In continuation with our Master Data Maestros interview series, today we are delighted to share the views of our good friend Henrik Liliendahl. Henrik is a seasoned data management consultant with a long track record from both the vendor and user side within data quality management, multidomain master data management and data governance. Henrik is also very active in the data management community on social media including running a popular blog and a market overview service. Henrik’s latest venture is a cloud service for sharing product information between parties in business ecosystems.
Thank you for allowing us to interview you Henrik. Really appreciate your time. Can you please tell our readers a bit about yourself? Your background, education, work experience.
In the 90’s, when I worked as a consultant in a private company, I became interested in data matching as I had a job around measuring the intersection of customers at two companies, who were in merger negotiations.
When I started as an independent consultant in 1998, I based my business on a data matching tool. I got into some exciting data matching business cases at Dun & Bradstreet and Experian, who are heavy users of data matching capabilities. I also started being active in the emerging community around data quality.
This led to numerous consulting assignments around data quality, data governance and Master Data Management (MDM) both at end user organisations and at tool and service vendors. In parallel, I have been involved in a couple of start-ups.
Editor’s note: Seems Henrik and we share our roots to our love for data matching 🙂
What’s your current role? What’s the business problem you are solving and what impact does it have on the organisation?
I still do consultancy. I am involved in a start-up called Product Data Lake. I run my blog for the 11th year and a sister site called The Disruptive MDM / PIM / DQM List.
In consultancy, I usually work along with other professionals in making innovative MDM and Product Information Management (PIM) solutions encompassing people, processes and technology with passion.
The Product Data Lake service solves the pains trading partners have, when they increasingly must exchange product information between each other. This discipline is called Product Data Syndication (PDS) and we want to be the best at that by establishing a win-win situation for manufacturers and merchants – and the vendors of the inhouse PIM solutions as well.
My aim with the blog is to share the insight I gather and have discussions with others having other experiences.
The Disruptive MDM / PIM / DQM List is a platform where solution and service vendors can present themselves and their unique resources. It is like an MDM / PIM / DQM conference – however online and running ongoing 24/7.
Editor’s Note: Wish you the very best with your startup – Product Data Lake.
Reifier is so proud to be on the MDM List, do check us here.
In the recent Magic Quadrant for MDM solutions, Gartner has stated that “MDM continues to shift from reluctant to indispensable spend across a broader range of industries”. What are your views on this?
That is indeed my observation too. When the term MDM emerged in the 00’s, this was quite niche and not commonly known among business folks and top-level management.
Today it is not about if an organisation should do MDM, but about how.
Solutions have matured considerable and we see that the growing count of available solutions have different capabilities allowing them to be the best choice depending on geographical reach, industry affiliation, business cases in scope, deployment approach and a range of other parameters.
Also, the growing maturity of the discipline encompassing the people and process sides and data governance frameworks allows for feasible implementations in a broader range of industries and business scenarios with a foreseeable business outcome to be included in the business case.
According to you, how has the MDM landscape changed since when you started to the present?
Customer Data integration (CDI) and Product Information Management (PIM) can be considered as the predecessors of MDM and the single domain view (customer MDM and product MDM) ruled in the first years.
Then multi-domain MDM became the norm, meaning that an MDM solution must encompass party master data related to several roles as customer, supplier, employee and the product side that is seen as a broader “thing” side encompassing most commonly assets too. As an example, I was in the early 10’s involved at a leading PIM vendor in the quest to scale their capability from purely PIM to multi-domain MDM. Several other PIM vendors have followed that path.
So today we have an MDM market with an adjacent and overlapping PIM market. At the same time innovative MDM vendors are doing extended MDM where more data than what is master data is covered under the same data governance umbrella. For example, we see MDM vendors compete with CRM vendors on the Customer Data Platform (CDP) theme.
Also, the move from on-premise to cloud is significant. This is natural as MDM follows where ERP, CRM and other applications go.
You maintain a service for solution finding for MDM/DQ/PIM. Please tell us a bit more about it.
The service started as the Disruptive MDM / PIM List being a list of the MDM and PIM solutions on the market who have seen the opportunity to be on a list that is not dependent on inclusion criteria (or bias) and favourable ranking by the established analyst firms and where the vendors can present their unique capabilities.
Last year the list was scaled up to include best-of-breed Data Quality Management (DQM) solutions as well.
Also, a service for making a bespoke list for an organisation on the look for an MDM, PIM and/or DQM solution was introduced. Your Solution List, as the service is called, takes into account the specific context, scope and requirements for the intended solution. More than 100 Your Solution Lists have been delivered to the requester within a few hours or days. Interestingly many of the requests are from consultancy firms who works for clients in solution selection.
The latest feature is The Resource List that allows registered solution and service providers to list their white papers, ebooks, webinars and other online content in addition to the general presentation. The next planned is a similar Case Study List sorted by industry sector. So here there is a lot of cross vendor content available for organisations with plans of implementing or optimizing their MDM / PIM / DQM capability.
Solutions who are included on the list do not need to have a multi-million-dollar revenue. I am besides the established ones on the look for rising stars on the MDM / PIM / DQM market. Several of the solutions starting as “unknowns” on the list are now on the radar of the established analyst firms. So, organisations on the look for a future proof solution may find the best fit vendor on the list today.
Analytics are a focus area for most enterprises, big and small. Can you please elaborate some examples you have seen where an MDM implementation fuelled the analytics journey?
Certainly. Let us take a customer MDM example and a multi-domain MDM example.
One often sees a business divided into B2B and B2C operations and the reporting follows that concept. However, when doing data cleansing and data matching, we could come out with a much more nuanced demographic and firmographic picture that was reflected in the MDM hub. Many customers that were considered private were in fact small businesses. Add to that getting a grip on the structures within the segments. By identifying households and other relations as well as the company family trees, the reporting became considerably more real-world aligned with better decision guidance as a result.
In another example fatal decisions were avoided. Looking at product profitability alone suggested that some products could be discontinued. But when combining with the customer hierarchies there were some products that were bought along with profitable products and discontinuing a part of the basket could lead to a risk of abandoning the full customer relationship.
Editor’s note: This resonates with our experience as well. Householding and customer data unification lead to unexplored customer life time value and segmentation, improved targeting, revenue growth through cross selling and better risk and compliance. The good thing is that with technical advancements, it has become very easy and fast to conduct this analysis.
You have written extensively on the challenges and nuances of fuzzy data matching and deduplication. It will be great if you can share your views here.
Indeed. This goes back to the days when I developed my own data matching tool and later merged that into a wider data matching suite.
Working for example with Dun & Bradsteet on their data matching capabilities allowed me to explore the full aspect of data matching. Matching person names and addresses is hard enough. Matching business entities is a level up. Doing that in a global context is yet a level up.
I learned that you must apply a range of methods to obtain the right balance of identifying the true positives, avoiding the false positives and not leaving out a bunch of false negatives. Doing simple methods of match codes and using soundex won’t do the job. Even the bit more advanced edit distance algorithms do not bring you near. Besides similarity you must also include an advanced candidate selection as you from a performance perspective can’t compare everything with everything. You must wisely combine a set of methods and technologies.
We also included machine learning at Dun & Bradstreet. As there were a team of people evaluating the dubious matching results from data sets from a range of organisations, we could record their decisions and apply those in the next match and thereby reducing the count of dubious results.
I am not blown away when I see the data matching engines that are included in the MDM and data quality tools around.
Machine learning and execution in AI deployments is the way forward. The challenge is not having to retrain in each case but being able to share the training outcome without the confidentiality issues that arise.
Editor’s Note: True that! AI is the way forward for data matching. Use of AI and ML in fuzzy data matching improves accuracy. AI based discovery of matching rules leads to blazing fight deployments. Intelligent indexing through AI enables data matching of millions of records in minutes, not days!
Which domains do you see most prominently covered for MDM – Customer MDM, Vendor MDM or Product?
Customer MDM is still the most frequent domain implemented and is the domain being relevant for most organisations. However, product MDM is catching up fast and have become a necessity in industries where the number of products is high and/or the complexity around product master data is significant.
Vendor MDM is not so frequent yet. As MDM solutions improve and the MDM discipline mature, we will see more business cases showing a positive Return of Investment (ROI) for that domain too.
What are some forums, websites and conferences you would recommend to update oneself about the MDM field?
I think LinkedIn is the place to start, because much of the linking to relevant sites is here. There is of course a lot of noise on this social media platform. It has become a fact of life that you need to see through the noise and select the serious links that matters to you.
It is also a matter of how you like to have the content served. Some like long articles and some like short posts. Some like reading and some like videos and podcasts. Some like funny inspirational stuff and some like the thorough academic approach. Many, including me, like a mix.
Conferences are often expensive, both for registration and then if you need travel and accommodation to attend. But they can be of great value too. And of course, everything offline is postponed at the moment.
So, in summary: Hook into LinkedIn, join the relevant groups, connect with like-minded people and consume the content that suite you.
Please name a few data practitioners whom you admire and would love to hear from in this interview series.
There are a lot. Some that comes to my mind are Scott Taylor, Daniel O’Connor, Nadim Wardé, Kersten Wirth, Ben Rund, Chris Jobse, Michael Fieg, Kate Koltunova, Sophie Angenot, Steve Jones, Salah Kamel, Ian Buttery, Kimmo Kontra, Julie Hunt, William McKnight, Julian Schwarzenbach, Ken O’Connor, Malcolm Chisholm, Mike Ferguson, Nicola Askham, Prash Chandramohan, Dylan Jones, Thomas Frisendal.
Editor’s note: Thanks for this exhaustive list, we will be reaching out to these maestros and hope to talk to them soon.
Pleasure learning from your experience Henrik, thanks for taking time out and we wish you the best.
Comments, questions, suggestions, feedback – contact us here. Thanks!Posted on April 6th, 2020
Check out our Observable notebook where we show some stats on how entity resolution or fuzzy matching – address matching, customer matching, customer deduplication etc blow up at scale.Posted on April 3rd, 2020
Hope you are enjoying the Master Data Maestros series. Our guest today is George Firican, a passionate advocate for the importance of data.
Thanks for spending your time with us George. Can you please tell us a little about yourself ?
Thank you very much. It’s an absolute pleasure.
I am a data governance and data management practitioner currently working as the Director of Data Governance and Business Intelligence at The University of British Columbia. I’ve been very lucky to work with amazing people and teams and together we received recognition through award-winning program and project implementations in data governance, data quality, business intelligence and data analytics.
I’m a frequent conference speaker and YouTuber and have been ranked among Top 10 Global Thought Leaders and Influencers on Digital Disruption and Top 20 on Innovation and Big Data. I am also the founder of https://www.LightsOnData.com, a website filled with practical information in the form of articles, templates, best practices, guides, and courses to address your data governance and data management questions and challenges.
How did your journey with data begin?
Depends how you want to look at it. For the most part, almost anyone working in an office, they work with data even if they like to think that they don’t. So I was a bit more hands-on with data while working as a software developer building e-commerce sites and membership sites that of course they all had databases to work with. But I became more aware of the importance of data when I was working for this business school and I was involved in different projects where data quality had a key role in the success of accomplishing the scope. Of course, you cannot sustain data quality without data governance and that led me to invest my focus in learning and practising data governance.
How is the master data management landscape changed from just when you started to the present?
Gartner refers to MDM as “a technology-enabled discipline”. So there are definitely a lot more vendors in the game providing MDM solutions or solutions supporting MDM. The rapidly improving technology has also enabled companies to collect and mine more data, posing new challenges in the MDM space when it comes to unstructured data. Plus today’s business environment is on the way of becoming more data focused and hopefully more agile, requiring information to be drawn out of real-time data, as much as possible.
Please tell us a real customer story which had a phenomenal impact on the organisation.
I remember this bank that was not making good investments when providing credit because their systems were not well integrated with one another. They had a system for mortgages, one for banking services offered to individuals, another to businesses, and so on. They were doing credit checks using external sources, but not so much based on their own data. This is because they were having a hard time to consolidate all the different records they had on a single customer. Because of this they couldn’t determine easily if a customer already had an active mortgage with a branch in a different country or a line of credit and so on. Once they invested in data governance and MDM they were able to better integrate all these separate sources of data which let them making better decisions in what customers to offer better services.
Editor’s note: What a valid Customer MDM use case! This one is in the banking sector, but our customers report similar stories in life sciences, manufacturing, retail and other industries.
What are the biggest challenges you foresee in the current MDM space?
I’ve been asked questions such as: “Which project do you recommend we invest in first, MDM or data governance?”. A lot of organisations think that they need to tackle them one at a time, that they need to choose between them. You shouldn’t. The reality is that while you can implement a data governance program without doing an MDM project, though an MDM project is a good driver for data governance if such a program is not in place, but you cannot implement MDM without data governance. The two efforts are not mutually exclusive. So a lot of the challenges in the current MDM space actually come from a lack of data governance as part of the implementation. You cannot achieve a good “golden record”, have successful data integrations and consolidations, and then accurate reporting on mastered data, without a common verbiage, data standards, accountability and ownership, clear business rules and policies on data creation, acquisition, maintenance, dissemination, archival, and so on.
MDM Implementations have a reputation for being extremely expensive and behind schedule. What are your thoughts on this?
Experiences vary, but you are right, in general they tend to be quite expensive and behind schedule. This is because it’s usually organisations with a bit more complexities and challenges that find the need to invest in MDM. So then the complexity of the MDM implementation is directly proportional to that. The technical challenges faced by the MDM are not as great as the data governance aspect. There are a lot of business needs, rules and requirements that need to be uncovered during the MDM journey and sometimes some can be in conflict with one another. Some of these lead to exceptions and customizations that need to be done to meet specific needs of individual business units, which in turn can have a scope creep.
What advice would you like to give to a customer starting their MDM journey?
Don’t underestimate the scope and complexity of your MDM implementation. As with any project, communication is key, as is the importance of change management. The MDM journey doesn’t just stop at the end of the project and it needs to be agile enough to adjust to new systems, new business rules and requirements, as well as new staff.
Please name few data practitioners whom you admire.
David Loshin is the person that first comes to mind when I’m thinking of MDM practitioners. On data stewardship I recommend Dr. Anne-Marie Smith, Robert Seiner, of course, on data governance and data strategy, Stephen Few on data visualisation, Scott Taylor on making your leadership understand the importance of master data, and there are many more :).
Editor’s note: Thanks for this, we will be reaching out to these master data maestro’s and hopefully our readers will learn from them soon!
Pleasure talking with you George, thanks for taking time out and we wish you the best.
Comments, questions, suggestions, feedback – contact us here. Thanks!Posted on March 29th, 2020