Join AI & data leaders at Transform 2021 on July 12th for the AI/ML Automation Technology Summit. Register today.
Master data management (MDM) has historically been a murky area many organizations simply ignored. In fact, no one is still quite sure who is responsible for data quality within many organizations.
But data quality has become an increasingly critical issue as organizations look to improve the customer experience and lay the groundwork for investments in AI.
VentureBeat caught up with Manouj Tahiliani, general manager for MDM at Informatica, to gain insights into how MDM is rapidly evolving into an autonomous process that needs to be infused within a larger data management strategy.
This interview has been edited for brevity and clarity.
VentureBeat: Why is there so much focus on MDM now? To some degree, it’s always been a challenge.
Manouj Tahiliani: The pandemic was this … event where companies have been forced to digitize. It’s happening across every area. The natural instinct is for businesses to roll out point solutions. It’s kind of the easy way out. All of this is leading to applications and technology sprawl. We’ve always had a fragmentation issue — now the problem is further compounded, so I think there is a realization. That is a trend that we are seeing now. It’s very apparent MDM is the foundation for any kind of digital transformation.
Above: Manouj Tahiliani, general manager for MDM at Informatica.
VentureBeat: Who is responsible for data quality these days?
Tahiliani: There is a classic problem within data management in terms of how to govern. You need to lay out a data strategy that aligns with your board strategy or your other business outcomes that an organization wants to get. But then you get to the next level of detail. How do you implement the data strategy? One of the foundations of any successful data management journey is having a good governance framework in place because governance is what brings the people process and technology together and puts in the necessary guardrails. Businesses own the data. IT teams can’t tell if two customers are the same or not, but IT does need to lay out the processes to give the business stewardship capabilities. It’s a collaboration between business and IT to be successful. It’s not a one-and-done kind of thing.
VentureBeat: Does every organization need a chief data officer?
Tahiliani: We’ve seen the chief data officer. Sometimes they’re also chief digital officers. It’s still an evolving space. I see that longer-term the chief data officer will be someone responsible for data. It may be a title with a role to it, but it’s certainly a role with certain responsibilities. When their role is not defined, it’s usually somebody from the business who’s kind of taking responsibility and kind of acting as a chief data officer without the title.
VentureBeat: It seems data management is ripe for automation. How much progress is being made on that front?
Tahiliani: MDM is evolving very rapidly. We are already using AI to improve. There is definitely a line of sight toward autonomous MDM. A core process of MDM is to get a single view of the customer. Then you have complex business processes on top of it, or what we call MDM applications. The MDM secret sauce is to get a single view of a customer that can be shared using a process that is autonomous. We’re able to discover metadata. We have the ability to automate pipelines from the standpoint of connectivity and then getting data to flow. We have a capability from the standpoint of matching that is based on machine learning and then being able to share data in a process once you have those governance policies laid out. We’re not all the way there, as yet. What we are doing currently is applying machine learning in different aspects of that MDM process. Matching is a big one. Being able to give recommendations around what should be the business model based on the metadata is another. It’s getting automated. It will move from the semi-automated process that is supervised to basically becoming an autonomous MDM process at some point.
VentureBeat: How will we know we can trust that autonomous outcome?
Tahiliani: Governance will become an important aspect if you’re automating everything. How do people trust that this data is really good enough just because the machine told me? You need to provide some explainability out there. We need to make sure the governance process around the front end in terms of deploying the machine learning algorithms is much more robust. It’s an exciting time because the core MDM process can definitely be fully autonomous. We see that happening soon.
VentureBeat: Business users don’t always trust the data. Will that improve as a result?
Tahiliani: As long as you can visualize how things were done, it’s not a black box process. There is a way to introspect and understand how the machine learning algorithm actually came to a conclusion. That would build confidence. I think we just have to make sure that we take everybody on that journey with us.
VentureBeat: Is there simply now too much data to be processed?
Tahiliani: It’s an aspect as the volumes increase. It’s just not humanly possible to have that stewardship scale. A lot more data is going to go through in an autonomous way.
VentureBeat: Is more of the data also now being processed in real time?
Tahiliani: Real time and streaming are in the future, primarily because data gets stale pretty fast. If you look at it from a standpoint of engaging with a customer, those kinds of interactions have a lifespan. You need to provide a recommendation to a customer in near real time to provide a relevant experience. It is important for any data management technology to be able to support streaming capability and provide real-time recommendations. We’re using a lot of exhaust data to get intelligence out of it. But that exhaust data also gets stale very fast and loses relevance.
VentureBeat: What impact has the cloud had on MDM?
Tahiliani: It’s an exciting time currently to be in MDM. We launched the intelligent data management cloud. MDM is a key service within the Informatica intelligent data management cloud. MDM cannot get deployed much faster. You no longer have two months of deployment. We’ve come a long way. MDM projects used to be six to nine months, now you have customers getting value in three to six months. There is a near-term opportunity for us to start harnessing that kind of community knowledge. Cloud is definitely facilitating that.
VentureBeat: What’s your best MDM advice then?
Tahiliani: I always tell customers it starts with the data strategy. Lay out a data strategy that is aligned to your business strategy and then really go through the journey. It’s not about boiling the ocean. You can start small. It’s also important to factor in the longer-term aspects of your journey and not be very near-term focused. It has to support your data needs, not just for today, but for the future too. It’s important that you look at your data strategy from the lens of a data platform. It’s important to have a data platform as a foundation that supports multiple domain needs.
VentureBeat
- up-to-date information on the subjects of interest to you
- our newsletters
- gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
- networking features, and more
Source: Read Full Article