Find the PPT here.
@Sandro Palleschi, Manager of Enterprise Data Services at Empire Life talks about the Business Benefits of the Reltio MDM Platform. Sandro covers the following:
1. High-level overview of Empire Life and how they got to Reltio
2. High-level overview of their implementation/data model
3. Review of how mastered data is being leveraged in their downstream applications/services
4. Review of how Reltio architecture has provided them with more of a modern and agile platform for mastering data
5. Planned enrichment and enhancements in their roadmap
Transcript here:
Chris Detzel (00:00:07):
Welcome everyone to another Reltio Community Show. My name is Chris Detzel. We do have special guests, Sandro Palleschi. Sandro, I really appreciate you coming in. He's from Empire Life, and the topic today is business benefits of the Reltio MDM platform. I had a chance on Friday to talk to Sandro and it's going to be really good. So, Sandro is a manager of enterprise data services. Sandro, here in a minute, I'll let you introduce yourself and everything else.
Chris Detzel (00:00:41):
Here's the rules of the show. Keep yourself unmute, use the chat to ask questions. I'll make sure to answer those questions or have Sandro answer those questions. I'll ask him from the chat and the Community Show will be recorded and posted to the Reltio Community in the next day or two. So, some upcoming Reltio Community Shows, today is Empire Life.
Chris Detzel (00:01:05):
Next week, we do have a show with Michael Burke called Evolution of the Data Quality Space, really excited to hear that. He's the director of ML and AI here at Reltio. We do have a new release coming out here in June. We'll talk about it on June 2nd. And then on the 23rd, we're doing something a little bit different. Chris Iloff, he's the Director of Customer Success. He's been in the MDM space and built MDM practices from ground up and who will be talking about some of those experiences in late June. I'm going to stop sharing my screen, Sandro, and let you share.
Sandro Palleschi (00:01:45):
Okay.
Chris Detzel (00:01:48):
Thanks, Michelle, for getting on video.
Sandro Palleschi (00:01:50):
That work? You guys can see my screen?
Chris Detzel (00:01:52):
Looks great.
Sandro Palleschi (00:01:53):
Super. Well, thanks for having me, everybody. I know that some folks at Empire Life that are also in the audience, so I'm sure they'll keep me on this as I go through this. Really quickly, the agenda I was going to follow was a brief introduction, although Chris has covered most of it already. I'm going to give you a bit of an MDM journey timeline that we've gone through. Reltio is not our first MDM platform. And I thought if I could give you some context as to what we went through with our first platform and to where we are today, it'll help understand some of the benefits that we're realizing just with the migration, as well as the MDM platform in general.
Sandro Palleschi (00:02:37):
Talk about those benefits of migrating to Reltio, a quick overview of our data model, a quick overview of our architecture for the MDM platform, bullet items on the business benefits of the platform. I'll give you a list of our current consumers of the mastered party data and what our future plans are for Reltio. Chris, you can just interject anytime someone has a question.
Chris Detzel (00:03:03):
Yup.
Sandro Palleschi (00:03:03):
And I'll take it then if that's okay.
Chris Detzel (00:03:05):
Absolutely.
Sandro Palleschi (00:03:06):
Super. I won't go through much detail here. This describes Empire Life. There is a link and I'll make the deck available to Chris for you folks. And there is an Empire Life story that's available at the Reltio site that goes through Empire Life and detailing a lot of the things that I'll be talking about today as well. So, I won't go through any detail here. As far as our MDM journey timeline, we started with Informatica on 9.6 back in June of 2013. It was an on-premise MDM platform. And it was the solution that was selected at the time with Oracle being the database back end. We had been running Informatica from June 2013. Up until July 2019, we had not done any upgrades and that got us into a lot of trouble.
Sandro Palleschi (00:04:00):
So, when we had to eventually migrate to 10.2, given the fact that we were no longer supported, it took us a long time, took us over a year to do the upgrade. And we actually had to get a third party to help us with that migration end. That was one of the areas where we thought we would never want to go through that again. So, at that point in time, we started looking at potential options as to how we would either deal with Informatica going forward or potentially look at another platform altogether. Our director had attended a conference. I can't remember which one it was when he met some representatives from Reltio. And that got things started. Matt Cook is our Director at Enterprise Data Services, who I report to, and I probably should have introduced myself.
Sandro Palleschi (00:04:49):
I'm Sandro Palleschi. Chris already did that, but yeah, I'm the manager of Enterprise Data Services. I report to Matt and we have the developers that report to me, which include now I would say one and a half MDM developers. We used to have up to three. In any event, we ended up with a POC with Reltio, like what we saw, and then we negotiated to move to Reltio at that point in time. We redid a whole data profiling, went through right from the beginning, relearned everything that we had done before, and got to the point where we ended up migrating the data from Reltio in March of 2021.
Sandro Palleschi (00:05:34):
So, at that point in time, as you can see in the graph, we ran Informatica and Reltio in parallel, where we would ingest our source system delta changes on a daily basis to both environments. And the reason why we did that was once we migrated the data, we had to then redevelop essentially all the downstream services of which there were approximately 15 applications. And we had to go through them one by one or some of them in parallel, go through that whole development effort, and introduce those applications and point them to Reltio as they became available.
Sandro Palleschi (00:06:09):
So, that took place between March 2021 up to December of 2021 when we cut over the last set of applications. At that point in time, we no longer had to run in Informatica. We did keep it running though, just in case we had to fall back for any reason, but everything worked out really well. We were pretty much done then in December of 2021, and then we were running Reltio by itself and then decommitted the Informatica master data management platform in January of this year.
Sandro Palleschi (00:06:44):
In terms of why we went to Reltio and the benefits that we accrued as a result of that work, we had to deliver a new data model. Obviously, the data model is a bit different in the Reltio world than it was in the relational world. So, we went through. We developed a new data model. And as a result, it required us to redo all and review all of our mappings in the way that we staged and ingested our data to conform to the new data model. Our internal resources when we were working on Informatica were very much in a bad position, because the solution was delivered by a third party. So, they pretty much had learn it on the fly. And we were often in react mode dealing with new business requirements and just managing the platform on premise.
Sandro Palleschi (00:07:33):
As a result, the migration really gave us a much better and complete understanding of the MDM delivery front to back. So, we got to relearn everything again and uncover a lot of design flaws with the original solution. We were attracted to the cloud native solution that Reltio provided, fully managed SaaS solution, obviously to lower our infrastructure costs and increase our agility to deliver business value. The other vendors that we had looked at, when we looked at their products, they were pretty much on-premise products that were reported to the cloud. Reltio was the only true cloud native MDM platform that we had looked at and ticked all the boxes for us.
Sandro Palleschi (00:08:19):
The high availability that's inherited in a SaaS solution was very important to us because we didn't have that with our existing platform and the cost of getting a high availability solution using Informatica and Oracle as our database was going to be very costly for us. And then ultimately, the whole API delivery of Reltio aligned to our API and microservice, the way that we develop code at Empire Life. So, it aligned nicely with APIs and microservices. And the rich library of API calls replaced the complex sequel that we had against Oracle in our previous environment.
Sandro Palleschi (00:09:03):
We reduced the risk of operating a mission critical system. Again, being a SaaS cloud platform, it had HA built in, had dynamic scaling, has high performance. It's secured in a very stable platform. And that was really important to us being a highly regulated industry. Getting the enhancements that we get on, I guess, a quarterly basis into our tenant with zero downtime and no internal IR involvement's also very important. And then it did allow us to decommit what was turned out to be a very extensive technical debt, whether exist in MDM platform. So, right now, we're getting to focus more on business delivery and less on maintaining the environment.
Sandro Palleschi (00:09:48):
As you could probably witness for yourselves, it's a much superior user interface and tenant console capabilities than what we were used to on the Informatica side. The licensing model was much better. We were used to be licensed on the number of crosswalks using the Reltio term as opposed to golden profiles like it is right now and the way that it's modeled in Reltio. And that's a significant cost savings for us.
Sandro Palleschi (00:10:14):
We had, I would say, not a great relationship with our previous vendor, very little focus on new features and customer value. We were assigned a customer success manager while we were going through the migration project. And the relationship with Reltio has remained very strong. Again, Reltio using Google Cloud Platform as their in-house digital development was also exclusively GCP. So, everything really aligned well with our environment.
Sandro Palleschi (00:10:46):
Here's our data model. Really quickly, everything you see in green is essentially what we define and configure in Reltio in the data model, which is very simple when you compare it to the data model, the way that it used to look like. As you can see, we have mastered party. Party for us are either customers or advisors. The empire Life is used in the Reltio Connected Data Platform with the Account 360 Solution Accelerator. It includes a data model for business to business, B2B business processes. And we use the master data for life and health insurance policyholders, as well as the related parties on those contracts and brokers and advisors that we sell through.
Sandro Palleschi (00:11:30):
So, the party is what we master. Parties can be either customers or advisors. It's broken up into three entity types. We have 3.5 million individual profiles, 21,700 company or organizational profiles, and 4.6 reference profiles. We have 3.75 contract profiles. Contracts were either individual policies or investment accounts with our customers or their advisor agreements with our distribution channel.
Sandro Palleschi (00:12:09):
Then between party and contract, essentially our relationships, we have 17.75 million party-to-contract relationships and there are different types of party-to-contract relationships. We have 7.11 million contract-to-contract relationships and 3.35 million party-to-party relationships. So, by looking at the numbers, relationships is really, really key to everything that we do in a master data management platform and that the numbers bear that out. Is there any questions on that?
Chris Detzel (00:12:45):
Not yet.
Sandro Palleschi (00:12:47):
Okay. A very high level architecture, we ingest our source data in batch today. So, on a daily basis, we will get the delta changes from all our backend systems. And through Informatica Data Quality Tool, we will land those changes after we've curated and standardized the data. The net changes go through a Pentaho ETL process where we load it from the staging tables that are in SQL Server. We actually use the Reltio ROCs utility by taking the JSON that we create through PDI off of all the delta changes by source system. And we load it into Reltio using the ROCs utility. We do have an initiative where we're going to move that to the data loader, but at the time that we were doing the migration, we had some issues with the data loader.
Sandro Palleschi (00:13:37):
So, we started using the ROCs utility and that's actually worked out really well for us. We have a number of back office systems on different formats. We ingest data directly from relational databases, like SQL Server, Sybase, and DB2, as well as extracts that we get from some of our older systems through relativity. They're VCM files. Once it gets into Reltio, then it does all the magic. It does all the matching and merging on the fly, everything that you can figure from survivorship rules and your match rules to create the coding profiles and manage the crosswalks and the relationships that we've loaded. And I'll talk about this later on, but we have a number of consumers either directly against Reltio through APIs or through the UI.
Sandro Palleschi (00:14:29):
We also go use the export API utility to create a relational copy of the Reltio data model. And that really comes in handy for a lot of reasons, which I'll get into in a bit. So, we consume data directly in real time with Reltio through APIs, but we also have access to the relational copy of the data usually for extracts, but also for other reasons. Metrics for example are much easier to do relationally than it is to do through the APIs.
Sandro Palleschi (00:15:03):
Jumping to the benefits of using Reltio, which is the core of what we wanted to present today. We have found a much more improved customer experience and customer service productivity with Reltio. The UI is very intuitive. The ability to access accurate timely data has really increased our first call resolution rate from what we were used to on the Informatica side. Increased data and data pipeline quality, the migrations helped us identify and remediate numerous data quality issues that we had with the MDM platform. Having to redo it all over again, we had to review all the mappings and reviewed the way that it was initially designed. So, that helped us uncover a whole bunch of data quality issues.
Sandro Palleschi (00:15:56):
The MDM platform in general, as you guys probably have seen, helps you also identify a lot of issues that you'll have in your back office systems. So, we've seen a lot of data quality issues that have come from source, things like missing contact information. Date of births are different for the same person in different source systems. We've had business processes that overwrote the crosswalk or the source record with a whole different person. When that comes into Reltio because it's an existing crosswalk, it tends to corrupt the golden profile. So, uncovering those business process issues has helped us clean those kinds of issues up and we don't run into those kinds of cross-pollination of individuals and organizations where the crosswalks got reused.
Sandro Palleschi (00:16:51):
The data profiling that we did from the beginning again helped us improve the match in survivorship rules. So, we got much better match performance as a result of the migration. The simplified data model has reduced our production support, troubleshooting, and triaging significantly. Today, one API call can return the entire profile. In the relational world because it's highly normalized, you had to create probably in our case one primary table with about five or six dependent tables and then you got complex sequel to join all that data and bring it in a view that we get through one API call in Reltio. So, that's been very productive for us. Again, being highly regulated, audit compliance and security is really important, obviously.
Sandro Palleschi (00:17:47):
So, the activity log, as you see in the UI, captures everything historically, who made the updates, where the updates came from, when they came. And having that out of the box is huge. In the environment that we were in before, we had to deliver those views of the history and the history of the changes through SQL against what we're called infrastructure tables, but we had to pretty much build that ourselves. This comes out of the box.
Sandro Palleschi (00:18:16):
And being able to see that at the dashboard that you can search and query through, but also be able to provide the activity log at the entity level and at the relationship level is huge for triaging. So, you can see the whole history of how the changes were applied to the relationships and to the entities themselves.
Sandro Palleschi (00:18:39):
The security model seemed a lot easier for us to integrate and using the console to maintain, the users of the user accounts and the roles that, and the privileges that they were allowed to have, the credentials that they were given. So, that's easy to manage and to see in the console. Higher business and IT productivity, using the contextual user interface, it's designed for business users, representatives. The color-coded crosswalk enables us to identify everything pretty much, the lineage back to the source systems, also helps us understand how to manage the data and the source systems by looking at it graphically through the crosswalk view in the UI.
Sandro Palleschi (00:19:27):
The manual match and unmerge is really, really easy as you've probably seen in the UI, completely auditable via the activity log. Another huge benefit is managing the potential match pool. You see it in the dashboard and you can search on it by the various manual match rules that you may have. That's really important for us. We started with a match full of about 75,000 potential matches when we are in the Informatica world. We had brought that down recently down to 20,000 potential matches, and we have action plans to bring it down to 10,000. You're never going to get to zero with potential matches, but you want to be able to get it down to a manageable number that can be actioned by the data stewards.
Sandro Palleschi (00:20:11):
So, getting that number down was a significant productivity improvement for us on the Reltio platform. And just again, being able to search on the potential matches has been very valuable for us, as well as being able to help tweak the match rules. When we went through the migration, we had to help the developers learn how to use the Reltio library of API calls. Once we went through that with the developers and we tend to only use three APIs and everything that we do in Reltio, we either use the entities, the relations, or the hops API calls. That's it.
Sandro Palleschi (00:20:56):
That covers, I would say, the equivalent of probably, if I can think back at our inventory, over 100 different SQL scripts that we would have used and built to deliver to our downstream services were essentially replaced with three API calls that had different filters and different selection criteria. So, that's how easy it was to do the migration and how easy it is to maintain the code. But once we went through that learning curve, developers that used to rely on us for the SQL for the MDM platform, they pretty much understand how the Reltio API calls were. And so, they can start building net new requirements themselves with very little consultation from my team.
Chris Detzel (00:21:44):
Hey, Sandro, quick question. And I was just curious, last time we spoke on Friday, you mentioned that you were going to start using or think about using the relevance-based matching. Does that come into play on bullet two or whatever to knock down some of the matching stuff that you're doing there?
Sandro Palleschi (00:22:03):
Yeah. I have a slide on our roadmap going forward, but yeah, I'll talk about it when I get there.
Chris Detzel (00:22:08):
Okay, cool.
Sandro Palleschi (00:22:09):
But thanks for bringing that up. The clone capability, so Reltio now allows you to request a clone of your production environment into your test environment, I believe once a month. We just started to take advantage of that. And it's probably easy to take for granted that capability is there. Trying to do that in the MDM platform was pretty much impossible. We pretty much had to reingest all the data through a one-time load to sync up our pilot environment with our production environment. That took days to run, believe it or not. This clone, the way that it works right now, it does affect our pilot tenant for about a day.
Sandro Palleschi (00:22:54):
But other than that, though, it's a complete copy of the data right down to the URIs. And so, that really does help with our development and a lot of our projects that will take advantage of production data in order to do a lot of their test cases. It's very difficult to recreate some of the use cases when we're testing new requirements without having production data and production volume data. So, the clone capability in Reltio is a really big advantage.
Chris Detzel (00:23:21):
Sandro, a few questions here are starting to pour in now. How was the Reltio API performance in your case?
Sandro Palleschi (00:23:29):
For us, it's been really, really good. I mean, when we replace the SQL calls, which were complex and not only were they complex, we had to use materialized views to improve performance if you folks know what materialized views are, but it's essentially been able to rejoin a lot of the data and create temporary tables. They're actually permanent tables, but it already encapsulated a lot of the navigation that we had to do and was causing us some grief from a performance perspective. But when we rewrote them into API calls using the Reltio APIs, for us, the performance was at least as good, but more often was better than what we were getting going after Oracle with SQL.
Chris Detzel (00:24:11):
Well, that's awesome. So, I see that you are using, I'm going to say this wrong, but Pentaho ETL to load in Reltio and extract the Reltio. So, a few questions, maybe he's jumping ahead. He's not sure, but were you using SQS and DBSCAN to extract? That's one question out of three.
Sandro Palleschi (00:24:33):
No, the source system data that was coming to us was a combination of extracts that were produced for us from some of the older systems, or we directly went to the databases and we were able to create the data that we needed to populate the MDM platform based on the parties, the related parties, and the contracts that were of interest to us in the MDM platform. We didn't have any other utilities other than using... Well, in our case, it was IDQ. The Informatica Data Quality was our ETL tool for ingesting data from source.
Chris Detzel (00:25:09):
What's the average time of extraction given the volume of your data?
Sandro Palleschi (00:25:15):
So, we do deltas now. It's actually a good question. The process of ingesting delta data, we have a change data capture process that was designed by our developers that takes full copies of the source data, does a CTC process, which gets the delta changes, and then we apply the delta changes. The batch window for us to absorb the data from all our back office source systems, it varies, right, but I would say it's... I wish I had JP here, but I think we're probably looking at five to eight hours to do that on a batch nightly basis. We've brought that down by three hours so far and probably going to bring it down even further. So, our ability to manage the daily cycle of our data coming in from our source systems is probably less than four hours now on a nightly basis.
Chris Detzel (00:26:12):
Speed is important. Assuming mastering of the data is done in Reltio, do you enrich data outside of Reltio?
Sandro Palleschi (00:26:19):
There is some enrichment done, more standardization and curation more than anything. I wouldn't say it's enriching. Reltio has Locate for address as you know. We had Address Doctor on the ETL side. Our experience with address cleansing and standardization has been much better using Locate than it was with Address Doctor. So, it's no longer part of the ETL process where we do that data cleansing and standardization is actually part of the MDM ingestion now. It performs better.
Sandro Palleschi (00:26:59):
And verifying the address based on in our case, Canada post for the most part, has been outstanding. So, that's the enrichment that we're getting out from Locate. We are looking at enriching the data with other third-party applications down the line, but for the most part, it's what we're ingesting from our back office systems.
Chris Detzel (00:27:20):
Great. Do you know the API return time in milliseconds?
Sandro Palleschi (00:27:25):
We can get it at and it is at milliseconds. Our SLAs, at the API level, it's three seconds or lower whatever 95% of the time. And we've met all our SLAs when it comes to the APIs. Like I mentioned, we export the data to a relational database, DB2 in our case, and we export it every night. So, we keep it in sync. That export runs really quickly. It's like less than an hour and a half, and we get an entire copy of the data into a relational format. And I haven't really given you a picture of what that looks like, but if you look at our data model, we have a table for the party entity, we have a table for the contract entity, and we have a separate table for every relationship, and then obviously, foreign key relationships between them all. But that export API works really, really quick.
Chris Detzel (00:28:22):
Great. and some comments here. It's actually extraction from Reltio to other SQS versus DBSCAN from the previous question. And then one comment is bold. Export API gives too good performance, if not great, but really good option to extract APIs. Let's see here. Scan API might degrade the performance and slower. Let me see if there's other questions here. So, time to market at, about what percent faster does the team deliver on the new business requirements and how many people work to deliver on delivery on new business requirements? Do you know?
Sandro Palleschi (00:29:09):
Well, so in the Informatica platform, we had up to three MDM developers working full time, managing the environment and making changes. In Reltio now, we're down to one and a half. And even then, most of the time now is on delivering new requirements, not managing the environment. So, our ability to deliver business requirements has been greatly enhanced. When I say time to market, the way that you can make changes to the data model and when you're configuring survivorship rules, changes to the match rules for example, it's days to do that work. In Informatica, it was weeks to do that work, to develop it, to change it, to be able to test it, and then be able to move it to production.
Sandro Palleschi (00:29:59):
You're going from a model where it took us weeks to do that work to being able to do that within our agile sprints, our two-week sprints. So, it's an order of magnitude, difference in time to market. And the number of developers that we have on the MDM side right now is down to one and a half developers essentially doing that work. But again, so people that ingest data from the party have their different development team, but like I said, they can turn around the new requirements so much more quicker, because again, it's essentially three different APIs that we leverage over and over again, right? So, once you've created the template, it just makes it easy for net new requirements as to how you would've consume that data through the Reltio APIs.
Sandro Palleschi (00:30:45):
I don't know if that helps. Just the previous question, getting data out of Reltio, from an analytics and business reporting perspective, I have to say right now, we have taken advantage of the relational copy of the data model. It's a little bit easier for us to do some of the heavy navigation that's required at the relationship side. And those relational tables that we've created through the export API makes that ability to do metrics and to do analytics on the data a lot quicker than it was going directly to Reltio APIs. Although, there are use cases where we want to use Reltio APIs directly. In our case, it's PowerBI as the platform that we use for reporting analytics.
Chris Detzel (00:31:31):
Got it. Just two more questions and then we'll let you go on. This is really good. Do you have additional layer between your consuming application and Reltio?
Sandro Palleschi (00:31:42):
What do you mean an additional layer? So, we have applications that directly access Reltio.
Speaker 3 (00:31:49):
I think you answered the question. So, between Reltio and the downstream application that uses your data, you have put a relational model in between.
Sandro Palleschi (00:32:02):
Well, only for some use cases. I will review the consumers of the Reltio platform in another slide and I'll give you the distinction there.
Speaker 3 (00:32:12):
And the reason only I ask is for one main reason is that you feel customer data is a lot of compliance. Security things comes into play when you replicate MDM data into a non-MDM environment, plus there is a governance part also. So, I just wanted to understand from your implementation, why did you create this and how useful it was in terms of performance?
Sandro Palleschi (00:32:40):
Yeah. So, performance, for sure. Again, because if you look at our data model, the value that you get from an MDM party, mastering party data, the golden profiles obviously are important, but to get to the golden profiles, everything is relationship driven, right? All the use cases are relationship driven. So, navigating the relationships is the key to performance. And so, the reason why we created the relational copy, it made that navigation a little bit easier. You make a good point about security and there is security that needs to be built into the relational copy, as well as in our case, leveraging the roles in the Reltio platform directly to manage what you can see and who can see it.
Speaker 3 (00:33:26):
Bear with me if I'm going-
Sandro Palleschi (00:33:29):
No worries.
Speaker 3 (00:33:30):
... to eat some of your time, but isn't it all just could be nested attributes in Reltio and you could have got all the payload from one single API? I'm just wondering, why would you say that you as a user of Reltio have to look into the relationship objects or how the data is connected? I think that's Reltio API would do for you.
Sandro Palleschi (00:33:57):
The best way to describe it, I guess, is that yes, absolutely, you can get the relationships through the APIs. It's when you're wanting to aggregate and you have use cases that you want to report on where you want to look at... An example would be if I want to find out someone that has individual business with us, as well as investment business with us, as well as potentially an active certificate on our group system, that kind of navigation and trying to do that in bulk when you're looking at metrics, that's much easier to do relationally.
Speaker 3 (00:34:31):
Yeah, it's more analysis work that you want to run.
Sandro Palleschi (00:34:34):
It's the analysis side. Yeah, exactly.
Speaker 3 (00:34:37):
Okay. Yeah.
Sandro Palleschi (00:34:37):
But yeah, the real time access is really focused on at the contract level or the profile level. So, you don't have to have the need to worry about it being relational. But the analytics side, it does help on the analytics side and the reporting side. Yeah.
Speaker 3 (00:34:53):
And the relational database, is it Oracle?
Sandro Palleschi (00:34:58):
No, in our case, it's DB2.
Chris Detzel (00:34:59):
DB2.
Sandro Palleschi (00:35:00):
Yeah. And we had the ability to create a copy of that, even when we were on the Oracle side, because even when we had Informatica on Oracle, it was the Informatica data model and the way that you define the business objects in Informatica. There was some structure to it. We put it into a structure that made sense to us afterwards. Did that help?
Speaker 3 (00:35:24):
Yeah.
Chris Detzel (00:35:25):
Great discussion. One last question and then we'll continue to go.
Sandro Palleschi (00:35:29):
Sure.
Chris Detzel (00:35:30):
So, you said one and a half hours for nightly sync to DB2. How long would it take to more frequent delta syncs? So let's say every 30 minutes.
Sandro Palleschi (00:35:41):
Minutes, we actually do deltas now to that relational environment. And so, we have talked about that as doing it more often and it takes minutes, less than 30 minutes to sync it up on a delta. Well, there's two things. We would like to do that obviously more frequently. And as I'll talk about in our slide near the end, we want to start taking advantage more of event streaming and being able to do things in real time using Pub/Sub or Kafka and do that as an event process, as opposed to a batch process, a batch ETL process.
Chris Detzel (00:36:18):
Great. Thanks, Sandro.
Sandro Palleschi (00:36:21):
How are we doing for time?
Chris Detzel (00:36:23):
We have 20 minutes. So, we're good.
Sandro Palleschi (00:36:25):
Yeah. I'll whip through this. I apologize. So, I go too fast here, just a few more slides to go through. And I won't go through every bullet item here. The business engagement on the MDM platform is critical for you to have a successful MDM platform. And we did not have that on the Informatica side initially. It was very much an IT driven initiative to bring in Informatica MDM platform. So, IT drove it, right? And so, adoption of what we call the party hub or the mastering of party data was probably not compromised, but we didn't take advantage of it as much as we should have had the business engagement been a lot higher.
Sandro Palleschi (00:37:07):
When we went to the migration of Reltio, so from the time we went from Informatica to Reltio, obviously we started to get applications that started to use the party data. But when we did the migration, the steering committee was made up of senior executives on the business level, as well as our Chief Technology Officer who sponsored the Reltio migration. That engagement alone, as well as demonstrating and doing presentations on the UI for Reltio really ramped up the understanding of master data management. One thing and I brought the UI up, I think, a few times already, but what we were used to in trying to explain or describe master data use in the Reltio UI is like day and night.
Sandro Palleschi (00:37:54):
It was so obvious by looking at the way that you guys have graphically represented profiles and the dashboard and the ability to look at the relationships, navigate to the party data from the contract and vice versa, looking at the activity log, looking at some of the dashboard metrics. It just made understanding master party data much easier.
Sandro Palleschi (00:38:17):
And so, different areas within the business obviously started to see the value, not only of what we've already delivered with the party hub, but could start to look and think of future use cases of how we would use the data, how we would ingest contact information from multiple sources for example, how we could sync up, say, the address across every footprint that that person had for that address back in the back office systems. So, lots of use cases are now coming up and they're coming from the business as opposed to IT driving it, which shows you how the adoption of master data has really improved as we've gone to Reltio than it was when we had Informatica. It was very, very much more technical presentations on Informatica side.
Sandro Palleschi (00:39:05):
The Reltio side allows us to deliver that message more in business terms, in business graphics. And that's why I was getting through the business engagement, the bullet items here. And the data stewards, our customer service reps, our business operations folks, all the users of the UI and/or the applications that I'll show you has really magnified the value and the use of party data going forward.
Sandro Palleschi (00:39:38):
Just the last bullet here and I briefly talked about it, but PowerBIs are reporting analytics platform. So, the single view of customer and the single view advisor context is provided by Reltio. Having said that, we leveraged the relational copy of that data to deliver the reporting analytics site, because it does deal with metrics, right, as opposed to actual profiles. So, that's a value added use case for the relational copy of the data. But the single view of customer, single view of advisor is critical pretty much to anything that we do in an insurance and an investment company. So, Reltio provides that context obviously through the natural survivorship tools. And we've been able to leverage that across a lot of use cases.
Sandro Palleschi (00:40:31):
I think I've covered probably most of this. We onboard applications quicker using BOTS. BOTS can access Reltio as well through the APIs. Our customer and advisor portals are probably the main users and the main benefits that you see visually with party data. So, that's covered in the bullet items that I've provided here. We have anti-money laundering and terrorist reporting extracts that are delivered from Reltio. We have reinsurance. So, these are the current consumers of party data. Customer portal, which is where our customers can register and look at all their holdings, they can see their investment and insurance policies. They can see their related parties like the beneficiaries, the insureds. They can see their holding values.
Sandro Palleschi (00:41:24):
Now the values come from a different application, but Reltio provides the context and the bridge to those policies and accounts and the details about the policies and accounts come from our integrated integration data platform, which is a DB2 environment. So, that's the other reason why creating a relational copy of the data model is really helpful. It allows us to join to the detailed contract data that we have in our enterprise platform. So, you can't do that obviously with joining APIs to other relational tables. Having a relational copy allows us to do that. So, we have an advisor portal. We have the distribution advisor portal itself.
Sandro Palleschi (00:42:09):
So, the customer's on the one side and the folks that actually sell and organizations that sell the business, they have the ability to view all the policies that they service from an Empire Life perspective. Hierarchy, which I haven't talked about, which you see as out of the box capability in Reltio, really came in handy for us. We had to programmatically deliver hierarchy in the old platform. Hierarchy as a facet or hierarchy as a capability that's built in out of the box with supporting API calls for us was really huge. And it gives you a graphical representation of the relationships that are managed through the hierarchy and then we use that for advisor hierarchies.
Chris Detzel (00:42:50):
Hey, Sandro.
Sandro Palleschi (00:42:51):
Yeah?
Chris Detzel (00:42:52):
What is your team size supporting your Reltio instance? So across like admins, ETL developers, configurator support, data stewards, et cetera. What does that look like?
Sandro Palleschi (00:43:01):
So, for developers, we have, as I mentioned, only down to one and a half now. And they do all the configuration, data modeling changes, matching, merging. I shouldn't say that. We also have an enterprise data analyst, and she's actually probably on this. She helps a lot with match and merge rules. She helps with working with the data stewards, helping with providing potential match reports that help the data stewards. We have, I would say, probably four or five active data stewards, but they're not full-time. They do that role part-time right now, but they use the UI obviously to do the data stewardship role.
Sandro Palleschi (00:43:40):
And then we probably have about out 80 customer service reps that use the UI. They don't develop or maintain anything, but they take advantage of the UI and they deal with customer calls using the UI, which for them, they've lauded the UI quite extensively, the value that they've gotten from the UI. And being able to see the single view of the customer and being able to relate with the customer on that has been really important.
Speaker 3 (00:44:08):
Hey, Sandro. Are you leveraging the sync back to your source systems or you are satisfied with the customer service reps accessing Reltio?
Sandro Palleschi (00:44:23):
We are satisfied with the customer services accessing Reltio. There are other things that they have done in the past where they've captured notes and things like that in different systems that they still rely on for other reasons. But when it comes to the customer data, the accuracy and the integrity of that customer data, absolutely, it is highly trusted.
Speaker 3 (00:44:48):
Thank you.
Chris Detzel (00:44:49):
So, one other question, let me see if I can do this right, but how the mastering app manages when there is an outage or downtime at Reltio on or on Reltio side, I guess, how do you manage that out, the mastering piece when there's downtime with Reltio?
Sandro Palleschi (00:45:08):
So, one of the things is there's very little downtime. In fact, there's no downtime with Reltio. So, anything that we need to do when it comes to updating the data model, for example, if we had to make a change, that doesn't require any downtime. When we do an enhancement, we get them out of the box from Reltio. There's no downtime.
Sandro Palleschi (00:45:27):
We've only encountered since we've cut over, which would've been March 2021, we've only ever had one hit where the tenant was unavailable for about... I'm trying to remember how long it was. It might have been about up to 30 minutes. And you guys now have a status page that tells us if there are issues. That was corrected by Reltio. We had a couple of consequences as a result of it that we were able to clean up, but that was really the only time we've ever had a hit on the tenant since we've gone live.
Chris Detzel (00:45:58):
Thank you.
Sandro Palleschi (00:45:59):
Does that answer your question?
Chris Detzel (00:46:01):
I think so.
Sandro Palleschi (00:46:02):
Okay. So, I talked about the other consumers of the portals. Obviously, the BOTS use Reltio. The consolidated view of customer advisor is applied to PowerBI reports through the relational copy that we have. We have extracts integrations to CRM systems. We have three CRM systems. One's on-prem and the other two is on the cloud. Reltio is used for anti-money laundering and terrorist reporting extracts, which are important obviously, but for being able to identify those use cases for the business. It also helps us with our reinsurance business.
Sandro Palleschi (00:46:40):
So, when you want to see the exposure you have at a life perspective, for someone, for example, may have multiple policies and they may be insured for millions of dollars, right? Once it gets to a certain threshold, the company will reinsure that risk to other reinsurers or other insurance companies. So, being able to identify when you hit the threshold and understand when that feeds into the reinsurance treaties is important, right? So you don't get exposed in case you have a lot of death claims in a period of time. And like I mentioned, being able to leverage the API to create the relational copy of the data has been really important to us.
Chris Detzel (00:47:23):
So quick question, have you had a chance to utilize the Reltio's workflows? And if so, can you add any points on complexity of workflows that you developed?
Sandro Palleschi (00:47:33):
So, we have started. The resource that was working on the workflows is no longer with us, but she did really good work for us and we are going to continue to leverage it, mostly on the helping with the data stewards and identifying potential matches as our initial use case.
Sandro Palleschi (00:47:50):
The out of the box workflow can take you so far, but I guess what we found out with it and again with the potential match use case that we were looking at was in the case of that particular use case, there's going to be some customized Java development that we weren't prepared for, that we're going to have to look into a little bit more detail and to truly exploit the workflow capabilities that are there, but there are some out of the box that are helpful. I think when you get into some of the more complex use cases that you want for workflow, you're probably going to have to use Java development as well on top of that.
Chris Detzel (00:48:32):
What's the biggest challenge you faced in Reltio and your approach to counteract?
Sandro Palleschi (00:48:41):
The biggest challenge to this point, I think, had been with the migration, just going from a whole different data model and being able to work through all the downstream services, as well as communicate the benefit of why we were going that route was a huge challenge for us. We did partner with Reltio. I'm not sure if Joel Snipes is on the call or not or on the webinar. He was our technical developer with Reltio and he was outstanding. So, he really helped us a lot. But that migration effort was the biggest challenge we've had at this point. It wasn't a Reltio issue particularly.
Sandro Palleschi (00:49:23):
It's just that it was such a big change and involved impact to a lot of applications and being mission critical, keeping the existing environment going while we were moving to a new environment and running them in parallel with a very small team. At that point, we were down to two folks. That was a huge challenge, and the team did a great job, I have to say. And I mean an outstanding job. And we were able to deliver the migration without impacting the business. And even when we did the [inaudible 00:49:54] of production, we minimized the impact of the business when we did those [inaudible 00:49:58].
Chris Detzel (00:49:59):
Great. And last question, because we got a slide to get through, but did you utilize LCA in Reltio? And if yes, was it useful?
Sandro Palleschi (00:50:09):
Remind me what that is again. I've forgotten the terminology.
Speaker 3 (00:50:16):
Life cycle actions.
Sandro Palleschi (00:50:19):
So, I'm sorry, missed that.
Speaker 3 (00:50:22):
Life cycle actions, Sandro.
Sandro Palleschi (00:50:26):
I mean, it sounds familiar. I can't off the top of head.
Sarah (00:50:31):
Sandro, it's Sarah. How are you doing?
Sandro Palleschi (00:50:33):
Hey, Sarah. Yeah.
Chris Detzel (00:50:34):
No, the life cycle action, we did talk about it on the implementation. It's a customization and I remember you guys did not end up...
Sandro Palleschi (00:50:42):
Okay. That's probably why I'm not familiar with it then. Yeah. So, sorry. I hope that answers your question.
Chris Detzel (00:50:51):
Thanks.
Sandro Palleschi (00:50:52):
Yeah, this is last slide. So, what we're planning to do going forward, the move to relevancy match rules. So, I know you brought that question up earlier, Chris. We see a lot of value in the relevancy matching. Anika who's on the webinar is doing that work and looking at investigating the use of relevancy match rules, because for us, we have a set of pairs of match rules. We have the auto rule that matches on the fly and then we have almost a somewhat equivalent manual rule with fuzzy matches and different criteria, but there's pairs, right? They come in pairs. With relevancy match rules, you can do that all in one rule and you set the thresholds. And a certain threshold is automatic. And then below the threshold, it becomes a potential match.
Sandro Palleschi (00:51:40):
So, there's a lot of value there that we certainly think we're going to be able to probably cut down our match rules in half as a result of going to relevancy match rules. So, Anika's already been looking into... I think the customer ones is the first one she's looked at. And I think the testing that she's done has shown that we're probably going to get more effective matching as a result and be able to manage our potential matches a little bit better as well. Real time data ingestion, so everything right now is essentially batch for the most part on our back office systems.
Sandro Palleschi (00:52:15):
But with third-party aggregators, looking at Reltio Integration Hub without having to do ETL development and directly connect to be able to bring data in real time, as well as obviously have our consumers access Reltio in real time, that either through the Reltio Integration Hub or through event driven updates in and out of Reltio, we want to make the experience in and out of Reltio to be more real time. Expanding the use, preference and consent is out of the box and works beautifully for us. We only have one use case that we migrated to, which was for e-statements, preferences that the customers had signed up for. But there are a whole slew of use cases where we want to take advantage of this.
Sandro Palleschi (00:53:07):
And certainly, the compliance with the Canadian Anti-Spam legislation comes into this. But those of you that have been involved with security and been involved with privacy, there are a lot of use cases here and all of them are we're facing in our industry. And we have confidence in the ability to be able to manage that information in Reltio with the out of box capabilities that are already there. Reference data management, we use them for lookups as they're intended, but the mapping to the canonical values from multiple back office systems where something like gender, for example, has different representations. We ended up doing that in the ETL process.
Sandro Palleschi (00:53:50):
And so, we want to migrate that and manage that directly in Reltio and taking full advantage of our DM. So, we didn't really take advantage of reference data management capabilities very well when we did the migration. And now that we've done it, we certainly want to circle back and take full advantage of what our DM can do and reduce the work that we're doing on the ETL side. Bi-directional updates back to source, so now, for example, we update addresses and contact information, but if that person exists in multiple back office systems, we don't go back and resync it across all those touch points.
Sandro Palleschi (00:54:27):
That's something that we want to do is to be able to have a full bi-directional ingestion and a maintenance of party data across Reltio and back to the store systems where the data came from. We're starting to see more data quality capabilities within the MDM platform in Reltio. So, we would like to take advantage of that and again, reduce where we're doing equivalent things in the ETL process. So, try to replace ETL cleansing and standardization with the corresponding capabilities that are becoming available directly in the MDM platform as you ingest the data from source is something that we want to get more involved with.
Sandro Palleschi (00:55:08):
We don't use interactions today, but we do want to start using them going forward. The one use case that we're looking at is keeping track of when customers log into the customer portal. There are a lot of communication use case touchpoints that the business wants to take advantage by knowing how often someone logs onto the customer portal, when's the last time they went on the customer portal. So, from a marketing perspective, from an overall customer experience perspective, we see interactions and some use cases there that we think will help the overall user experience.
Sandro Palleschi (00:55:45):
I talked a little bit about the privacy and compliance. So, there's a lot in data privacy and compliance that we need to align to. So, what you can see in the UI, what you can see through the APIs, who can see them, who should be allowed to see them. We have a unique situation where our staff policy obviously and the profiles that come under our staff policy are in Reltio. We don't want everyone to be able to see, obviously, everybody else's investments from internal Empire Life employees. So, we got to build in through the roles, the ability only for certain customer service reps to be able to look at employee information. And they don't want anyone else to be able to see that, which makes a lot of sense.
Sandro Palleschi (00:56:33):
Things like SID number, maybe only allow the SID number to be viewed by certain folks and not everybody and things like that. There's a lot of use cases there. We have to do it in two places. I forget who the gentleman was that asked the question earlier, but data privacy and compliance applies to not only our DM platform, but the copy that we've made as well in the relational format. And then leveraging the workflow to be able to automatically notify customer service reps and data stewards for various use cases like potential matches that have just came through that night, that kind of thing. So, I know I whipped through that. I hope I didn't go too fast, Chris, but hopefully, I covered what most folks are looking for and the values.
Chris Detzel (00:57:24):
Sandro, this is really great. I really appreciate one, you coming on and talking about your experience. I think you're the first person to do that on the community. And so, we are looking for others to do this as well. There are a couple questions unfortunately we didn't get to.
Sandro Palleschi (00:57:43):
I can hang on if they want to hang on for a few minutes.
Chris Detzel (00:57:46):
Okay, great. I'm going to ask two questions then. The first question is, is it for global implementation or data for the only US insurers or global customers?
Sandro Palleschi (00:57:59):
Actually, Canadian, we're a Canadian insurance company. We're a Canadian insurance company. Most of our customers, well, overwhelming number of customers are Canadians, but we do have international customers as well. And by the way, Locate comes in handy there for our addresses when they're international addresses. Out of the box, we get to standardize the verification and standardization across the international addresses as well.
Chris Detzel (00:58:24):
Great. Thanks for putting your video on, but what would be your feedback on implementing and prod environment relevance-based match rules and match IQ?
Sandro Palleschi (00:58:36):
I don't know if I mentioned match IQ. That's something we do want to take advantage of. So, being able to better understand the effectiveness of our match rules, so being able to exploit that a little bit more. The work that Anika has done on relevancy matching in our team, we're getting positive results or the ability to define that as a potential match initially, to do the testing really helps, right? So, we can compare how things are being matched right now with our existing rules. We introduce the relevancy rules and we see what the behavior is like. And it helps us gauge the effectiveness of those relevancy rules and help us understand what thresholds we should be setting to determine whether it should be an automatic match versus a potential match.
Chris Detzel (00:59:23):
So, just some comments here. Great presentation. Thank you, Sandro. It was really helpful and then this was awesome. And thanks for organizing this. Very helpful. Fantastic presentation from somebody that has been doing this for a long time as well. So, thank you, Chad. Just over and over, it was really appreciated. And so, this was really awesome.
Chris Detzel (00:59:46):
So, Sandro, thank you so much. This was a great session. Somebody said, "I would really appreciate if we can do some sessions of the best practice and biggest challenges faced." I like it. And I will try to push some people, customers and alike, to try to do some more of these, because I think they're extremely valuable. This will be posted by the end of the week. It's my goal. And it will be on community.reltio.com under the Events and then Past Events on the Community.