​@deloitte Shares How to Avoid Migration Traps | Tableau Conference 2026

a lot of these small things that we did to be successful with our migration. Full disclosure, we're still in our migration as we speak, but I'll talk a little bit more about that in a bit. So first, I'm going to discuss the motivation and the scope. So why did we decide to migrate over to cloud? And also the scope of our size, and what you're gonna learn is that's one of the things that made our migration unique. So really there's 3 things when we talk about the why we're doing this. One was around the governance side of things. And when I talk about governance, we were spending a lot of time governing our platform, our platform OSs, our platform, just keeping the systems up and running. Instead of doing that, we're more interested in governing the data that's going into our environment, the content that's going and being displayed in our environments, as well as the users that are using our environment. So the governance side of things was a really important thing for us. And it's not about, it's not about not doing governance in cloud. You still have to do your governance. It's just we can focus, we can change that focus on what we're governing, and we can also spend more time training our users, our architects into worrying more about becoming thought leaders, and knowledge architects, and decision architects. The innovation and AI side of things, that's really about getting the latest, greatest tools in the hands of our users. It was an introduction for many of our users to get into Tableau Pulse and Tableau Pulse AI. We gave them that little taste of AI that comes with Tableau Pulse. And that's something that, again, as we go forward, one of the things Moritz is going to talk about is what's going to— this alone is going to help us lead into getting into the things like Tableau Next going forward. Scalability. So you all know the scalability story, right? Tableau Cloud, elastic scalability. It can grow with your user base as that grows. That was obviously really important to us. But in our current server environment, we also limited our users from using things like extensions., right? Because we were always a little bit concerned that those extensions would cause server issues, and in order for us to govern that, we would have to add more resources. Okay, as far as what we are migrating. So the thing that makes us unique is we have a very large production environment, 10 nodes, 256-gig servers, 16-core, 32 virtual CPUs. And it's not just, that's the one production environment. So we have the same environment for stage, we have the same environment for our disaster recovery, and then we have a similar environment for QA and dev. So we're talking close to 40 nodes sitting out there that we maintain today. So we're an extremely large environment. We have roughly 65,000 users. That use our environment. That's our kind of our rolling 6-month average is 65,000. These are all internal Deloitte employees using this environment. 1,200 Tableau projects, and that's just a top-level project. So we probably have closer to 10,000 Tableau projects when we started this migration. And again, 15,000 workbooks. Our transition In order to help automate part of the migration, we did use some AI vibe coding to help in that process, because again, we're a little bit unique, and Praveen's gonna talk a little bit more about that. So that's the why. Praveen's gonna talk about the how, and—

Thank you, David.

This one, right?

Green button. Yeah, thank you. I'm gonna share you the practical lens of migrating the environment from server to cloud, starting with the cleanup activity, understand your technical constraints, move in waves, and have the governance in place so that the cloud environment stays healthy over the time. Cleanup. Cleanup provides the, reduces the clutter on the environment and make sure that migrations are happening smoother as you expected. Coming to the cleanup, you start with the user cleanup. Remove your inactive users, redundant users, any outdated license users. Identify the content who doesn't have any ownership to it. Moving to the audit and the content cleanup, identify the workbooks which can be archived or removed, which are unused in the environment. Have some kind of a standardization in place in, in terms of the naming conventions. Consolidate all your duplicate projects, have assignment or have a clear ownership on all your content including the data source, workbooks, and the flows. Take the backup of your content while you're doing deletion process. Identify your content dependencies. Once you have the governance process in place on your Tableau application environment, make sure you have audits are in place. It could be with your PostgreSQL databases or it could be Admin Insights. At this point, you know what content to be migrated, what content to be cleaned. You know what your data sources are and what your databases are. Will your Tableau Cloud able to connect your data sources sitting on your on-premise? No, not directly. So we, we introduce Tableau Bridges. What does the Tableau Bridge do is connect your on-premise databases, whether it could be, you know, residing on your local tenants, or maybe if you have databases having Kerberos or Windows authentication or connecting to file systems. In short, if you have data sources in the private network which doesn't have public access, then you need a Tableau Bridge. Where does this Tableau Bridge reside? Our recommendation is to have the Tableau Bridge reside near to your databases. For example, if you have databases on your local data centers, local, then you have a standup Tableau Bridge stored there. If you have databases residing on AWS environment, then you have a Tableau Bridge on an EC2 instance. How many bridges that you want? It all depends on how many live versus the extracts that you have it on your environment that runs concurrently. And how do you want to build those environment? What's the sizing of those environments? Start with the minimum basic requirements that the Tableau recommends and scale up depending on your usability. Firewall planning. So our Tableau Cloud doesn't connect to our local databases. Make sure you have your IP addresses in place. The firewall openings happen prior in advance before migrating the content. Since it is a locally hosted application, service, Tableau Bridge. Make sure you have automations are in place to patch the Tableau Bridges to the latest and greatest versions. Make sure you have monitoring in place as well. And few other technical considerations that you need to consider. Understand your live versus extract connections. Understand, check your time limitations, size limitations, and also supported connectors. Sometimes not all the server-hosted data connectors may not be compatible with the server, with the cloud. So make sure you have the compatibility check as well as you have some kind of performance-based lines to compare with the cloud. Once you have all this consideration, next to think about the user administration and users onboarding onto your application. We followed this SAML approach. We used Azure Entra and also with a SCIM process to sync up all your AD groups or onboard all your AD groups and the users into the Tableau Cloud environment. But at this stage on the Tableau Cloud, do we have this leverage or the luxury that we have on the, like we have on the Tableau Server? No, thing here, there is that not, synchronization delays. We do not, cannot perform any kind of ad hoc synchronization for an AD group on this during the sync process or the SAML setups. And if there is an email address change for a user, yeah, then the new email, the new user is not imported onto the new and the cloud environment. If there are any special characters onto your AD group, those are not supported. And also the nested AD groups, if you have any nested AD groups which are supporting on Tableau Server. Now it's not supported on the Tableau Cloud. Change management. Ensure you have the user impact assessment done for each and every project that you are kind of trying to migrate. User trainings are in place. Communication strategy and make sure your project leaders are given enough time for the migration and have the validations are in place. Can we migrate every everything to the Tableau Cloud. There are a few limitations that are constraints that we are migrating to the cloud. We cannot have custom views, subscriptions and alerts, refresh schedules, and any kind of the history that we have in a PostgreSQL, the usage history cannot be migrated. And any kind of web embeddings that we have it on the Tableau workbooks, that cannot be migrated along with any kind of unsupported data sources. At this point, you know what your content to be migrated and what your technical constraints are. Now we're in a position to know how we are going to migrate the content, right? Whether we're going to migrate the content with the Tableau-supported product content migration tool or build your own scripts where you can migrate, automate through an AI. Our recommendation is go through a web-based approach. The first wave, start with a simple project, less number of workbooks, a minimal user migration, security in place, and increase your complexity from moderate to complex in the next wave. During this process, you will have your challenges. There will be pushback from your project leaders and the business to have the migration in the future waves. Have all those projects part of your Wave 4. Every wave, end of every wave, make sure you have your project leader sign off. Document all your learnings in each wave. Track your metrics. It could be user usage or maybe performance. Decommission all your content that's been migrated to cloud on this— decommission on the server. And repeat this process for every wave. Thank you.

OK, and we're going to quickly just show the next Next slide here. Again, whether you choose to do this on your own or use a service provider, these are a lot of the questions you should be asking either yourself or obviously your potential service provider. These contacts, that's the three of us. Feel free to contact us. Track us down during the presentation or even afterwards. I'm gonna hand this over to Moritz now, and he's gonna talk a little bit about our, our AI use cases and where we plan on eventually bringing all of this. Moritz.

Perfect. Thanks. Thanks, David. And thanks, Praveen. So I hope you can hear me. So I'm responsible for our client-facing Tableau practice globally, and I'm super, super happy to see that we as one of the biggest Tableau clients are now moving finally to Tableau Cloud because this is really what I was looking for. Being part of our external-facing consultancy workforce for over 3 years now. And usually, I mean, my team is doing many, many cloud migrations for our clients, and quite often the objective of our clients is that the user doesn't feel any change compared to being on-prem, right? And maybe this resonates with some of you that you don't want actually that nobody actually feels any difference compared to the old world. And actually, this is something where I believe migrating to Tableau in the cloud is making a huge difference for your users because this is actually unlocking a lot of additional value for them. So I hope that you are not having the objective that nobody is actually experiencing any change over the past. And the second thing I would like to share with you, obviously, is that we are super keen to also support you in that way. So as we also internally have lots of experience, we would love— and that's a bit my, my marketing slot I would like to use here. So if you're interested in getting our support, please talk to us. But as David said, for us, it's also very important to talk a bit about the value which is actually unlocked by moving to the cloud. And we and the clients I'm working with, we usually look at 3 different dimensions. So one dimension is obviously getting more reach or getting more people using data-driven decision-making. And we still see that there's a huge gap out in most of our organizations because we see that usually you maybe have 20 to 30% of your users are using dashboards, using data-driven decision-making. And this is a huge gap which we all should aim for closing. And moving to the cloud, moving to new products like Tableau Pulse is a good way to actually close this gap and actually increase the reach and have more people using data-driven decision-making by using the capabilities of, of Pulse, for example, which unlocks new capabilities there. But the next and the second pillar for me is, is the trust layer. Quite often in the good old on-prem time, people have been using their own data sources. They have been uploading data from, from Excel or any other source, and this is not really something where there was a lot of trust involved. And having a more governed environment, which is like really top-notch in terms of the capabilities you can bring to your organization, I think this is again something which is unlocking a lot of value. And a lot of trust, which is a big theme specifically if we talk about using AI in the future where, I mean, there's lots of excitement, but if you get one answer right or if— sorry, if you get one answer wrong by asking your LLM, you will lose all the trust of your users. And this is actually something where the Tableau Cloud environments and everything which is coming with that is increasing the level of trust. And last, I mean, for me, the ultimate goal, of course, is to increase the value, to increase the benefit for the organization overall. And there again, I see that moving away from just pure static descriptive dashboards, but having the capabilities of using the modern possibilities of AI and conversational AI and agents is really increasing a lot of potential. But of course it's an effort which everybody needs to manage in the organizations to make sure that you really have a value case in place, right? We see quite often that there are a lot of crazy ideas. People want to do stuff because they can simply do it because the technology is there. But something we usually advise is to really look at what is the expected value of the topic you would like to achieve, then make a little cost-benefit analysis, make a little business case, show it to your business users, show it to your management, and then it's way easier to actually argue why you maybe need an additional investment in actually executing and implementing that beyond the pilot. Because I mean, we all see in many companies that doing a pilot, everybody's doing right now, right? But moving beyond pilots and bringing AI use case in production, is something which costs resources, which costs tokens, which costs additional capacity of you and your teams. So that's my last advice for today. Always think about the potential value and make that a great case to do what you want to do. So with that, I think we are— again, I would like to thank you for being with us tonight.

Yes, thank you.

And we open for questions.

Oh, oh.