voice is heard. So we wanna make sure that we're sending those surveys, interviews out, and getting the collected information so we can continue to get better and build a better relationship. Further, as we get questions, as we have to explain things to the COE, we should be looking to help encourage those ambassadors and explain those items to them because those ambassadors can then take that information and explain it further to their business peers and partners. So almost like a network effect. You also wanna make sure you prioritize support. So how do you ensure that the support you're giving really gives the right amount of support and care for those individuals so that they feel comfortable providing support to other users or passing those potential other peers over to your team to work with self-service? You also wanna think about your backlog duration. You'll have many things that you requested from the community. How do you take a look at all of that and kinda stack that into priorities and make sure you're executing? And, importantly, while you do that, look to prototype and test that, especially with your ambassador community. Give them a bit of a chance to try things out and have their voice be heard. And And then finally, you wanna make sure that you're regularly updating. That's a usual recommendation to have monthly meetings with this group of ambassadors to share best practices and to hear feedback from the field as well as give updates in the overall program. I think, additionally, I just wanna add that when you're setting up a COE, it really is you get a lot of engagement when you're setting up at the team level. Of course, throughout your enterprise, you may have different users, different teams, using the application, using it differently. And that's one of the things that you wanna get to know and assess. Like, how are each team planning on using your application, and what kind of support will be needed? So one of the things you'll find as you start maturing your COE practice is that you're gonna be needing to, create new incentives engagement opportunities that might be different from one team versus another. Teams that might be new in your analytics journey might need a little bit more hand holding than teams where you're dealing with more senior data scientists, developers, more used to just reading the documentation and getting started. So do be prepared that support is gonna look a little differently, engagement's gonna look a little bit differently, and you might have to assess what services and support may be provided across the different team levels or user personas. Great. So now we've talked a little bit about how do we think about ambassadors and how we maybe differ based on user percentage we're working with. Let's talk a bit about what I would call the baseline kind of support mechanism for building community, and that's actually documentation. Now when I work with our clients on this, I usually hear a little bit of a yawn, a little bit of boredom here, but, actually, it's very key to how you support a program like this. Because, ultimately, those users, your prospective users, your current users, are gonna need a place to go to get information they need to be successful. And, importantly, it also depends on how do you actually define how this content is viewed. So how do you organize the content? So So first, you wanna make sure you put together from a, you know, user perspective. What is that user gonna wanna know first, second, third? So as an example, you're probably gonna wanna have an easy way to get to, like, a fat page, right, for you you early users or potentially get easy pass to get the licenses. So what is some way to know when and where in the journey? Another best practice is to ensure you have a central splash page with a very simple URL, a place where everyone can go and pick up the information about the program or about Dataiku and learn more so they can be successful. And importantly, make that a very easy to access short link because you may have, you know, hundreds or thousands of people trying to get to it. If you were part of our first part of the series in the webinar, we talked about this idea about a branded mental model and organizing your program and structure around this to make it easier for the prospective users to consume and understand and for executives to align to. The same idea applies here when you're building documentation. You should structure your documentation around the branded mental model that you've used to guide the organization so that you can have a better structure that's easier to follow and understand. And then finally, if you have multiple team members working on documentation, it's recommended to utilize the template to drive consistency across that documentation while multiple teams contribute to it. And what you'll see here is a very common practice that comes up when we talk with customers, which is this idea that at Dataiku, we have a lot of documentation that we provide to you out of the box to work with with your users on. But by the same token, as an organization, you're gonna wanna create your own documentation because the way that you operate will be different than maybe what's necessarily public and out in the world. So if you look here on the Dataiku provided content column versus your content, you'll see some comparisons here. So as an example, you have the Dataiku knowledge base, which provides details on all of our technical capabilities and architecture. Well, similarly, you may have your own governance processes as an organization or maybe you don't want all that capability being used. Maybe you have a certain reason for that from a compliance standpoint, a regulatory, etcetera, or maybe you just have different usage patterns in mind. Regardless, you're gonna wanna document those type of processes to make sure that the organization understands that and so you have clear direction as opposed to maybe the generic documentation that's available. Another example might be our Dataiku Academy, which is dedicated training for those who wanna learn how to use more Dataiku. Similarly, you may have your own internal training and schedule nomination processes. So you're gonna wanna document that and provide that to users as well. Another example would be our Dataiku video library. These are videos that show how to use Dataiku and demo working with the software. Well, you may also wanna have demos within your own business to do this. It's recommended to create demos that show your potential users how they'll walk through a data workflow, that do so in the context of your organization. So if you're a CPG company, look at how you can look at your products within those categories and track that from a data pipeline. If you're in manufacturing, maybe you apply supply chain processes. But the goal is to have your examples be as close as possible to the day to day work those users are gonna do. You may also wanna have sample projects that are provided by the Dataiku, but, similarly, you may wanna have those similar distinct connection and configuration details for your organization. So while a sample project may show how to do this, you may have something unique to your company. How do you represent that? Further examples include our business solutions, which provide example use cases that you can directly deploy in your organization. You may also have specific industry and percent of workflows for your company that you wanna highlight. Similarly, we have webinars such as this series here, which you may want your own getting started and how to get licensed content. We have our Dataiku blog where you can learn more about what Dataiku is kinda thinking and some thought leadership. Well, you may also want to have good contact information for your core team so they can have your users can have those resources to reach out to. And then finally, we have success stories that we share from a Dataiku perspective about success from other customers. We may wanna do something similar where you look to share examples of successful Dataiku projects and self-service projects in your organization to other users. So in addition to the Dataiku provided content, as a COE, especially if you are supporting a data science machine learning or DSML platform, some of the common themes that you can expect to have to be responsible for documenting in addition to, you know, giving users access to the Dataiku or any other platform content is usually customizations. A lot of the DSML, solutions on the market may allow the platform team to customize access accessibility in certain ways, especially when it comes to compute resources. So if your solution that you're using, your application, your platform, if it's not the out of the box vendor provided and if there's customizations that are included, that's definitely gonna be one of the first things you're gonna document. The second thing you're gonna have to document is access to source systems. So whether how should users access different database systems, warehouses, cloud storage, graph databases. Graph databases, for example, may operate a little bit differently from how a user can expect to retrieve data from that particular source versus, you know, reaching the cloud. There might also be limits that your administrator may place on how much data you can pull in at a time or how queries should be written, how much how many queries they will allow you to be be written to pull into the source system. So all of the database, the data layer, customizations or compliance standards at your organization, that definitely is something that you will most definitely need to have documented for a DSML solution. Another common theme, APIs. Different organizations treat APIs differently. Larger organizations may treat that as an actual, product. So there might be different deployment patterns, different governance standards. There may even be a register, capability for the API. So API components in data science solutions may have different deployment routes that are set by the enterprise that need to be documented before users can just freely use those features within a tool. Other things that might be documented, different deployment patterns, how compute may work. You know, some processes may users may easily be able to run CPU related jobs. GPU jobs might need approvals. There might also be, cases for customization, cost optimization that might need to be used, that may need to be specified as well, and then frequently asked questions. So as a COE practice, you always want to keep a pulse and track and monitor what kinds of inquiries are the business users asking. The inquiries that seem to be more frequent, that might be an indicator that you may need to create documentation for those use cases. So that just provides some examples of how documentation, internally may be different from the Dataiku provided or any vendor provided documentation for DSML solutions. Awesome. So now that you've had a chance to talk about building a better sense of community with our ambassadors and building more trust, as well as building a foundation around our documentation, Next question is, how do we really start to define a concrete user journey? And this becomes very important because, ultimately, as users work in your environment, after they first gotten access to Dataiku and your self-service platform, they're gonna have a lot of questions as they walk through their user journey. And if you want users to be sticky and you wanna continue to have them work within the environment, it's very important they have a good experience. They don't feel held up by processes. Right? So why don't we like to look at this as the idea of flow, which you'll see a quote here at the top of the screen. And flow is the state in which people are so involved in activity that nothing else seems to matter. And what this really relates to is this idea that if we set things up correctly as a COE and as an architecture, we're able to go through and actually create a process in a a flow that feels conducive to the user's work and allows them to focus on using their own expertise to solve problems of data. So this means making it easy to get a license, making it easy to connect the data, not having a lot of questions about where to find data sources or getting held up by various activities. A great example would be, you know, writing code to solve a problem versus using the code or providing no code such as a platform like Dataiku. It's much easier to use our own expertise as a business individual to actually solve these problems because we're not worried about learning how to write code at the same time. So as we think about how to write this correct user journey, you need to think about first that really what you're doing is defining how employees are going to do their work. Right? What is work gonna look like on a day to day basis? In order to guide these employees, you just need to make sure they have a process flow to guide them through this. And Dataiku is a universal AI platform can really help carry that entire journey from an end to end perspective. So from the very beginning of developing a new idea, all the way to productionization, authorization of an AI product. So in order to build this, the goal is to have to strive to build that psychological state of flow. Right? So how do you get that highly focused mental state? And, ultimately, you wanna wanna make sure that the processes we create really allow employees to focus on that work and apply it. So as we can develop this, we wanna make sure as well that in the development of it, that it's not intrusive. Right? That it's not something that holds folks back from being able to focus. So what's the approach for how we develop this? So first, as with the documentation and tying back again to our first webinar, it's highly recommend to use that branded dental model as really the high level process steps. If you remember from our earlier sessions talking about the success model and thinking about how that walks through every step of a data product and AI life cycle, You'll see a similar example here in the bottom. This is actually using the previous version of that model, the five s. But what it does is walks through at a high level what each of those steps would be. Right? So search, stitch, science, show, and share. That gives you a high level construct to work against, then you can start to build the dispute process steps that go underneath that. Now another key part of this is identifying the decision points on the user's path, including focusing on their persona, because different personas may have different decision points and directions they go. And And then you need to make it clear about when to make a decision about where to go. So a common challenge I'll see with customers is that maybe there's multiple tools that have some overlapping functionality, and users don't really have any guidance about when they use each tool. And that should really be intentional. So as part of your user journey, you should be defining if you're at this decision point, how do I decide which tool to use, what technology, what is the best practice? And you need to make that very clear from a user journey standpoint. Additionally, you wanna make sure that you're really embedding your documentation. So the kind of from our previous slide directly into the journey. Each step should have corresponding documentation that guides users through how to actually use and perform the step. How do they use data? I could do these actions. That's very, very important so they don't feel like they're lost along the journey. Additionally, you know, once you get to a training perspective, you're gonna wanna make sure that you use this user journey as a backbone of that training program. The steps you define here are the same steps that those users will go through in training to make sure it's all aligned and connected. You wanna make sure that you're always focused on that user experience. So, ultimately, similar to our governance conversation in the last webinar, this process needs to be unobtrusive. It needs to be easy to follow. The more difficult it is, the more friction there is, the harder it's gonna be to get those users to be sticky. You wanna make sure you're reinforcing important concepts throughout the journey. A huge one that I always hear about is reuse. Almost every company I've ever spoken with has a challenge around reuse. Right? So how do we think about ensuring that throughout the user journey, we're coaching our teams to think about how do you reuse first and then create second. So that makes sure that you put those process steps directly into the flow, and you encourage those users to look for something first. And then finally, make sure the process flow is easily accessible, whether that's up on a web page, whether that's printed out and stuck up on a wall. I've had experiences where these user journeys have been a four foot piece of paper put up around offices for a company globally. How do you really start to make this part of the company's culture? Then finally, the highlight at the bottom, I won't walk through all of this, but what you'll see here are the discrete steps that someone may go through on a journey, and it works to act as an example of how someone could think through this. So you might identify initial business problem, acquire licenses, and then you're gonna walk through each of these steps. Right? So you might look for reusable data products in your catalog and then start to develop, how do I do my data engineering? How do I do my data science and eventually build my data product? And then finally automate this to production. But the important part here is having these very discreet and clear steps with documentation that makes sense for users. It can have a very big benefit on how you actually encourage them to continue to focus, use it, and produce better outcomes. I think one thing I'll add here is that along the way, you may have different license types. And so some users might have, like, a developer or data science license where they can build out solutions. Other users might just be explorers or business users. Well, they'll come in, and they'll just consume the results of what a designer or data scientist or developer has built out. So for them, their route is just, I'm not necessarily a developer in the application or the platform. I need to just come in and just access the dashboard, or I just need to come in and just export, you know, the end result of a model that has been running. And so just be cognizant as well that as you're designing the workflow for your user journey, you're being cognizant of the different license types and the different types of users because business users may have a different workflow, versus the people that are actually responsible for designing and developing the solution. Alright. So now we've talked a little bit more about our user journey, our documentation. Another key concept, and I'd argue one of the most key concepts, is this idea of communication. So So as you build out a program of self-service, one of the biggest roles from a CUE perspective is to make sure the organization feels a consistent update and energy around the program. Right? And I don't know how many of you worked in larger organizations that are here on the call today, but one thing that I tend to see is that you'll have a flash in the pan. You'll have a new idea, bad idea will show up. Maybe you don't get a lot of traction because it's not talking about a lot, and the business has a hard time believing this will be a reality. How do they know this isn't gonna change six months or a year from now? Well, a lot of times that comes in the form of communication. Consistent and constant communication really proves this idea that this is something we're investing in, something we care about. There's activity happening. It may be something that you wanna check out too as a prospective user. So what you'll see here on the screen is a series of various forms of communication that are recommended and the kinds of percent as a frequency we would recommend. And generally speaking, you're wanna gonna focus on the higher priority items being the more frequent items and the ones with the lower frequency or lower priority. And not every organization will use every one of these, so think about this as a bit of a choose your adventure. Focus on the ones that'll make the biggest impact for you. But to cover some examples that we would recommend, first and foremost is open office hours. You may remember this from our third webinar, we talked about the idea of support and how that changes in self-service. Open office hours are a big part of that. You wanna have a space that's available where individuals can come in and ask questions they have and do it in a live space where you can get help from others on the call. Right? Whether that's a technical team member in the CVV or maybe, ideally, maybe another business user, another self-service user could help train those other users. But ultimately, having this open environment of questions can be asked live and maybe in person or at least virtual format can be really valuable for certain to develop that community and that crosstalk in the organization. Similarly, you wanna make sure you continue to do things like user shadowing. Right? So you have many users across the organization, different parts, different divisions, and they will have different personas of work they're doing, tying back to what Christie just mentioned regarding the different types of users from the user journey. So it's important that you're actually shadowing those users and following them on a regular basis to understand the challenges they see and then really roll that back into how you build the program. Some other items you wanna look at is your Wiki. Right? So if you're looking to release new content, release new documentation, it's recommended that you do that biweekly or on a fortnightly basis. That way, you've got consistent documentation updated, potentially align that with scrum if you're doing two week sprints. Anything related to governance, similarly, you're gonna wanna update that pretty frequently. That way, users are not confused when you maybe made a decision or a change in their user journey based on the governance policy. Tied close to that is also newsletters, and I actually think these are probably some of the most key elements of communication. These newsletters can go out to your broader user base, existing as well as prospective users, and they can explain updates in the program. They can talk about what new features and functionality have been introduced. They can highlight various projects that have been successful or great examples to show the rest of the organization. They can talk about reusable data artifacts. There's a lot of things you can include there, but what's important is that there's consistent communication going out, showing there's activity happening within the organization. On more of a monthly basis, things you wanna think about is, you know, the data active homepage itself. You could update and make changes to that page to direct users to various content or information. You're gonna wanna make sure to highlight that, especially if there's specific things you wanna drive users to immediately when they log in to Dataiku. You know, if you have enterprise data catalog, similarly, you wanna update that on a monthly basis to make sure users know know more information on what's going on in that catalog. If you have ambassador meetings that we discussed here a little bit earlier today, it's recommended to do that monthly to really have that touch base with ambassadors across the business, both for feedback as well as, you know, sharing new updates. There's this idea of an innovation funnel. So as you have more and more self-service projects and they grow in their utilization and value, you may wanna be talking with business leadership or CIO leadership regularly to understand, do we take these self-service projects and move them into a broader portfolio? Do they need more support? How do you start to build those cadence? I think that's very important for building a connection to your more executive leadership, especially on the IT side of the house. Another consideration is also data councils. So if you have data counselors and data stewards across your business managing datasets, it's recommended to have monthly meetings to help coordinate that. On a quarterly basis, you wanna make sure that you're reviewing your wins, values, and metrics. Now as you can kinda tell, to us, it's very important that we talk about the value that's coming out of this, and we talk about the metrics because we really manage and measure through data. Right? So it's important to make sure you keep your head in the swivel and focus on that and use that to make more decisions. As Christy kinda mentioned, you can also have a fact page. I think it's important to update that fact page probably about every quarter or so to ensure that it's relevant to new users onboarding to the platform. And, again, welcome communications. You should have a communication that goes out when you have new users stored on the platform. Says, hey. We're glad to have you here. Here are your first resources and next steps you should take. And then finally, from a biannual standpoint, we recommend updating training probably biannually, whether it's live recorded. It takes a lot of time to update training, so we recommend doing that a little less frequently just because of the effort going into it. And then finally, if you're not already doing it more frequently, we do recommend doing roadshows across the business to IT and business leadership as well as potential users, annual basis so they understand the actual projects and work that are going on. So additional thoughts I wanna, add here is that when it comes to communication, of course, you're gonna have frequent daily interaction with the users. But you also want to have, interaction with their leaders, the core stakeholders as well. So when you're having quarterly calls, it usually is the leader and the stakeholder. It's very easy to onboard a team when you have the leader involved. When the leader understands how many individuals they directly lead are using your application that have a license. Getting to understand what are the leaders' goals at a quarterly basis, you know, annually. What are some of the project initiatives that that they're trying to build out? And that just helps you in the daily interaction with their users. So in addition to having that communication with the leader, and that can, you know, vary based on the project. Sometimes when I've ran COE initiatives, I've had project based support where a team maybe said that they wanted help really scaling a large scale project. And in that instance, I met more frequently with them than I might have would have if it was traditional office hours. One thing to also be cognizant is that when you're working with users and they're relying on you as a COE practice, it kinda takes on more of a mentorship model. Especially the closer, the more engaged your user community becomes, they rely on you, the the COE team, as experts. And so everyone has a different learning style. Office hours might work for some users. Other users might benefit from more one on one, learning sessions. So when you get into learning, you know, there's more than five different ways in which which people like to learn or can absorb information. Some people are more oral. You can do just a demonstration, and they they got it. You can have an office hour or a quarterly call, and they understand. Some users may prefer communication that's written, and that's where it goes back to the documentation. And these are users that are kind of like, oh, do you have links to the docs? They don't necessarily need to meet with you face to face. They just really want documentation provided on what they're trying to do. Then you have other users that are more visual. These are the users where they prefer more screenshots or if you have to send instructions, if you can show them step by step through some kind of graphical interface how to do something. They they are more receptive to the visual. And then you have kinesthetic learners, and these are learners where they want access to the template. They also want to kind of do the hands on with you. So that's where you might have more of a workshop format with these users. They maybe don't wanna learn just reading documentation. They wanna be in the actual setting where they can, follow along with your examples and ask questions along the way. And then you'll have a multimodial learner where they want a combination of different styles. For some things, they may want, you know, just watch a recorded video, how to how to do certain things. Other things, they're fine with reading the documentation. So keep in mind that when it comes to communication, one format is not gonna work when you're dealing with multiple users. You will have to adjust the communication style and cater to how everyone else learns. And so just wanted to highlight that as well. Great. And, actually, we've got some questions here. So we'll pause for a few minutes and kinda walk through some of these. So one question we have here is when a user creates something valuable but not quite reusable, how do you coach revolve that asset towards reuse? No. It's an excellent question. So I'll kinda point you first what we have currently on the slide here, the innovation funnel here on the bottom left. Part of the reason we have that communication mechanism is to really ensure we're having conversations around the pipeline. Right? So we're saying there's new self-service projects that are being built, and these may be getting more important. They may be becoming more valuable. Right? So common occurrence could be, hey. Particularly user in self-service build something. Maybe that user got sick for a week. That project started to fail, and then we realized it impacted the business operation. Was just speaking with the customer this week where they actually had, the delivery trucks did not go out because of the exact situation with a different platform, so related to self-service. So with that in mind, I think it's really important to keep an eye on what is the operational stability of these items being built and to ensure you have the right staffing and control around them. And I think that also applies to the idea of reuse. Right? Because if you're something that's reusable, you wanna make sure you're sharing it more of the organization. So to me, once you've kinda got that baseline, if we're observing, say, a new project by a self-service user, and that project's now got over, you know, 50 unique users a month. We've got a lot of adoption around it. Then the question may be, hey. How do we further evolve this? Or maybe the word may be hard net or operationalize it further. And that could happen in a few ways. Right? If we decide this is something that maybe is reusable but also something that needs a little bit more operational support, then maybe we're gonna move that into an IT or data team to manage. Another option may be, hey. This is really high value, but maybe it just needs some more work to make it better. Maybe in that situation, you have folks off your team. So, like, from a data crew perspective, maybe the DSIR on the account, you know, your subject matter expert working in data and analytics can maybe work with that individual and help them take that project further. So how do they take that more of that pilot phase, make it a full product, and make sure it's something we're usable to highlight to the organization. And I think, importantly, you wanna make sure that with that, that you're also communicating that reuse as the slide says. It's really hard to drive reuse. That communication is really important to that. Christy, anything else to add to that one before I jump to next question? Sure. I think the only tip that I I would, add is that I came across this several months ago supporting an organization where reusability was not really there. It was more just static projects. And I found that, my enterprise needed to be trained on how to work with data pipeline automation, and how to embed those techniques to make sure that they are building reusable pipelines. So I actually had to build a whole entire workshop. I had to build two. I had to build one for coders and one for non coders, you know, just teaching them how to automate. You know, if you're leveraging a tool like Dataiku where you're having, you know, scenarios and how to trigger, how to think about, you know, okay. You don't have to necessarily send this out manually. You can automate it. You can automate the whole project and have email notifications go out. So sometimes that part might be a training and making users aware. I found in my years of experience that when it comes to data scientists or data analysts, they may not be as train robustly trained on pipeline automation as, like, a DevOps engineer or data engineer or automation is second nature, building reusable pipelines, reusable workflows. So that might actually be a training or workshop that you might have to build out or documentation on that, getting them to think about how they can automate the various stages within their workflow. Great. Great. Awesome. So jumping on to another question we have here. So what surprised you most and how users respond to your communication strategies? What works better than expected? So for my own personal career, something that surprised me early on in working in this space was that when I was sending out communication, especially as related to governance, sometimes the communication was too direct on here's what we should just directly do and didn't really provide the reasoning. Right? And at one point, someone who was a colleague of mine, different department, within our engineering group, had reached out to me and said, you know, when you send these, it almost feels like you're up in the kinda ivory tower here telling everybody what to do but not really living on the ground level. And, honestly, that piece of feedback stuck with me. It's been almost a decade now. And what I realized from that is that you really have to be focused on providing the context and putting a bit of personal touch in communication. Right? At the end of the day, we're all people. We're all trying to achieve good outcomes for an organization. So for me, it taught me that I need to be a lot more focused on that element of it, explaining why we're doing things, explaining why that matters. So as an example, if I'm communicating governance policies now, I'm gonna talk about here's why we're doing this. Here's how it's better for the collective organization, and why we care about from an abatement of risk. I'm not just gonna say, thou shalt go do something. And I think that really applies across a lot of this. If you can do a little extra work to make something a bit more personal, to explain more of the detail, and they really connect the reasoning into something that individual or group will understand, I find communication resonates much more strongly. So, again, that's kinda why I focus a lot on this topic. That good communication is really foundational to how you build out a community within an organization, and it requires a bit more of that personal detailed touch. Christy, same question here. Would you add anything to that? What I have found and this is also, I think, at the professional level as well. So when I'm working with, data science or AI, ML engineers, you know, in their professions, you know, it's predominantly they have bachelor's degrees. But the higher up you go where if someone has a master's degree so I've supported a whole data science teams where everyone had a PhD. And so what I found is that, they asked a lot more signature questions. So sometimes my traditional documentation or, you know, just showing a demo wasn't necessarily gonna be as effective because they really want to get in there. They were more kinesthetic, and they really wanted to, ask questions on their specific use cases. So I just find that the different levels and if people are working on more original projects, that's where they may need more help and more assistance. And so that was one of the things that actually surprised me. So whenever I would do a workshop, if it was a newer user, then I could kinda follow, like, along with my template guides that I had. And a new user you knew on their analytics journey, they can just follow along with the template guides, and we can just, you know, go through demos and it was a little bit more easier. Whenever I dealt with individuals that were used to doing original work, original research, individuals that want to build custom algorithms, they had the most amazing, like, innovative deployment patterns. And so this was where I really, really had to spend a lot of workshop more kinesthetic and dive deep deeper with them. So the communication and the interaction will vary depending on, the level of and maturity level of the analytics user as well. So I just found in my experience, the more advanced a user is, the more you're gonna get more technical, deep dive, original type of inquiries, and types of support that you've never even thought of to write in your documentation, and you will need to have more of a deeper close connection interaction with them because you're really helping them scale original work. And that's that's what the field of really, you know, data science and AI is right now. A lot of times, they are tasked with scaling original work. So your interaction might be a little bit more in-depth and more involved with those users. I remember you were talking on mute. I was saying really well said, Christy. I couldn't agree more of that. One more question here, and then we'll keep kinda deep diving for the rest of the slides. So this one's from Koichi, who I believe is attend every webinar. So, hi, Koichi. Thanks for joining again. So, another question here. How do you manage the risk of becoming a victim of your own success or rate of uptake and interest over on CV's capacity? So the question is when you have a problem, but it's a good problem because you scale enough, you've been successful, how do you manage that? To me, it kinda comes out in two ways. So first would be actually what this webinar is focusing on, the idea of community, the idea of reaching out across the business. I'm a very big fan of having the business take a very large part in building these programs and ultimately helping drive them forward. What you'll find traditionally is that if you have a really highly leveraged COE with a lot of people in it, that can actually slow things down because it's so centrally focused. You wanna have the right number of people that then can work across the business and amplify the effect across business teams. So just as an example, in my former career, most of the time that I worked on these programs from a business perspective, about 80% or so of the folks who are doing the training and the outreach are actually from the business side, but maybe 20% coming from the data and ITCOE team. And that's because those business team members have bought into the idea into this idea of democratization and empowering the business, and they wanna help drive that forward. So to me, that's an ideal state because now the business is working with you and partnering together to make this happen, and all the work falls to a smaller group of individuals on the COE. The other element I would add, and this is not something we're necessarily focusing on this series, is the idea of using more automation to scale further in your operations so that you have less and less manual labor to do, especially in the age of AI, especially with tools like Dataiku. You can automate many of the processes that CVT members spend manual time working on. And through that automation, you can also scale further because you spend less and less time working on those manual items and more time working with the business. So to me, it's both the idea of stretching out your community and leveraging the business to help you scale effectively, but also looking how you can further automate and reduce the amount of manual processes the COE team works on. Now, Christie, same as the other ones. Anything to add here? Sure. So I actually encountered this problem in my last organization. So I had a very tiny COE practice, and then we had more than 1,200 users, and And we were constantly growing and extending licenses. And so one of the first things, how the COE is structured, I know this is gonna vary, but, our COE was built within the platform team that was supporting our Dataiku application. And that allowed us to flex up with more support. So sometimes you can build your COE as a stand alone from the if especially if a COE that's tasked with supporting an application, you have two options. You can build them as a stand alone team that doesn't have direct connection with the underlying platform team, which you imagine, you know, your developers that support your organization, or you can build them under one team. When you build them under one team, now you can flex up resources. So if me and the other members that were allocated for the COE, if we were, you know, unavailable or doing a workshop, then we would have additional tickets or support picked up by the platform team because they know the platform just as well as the COE would because they support it. So you have flexibility there with how you structure it. Another thing that you wanna do is you wanna have community channels and chats, and this was so essential. So we had a, community chat. We use Teams or whatever, you know, software that you wanna use, and everyone that had a dedicated license was on that chat. We would encourage that if users needed support or had questions, they would post a question in the chat. And so many tickets were resolved by other users. Our champion users would sometimes solve the increase before we got a chance to, and that helped offset the burden for us significantly. So, you know, how you structure the COE and definitely having a community chat. Now, one additional area that, like, for me, I was busy scaling web app development across many different teams, and it was a bit more of extensive. And it it kinda became a little bit difficult because I had so many teams ramping up and one developed like web apps at the same time. I just held a symposium and brought them all together. So a lot of times that helps. So we would just have, biweekly workshops where I call it web app symposium. And it doesn't have to be about web apps. If you have quite a few users that are working with, you know, some type of models or GenAI solutions, but they may be across different teams and you just start seeing where, you know, the influx of increase are pouring in. It may be advantageous to have a separate workshop for that whole group and you bring them all together. And that way, you're tackling their concerns within that same block of time and you're giving them the opportunity where other members and other champion users can help them out. So leveraging the champion network and creating opportunities where users aren't just coming to the COE, they can also rely on each other and providing those avenues for them. That is very helpful when you're dealing with a tiny team and and you're you're constantly ramping up support that way. Great. Great. Awesome. Appreciate you all for the excellent questions. We're gonna come back to the, slide content here a little bit. So now we've talked quite a bit about communication, building community, things like that. Another key element that comes up is this idea of implementing train the trainer, which actually ties really well to Koichi's last question and also to what Christy was just mentioning, which is this idea that when we think about training and upsealing the organization, generally, we tend to learn better from those who have a similar work experience to us. Those who have more similarities. Right? And that's because we know that they know what our job function is, the type of work we do, etcetera. So the idea behind train the trainers to have those who are actually in the business, those with that expertise in those business units to actually work with the business and doing that training itself and actually execute that training. Now there's a number of prerequisites to be able to do this. First, a lot of things we've talked about so far in the series, so establishing a vision, a branded mental model, user journey, documentation, percentage communication. You need to have all these things in place as a baseline because they build the foundation to enable this. They create all the artifacts, all the material, the educational learning you need to really start democratizing out the actual training and really federating that out a little bit. You also need to make sure that your internal team has really the right knowledge around it. So if we go back to the idea of what a team structure looks like from our third webinar, you need to make sure you have the person who's your product owner. You need to make sure you have your applied SME, right, and your data or your platform architect. Those roles need to be in place, and those individuals need to be upskilled enough that they can work with the business to handle additional capacity. Importantly, you need to make sure you build strong enough trust. So going back to the first slide of this presentation, build enough trust with the ambassadors and champions in the business so that you can actually do this effectively. And then finally, wanna make sure that you really document your nonnegotiable. So when we talk about governance and things like that, one risk of federating this training out more towards the business is that those business users may not be as attuned with what's happening from a government's perspective or some of the decisions you've made to protect the environment. So you need to make sure you really know what those items are and actually really are clear with that with your ambassadors and those in the business to make sure they reinforce that message. Now when you go through to build this training, there's some considerations. Right? So first and foremost, again, going back to the branded mental model, you wanna make sure you align your training curriculum through that. Once you build in your training, it should follow your documentation structure, it should follow your branded mental model. That way, users have a very clear way of what they need to walk through just like the user journey. Again, you wanna document that user journey and make sure you have the documentation around that to help ensure it's enforced and supported because, again, these teams will be training without you being present. You also wanna think about the levels of training. Right? I've seen many different forms of this. Some organizations do it like college courses, 100, 203 level type stuff. Some organizations have executive level training. But you wanna think about what is specific kind of training you wanna offer and does that change per personas, business divisions, etcetera. Now can your training be coordinated with your learning development team? It's very common to have a COE team trying to run this fully independently. I do think it can be more successful if you look at the leadership and development team of the organization and look to plug it directly into training for the company. If you make that more of a requirement or something that aligns with that, it'll be more ingrained in business process for the organization. The training curriculum itself should actually be designed collaboratively. You should look to define the framework and the structure, but you look for those folks as ambassadors as they'll be delivering that training. You should look for them to help complete that content with you. That way, you jointly work out together and it's being effectively done. Importantly, the type of training that you use for a use case is important. If you can support those users bringing in their own individual use cases to walk through, that's best. Next would be, you know, divisionalized use case, the industry, the generic. But the idea is to be specific. The more you can put the training directly to what that user's gonna deal with day to day, you wanna focus on that. Then finally, you gotta define the operating model. How will you update training content? Who's gonna assign the training sessions? Who's gonna do the scheduling? All those pieces that go into it. And then finally, jump to the right side here a little bit. I think there's kind of three kinda steps to this joint approach. First is building a relationship, increasing trust, which is really the idea of building trust with your ambassadors and your power users. From there, you wanna define the framework. That framework should be built on that branded mental model on user journey, the structure of what you wanna provide. And then you take that and you develop the content in a codeveloped way. That way, it's relevant to the business. And then finally, the COE team should focus on the strategy and the scheduling while the local training is done more by the business teams and ambassadors in the business. Christy, anything to add on this one? Oh, I think you summed it up perfectly. So, yeah, no additional thoughts there. K. Awesome. Keep scrolling. So, again, kinda tying back to some of our conversations we've already had today around reuse. Right? So the idea of reuse, I think most of us get it, is that the more you reuse, you can shorten your development cycle times, you can drive more consistency across synchronization, but it's also really hard to do. As I've said for many years, if a financial leader has 10 analysts to go develop a KPI, those 10 analysts will go out and build 10 KPIs of the same name, but the logic is different. So what you really wanna focus on is identifying the right out of those 10 KPIs, which one's the right one, and then make that the standard across the organization. So that's really what we're talking about here. How do you drive more reusability? There's a lot of different ways to think about reuse, but on the left side here, you see types of reusable data artifacts. There are many from the Dataiku perspective. Important one here would be your best practice methodology, so how do teams actually approach working in the software. But you also have items such as your projects, your pipelines, your agents and LLMs. All these elements that are part of Dataiku are also reusable in the software. And this is very powerful because you can use this to drive more reuse as users in their flow and doing their work day to day, and Dataiku has a lot of built in functionality to help support that. Similarly, within the data ecosystem, you may have additional things that you wanna focus on. So, again, you may have those best practice methodologies, data products, datasets, but largely, it's gonna be similar items to what you may see within Dataiku, just more broadly spread across, say, your enterprise data lake or lake house environment. Now there's a few different methods to encourage that reuse. The first one we've already talked about a bit is training, making sure you're reinforcing throughout the entire process of training that it's a reuse approach is pretty key. You wanna make sure you're measuring. So this idea of measuring the consumption of data products and then using that to drive reuse in the organization. You will also talk about how you're gonna make decisions. So how you decide if something's reusable? Is that through data stewards or a decision committee? How does that process actually happen? It is recommended that includes functional businesses. It shouldn't just be an IT or data team centric decision. Again, Again, you wanna make sure you communicate, seizing some of the communication strategies from the previous slide or previous two slides ago. How do we ensure that we are communicating these high quality data artifacts to drive reuse? And then finally, how do you gamify this? So how do you gamify the reuse of those data artifacts? And then finally, where can you share these? The Dataiku data catalog itself is an excellent place. You can reshare a lot of this within that environment. You can also use your own Wiki or documentation software to do this. Other enterprise data catalogs such as Elation or Calibra can be used for part of this reuse type approach. Again, through training, you can help align that. And then finally, you can also do that through sharing forms or symposiums that focus on this kind of reuse and best practice approach. K. Awesome. Finally, jumping on to the account, the next topic here is the idea of, gamification. So if you're not familiar with this idea, it's the idea of taking something elements of gaming and applying it to a process within a business or other processes to help encourage more direct engagement with that process, right, to make it more fun and make it more competitive. And here I kinda start with a quote, from Vietnam game researcher Jane McGonigal. When we play a game, we tackle tough challenges with more creativity, more determination, more optimism, and we're more likely to reach out to others for help. So that's kind of the idea. Right? If we can gamify the use of data analytics and working on self-service, we can help drive all this more determination, this more creativity, and actually more collaboration as well. Now when we think about the application identification, right, is that's really to increase in engagement, resiliency, and enjoyment. So how do we really do that for a user? They can also help introduce this data flow as we talked about from the user journey example. By gamifying, you can make it a little bit more prone to actually focus on what that journey looks like. And then finally, and I think this is very important, if you're gonna apply gamification and make sure it's natural to the activity, you don't want that process to feel unnatural like an extra ad because that's actually just going probably annoy users. You wanna focus a bit more on something that feels directly aligned with the work they're doing, and it really just further encourages more of that. So from an approach standpoint, the first thing you wanna do is focus on one of the things you want to incur. It'll be a little more focused on encouraging those users to do. That could be putting more data into your data catalog. That could be creating more projects within Dataiku. How do you think about what those key elements are that you wanna encourage? Now you wanna make sure you collect the metadata around the use of those items. So if I'm collecting from Dataiku, I I may wanna pull up my future utilization data around creating new datasets using a no code recipe, using a code recipe, and assign point scales to that. That way I can understand, hey. This user did this much work. This is how many points they're gonna get. Right? From there, you can then create a leaderboard that links to that. So a simple visualization can do this directly in Dataiku. But that reads in that metadata and says, okay. These users have done this much work, and here's what they're showing up like a bar chart. Right? So one's competing against each other. Importantly, I think it's probably the most fun part. Implement awards, have stickers, have things you give people. So as they level up, as they do better in gamification, they actually get little things that they can show off about their effort. And then finally, make sure that you continue to communicate and reshare about this, talk about it in the newsletters, make it part of the overall community engagement. On the right side here, I'll give a simple example of, from my own experience, from my time when I was at GE Aerospace. Here we had a program called self-service data dual that focused on this idea of how do you enable self-service across the organization just like we're talking about here today. And what we did is we tracked activity in both data accumulation, two of our three self-service tools, and look to use that to launch the activities and the value that we want to drive. We had leaderboards across the organization, have many fond memories of, various users competing out who could be best on those leaderboards. And we also created levels that those could actually level up into. And then some fun way we actually did this on eighties themes. You'll see in the very bottom right here, could have a small image, but we actually had various stickers based on levels you got to. So you'd start as a basic level, then maybe you become a pilot because now it's really the top gun, and maybe you become a Terminator, right, as you further go up the scale with the final one being a Jedi. But the idea being we made it fun. What was cool about that is that as you looked across the organization, internationally, locally, you'd see many employees walking off these stickers on their laptop. That to me was a great visual indicator of traction and adoption and interest. Now as part of what doing this work, we also focus on a specific competition around metadata. So we're able to get millions and millions of metadata tags done through a competition like this. And then finally, regularly communicate and have newsletters and awards tied to the program. But ultimately, it was a great way to drive further engagement. I've seen many other customers now do this since I joined you at Yahoo a couple years ago. I think it's a very powerful way to really drive more engagement across organization and community. So oh, yeah. Go ahead, Christie. Oh, I just wanna add in. So at my previous organization, we had annually, we had a create a thon event where we allowed users to build out creative, solutions. So we gave them a theme. Of course, there was a theme for GenAI. There was a theme for machine learning. And then sometimes it was a business challenge that the organization had. And then we had a whole week where, users were encouraged to build out their most innovative solution within our platforms, and then we had first, second, and third winners across different categories. And it was a really great way because we had a lot of teams from different departments working cross functionally, just trying to build out and solve the use case. So if at the enterprise level as well, if there are some organizational challenges, some data related issues, if you have, like, a theme event where there are incentivized awards, and we actually gave out actual cash prizes and really, really nice gifts based on the winners of the actual activities. So that's a great way of encouraging, engagement within the platform, cross functional team collaboration, and then kind of incentivizing users to use the platform in a different way than your traditional work. And that's exactly what these competitions were. So annually, we would hold them. It would be one whole month. We would, advertise them and advertise. Here's the categories and the use cases that we're hoping that you can solve with data, and you can be as creative as you want. And that way, you know, it really got the platform more visibility and also help with onboarding more users that maybe didn't have a license. Now they're watching the winners, you know, win these awards and also seeing what the platform can do and how far you just can stretch your innovation. So there's all kinds of creativity that you can, embrace as well from an enterprise level to kind of get more visibility back into onto the platforms that you're supporting. Yeah. Awesome. And with that, Christy, it's back over to you. Sure. So this is an actual use case example. So I've supported the Dataiku platform across several different organizations. And so this is an example from, one of the financial service provider that I supported the platform, which had more than 1,200 users, And I had mentioned that before. And so coming in, the platform was adopted for several years, but it was a heavily customized platform. So some of the out of the box solutions that users would normally just leverage, you know, the Dataiku Academy or some of the documentation to be able to implement their workloads were it was very different. And, you know, as I mentioned earlier about with the SML platforms, every enterprise gets to customize how they want users to use it. And so there was a little bit of a disconnect between the way that IT implemented the application versus the way the business needed to use it. And so the engagement level had declined significantly, and there were more than 20 different teams that were using the platform but were using it differently. So the first thing when I came in is I documented the user journey, at the individual team level and then also identify different types of personas. So I was able to identify that there were seven different types of user personas using the platform from no to low code users to, developers wanting to scale enterprise financial web applications and have them natively hosted within the platform. And so what you see here is you see, you know, the actual support tiers. A tier you see here is my, windmill of support that I provided. So starting out in the existing state, you have about nine different services. So these are some of the things that we've talked about throughout the slides, self-service support, office hours. But because web development was important, for this organization, I had a whole entire different tier of support around web apps. Different organizations, it might be maybe the platform is used for GenAI use cases. Maybe it's used for ML ops. But whatever that targeted level support is needed, that's what you wanna, zone in on. And I mentioned before that, you know, after, you know, supporting users, there was some issues with how do they create reusable pipelines. And so provide a lot of support with helping them with automating their pipelines. Now this organization also work with big data. So we had to do cost optimization workshops and really make sure that it wasn't just about teams building reusable workflows. It was about are they building cost efficient workflows that can easily scale and be more efficient for cloud deployment. So you can kind of see that you wanna tailor this according to how the enterprises and the various teams expect to utilize the support. And when we look at the emerging state, as this organization, we were pushing more towards leveraging more generic use cases, then you can kinda see some additional services that I was starting to add. So as a COE practice, you definitely want to, identify what services are needed to support different personas or different teams and also think about your future state. And so what you're seeing here in the future state as we were identifying more AI use cases working collaboratively with the enterprise, with core stakeholders, we started building out different ways where we knew we needed to support the team. If it was, you know, how do we help teams get started adding chatbots to some of the web apps they've already hosted in the platform? That was a support service that we identified that was needed. You know, how do we help teams with, pretty much being able to think about how they can leverage AI? And so that's where we work with teams to build out AI ideation canvases. So this is just an example of how you want to be able to identify the different levels of support, and then this will also help with tracking. So just wanted to highlight this example from a previous organization where I stood up and scaled the COE practice there. Thank you. Awesome. Thank you, Christy, for the highlight. Well, everyone, thank you for joining in today. We're getting in just under time, being 11:59 here eastern. That said, we'll hold on for a few minutes here. If there's any additional questions, please feel free to add them to the chat. But thank you for taking the time to join in here today, and we appreciate it. If you wanna check out our, next webinar or the full webinar series, please be sure to scan the QR code that you hear see here up on the screen.