MDM is Still Failing 2020
Download as PPTX, PDF1 like87 views
The document discusses Master Data Management (MDM) solutions and their evolution. It notes that past MDM approaches relied on procedural rules that were difficult for businesses to use and change. Modern MDM solutions aim to empower data stewards by incorporating machine learning to more easily train and update mastering logic. The document argues next-generation MDM will focus on enabling data stewards and business users, simplifying the life of data stewards through easy collaboration tools, and leveraging machine learning for faster time to value. Selecting the right MDM vendor is important - priorities include how easily data stewards can train models, enable collaboration, support multiple data domains, and facilitate ongoing operations.
MDM is Still Failing 2020
- 1. Master Data Management Still Failing After All These Years
- 2. Paul Balas pbalas@303Computing.com 25 Years Experience Leading Digital Transformations Multiple MDM Implementations, Data Governance, and Data Warehouse Initiatives ❖ Digital Transformation Consultant ❖ IT Executive ❖ Enterprise Architect ❖ Developer
- 3. Confidential – Tamr, Inc. MDM Past, Present, and Future Past - Legacy MDM Approaches Present - What Vendors have done to address past failures Future - What will differentiate the winners?
- 4. Confidential – Tamr, Inc. What is MDM? Master data management ("MDM") is a technology-enabled discipline in which business and Information Technology ("IT") work together to ensure the uniformity, accuracy, stewardship, semantic consistency and accountability of the enterprise's official shared master data assets. - https://en.wikipedia.org/wiki/Master_data_management
- 5. Confidential – Tamr, Inc. If you don’t have trust In your data You can’t use it to drive Company Performance
- 6. Confidential – Tamr, Inc. What Should You Achieve With Best-in-Class MDM? Stakeholders have TRUST in their data Stewards benefit from their INVESTMENT in time Better ALIGNMENT across teams Spend MORE time on Analysis and LESS on Data Preparation Faster TIME-TO-VALUE
- 7. Confidential – Tamr, Inc. The Past
- 8. Confidential – Tamr, Inc. PIM and Customer MDM Customer 360 and Product Information Management (PIM) ❖ Match/Merge ❖ Right of Survivorship ❖ Procedural Rules ❖ Hard-coded ❖ Data Stewardship a solo activity and not very efficient ❖ Little else ❖ Larger companies had to buy both solutions
- 9. Confidential – Tamr, Inc. The Present What are vendors promoting to solve your MDM challenges today?
- 10. Confidential – Tamr, Inc. MDM is Expensive! Gartner says on average the people costs of MDM projects are four times that of the software licenses Reducing the time to implement is a critical driving factor to reduce implementation cost and risk
- 11. Confidential – Tamr, Inc. “Enterprise MDM is No Silver Bullet” “Enterprise MDM software is no silver bullet,” says Sally Parker, Senior Director Analyst, Gartner. “Technology alone is insufficient to solve the bigger problem of data and analytics governance, which involves people, process and technology across the enterprise.” - Gartner
- 12. Confidential – Tamr, Inc. Why Are Analysts Scared of MDM Projects? MDM initiatives often fail when organizations don’t ensure organizational readiness before starting. Many also confuse what is and isn’t master data. They treat all data as equal and fail to identify and prioritize their master data. - Susan Moore “Be realistic about your readiness to adopt an enterprise MDM solution,” Parker says. “Ask first if you have the type of culture, data and analytics maturity, and the right level of executive support required for cross-organizational collaboration.” - Sally Parker
- 13. Confidential – Tamr, Inc. It’s People ... Data Hoarding and protectionism are real Departmental silos are real MDM requires people to collaborate irrespective of organizational units Cross-functional teams are a necessity in most hi-value MDM projects
- 14. Confidential – Tamr, Inc. Executive Support The reason why the Analysts and Consultants start with “You need top-down support” is for those reasons BUT ... The real reason is that traditional MDM Vendors don’t address the real issue: ❖ It’s time-consuming to deliver a solution, and ❖ It’s painful to get good results on legacy technologies... ❖ Which creates a high burden on the organization, and Data Stewards
- 15. Confidential – Tamr, Inc. Data Steward Objections I’ve heard every one of these objections … ❖ I already have a day job ❖ We need a dedicated team to fix this data ❖ Can’t we hire someone to do this? I’ll oversee them ❖ How can I spend 20 hours a week fixing data and do my job? ❖ I have been doing this with spreadsheets for 10 years and it’s inexpensive ❖ Did you get Bob’s buy-in in Customer Service? It won’t work without him
- 16. Confidential – Tamr, Inc. Every MDM Vendor Says - “Empower The Steward” Ask yourself - How long does it take to train a steward? How much of their day is fixing rules or correcting the results? How does you platform enable subject matter experts to collaborate without getting in a room or on the phone?
- 17. Confidential – Tamr, Inc. Every CXO Says “No More Tribal Knowledge” They know they are beholden to the tribal knowledge on how the data works They know that without Bob or Jan in Department X their reporting would suffer It’s intellectual property that belongs to the company… MDM Vendors will say they solve for this But how they solve for it will make or break your success… What are some other problems?
- 18. Confidential – Tamr, Inc. Business vs. IT - The Battle of The Titans What is a business rule? It’s any rule built in your MDM system used to standardize or organize master data MDM fails when your platform requires an act of IT to make a change to the rules or a long training to let business users implement some of the simpler rules ❖ It’s slow ❖ It’s frustrating ❖ Your building technical debt and complexity that will fail in time ❖ It’s costing your business money in opportunity and operating costs ❖ Most Data Stewards hate manual data entry
- 19. Confidential – Tamr, Inc. Agile to The Rescue! Yes, agile is the way to structure your project (most projects) It’s a way to align business value with achievable outcomes for specific stakeholders that have explained how they will use your software to achieve that value Agile will increase your odds of getting stakeholder buy-in more quickly BUT... It can’t cure for an archaic MDM platform that does things the old way
- 20. Confidential – Tamr, Inc. Aging MDM Platforms kill the agility to adapt to change
- 21. Confidential – Tamr, Inc. Especially When IT Works Like This... Fill out this form… Learn The IT Process… I need a month to do this…
- 22. Confidential – Tamr, Inc. MDM Agility Killers Procedural rules-based code (if then else…) Software requires a lot of training Changing rules is expensive and requires significant testing Software doesn’t make the Data Stewards life easier No mechanism for e-collaboration Adding a new data source is an act of G-d
- 23. Confidential – Tamr, Inc. On-Prem Implementations are DOA Cloud Solutions are a given
- 24. Confidential – Tamr, Inc. Best of Breed vs. Integrated Offering MDM solutions—with embedded, proprietary integration tools—compete with companies that have data integration as a competency Some vendors offer data quality and integrated dashboards Best-of-Breed vendors tend to be more agile and do ‘one-thing-extremely- well’.
- 25. Confidential – Tamr, Inc. Multi-Entity Mastering is Now Table-Stakes Your business is more than Customers or Products Your MDM solution must be able to master all of your data ❖ Lower TCO ❖ Less complex to administer ❖ Fewer people to manage
- 26. Confidential – Tamr, Inc. The Future
- 27. Confidential – Tamr, Inc. What Exactly is Next-Gen MDM? Next-generation means that the capability offered by your vendor provides significantly measurable improvements over pre-existing solutions and techniques for solving a problem.
- 28. Confidential – Tamr, Inc. Machine Learning is Next-Gen (Now) ML is the next step-change in the MDM World ML allows your match/merge logic to be ‘trained’ efficiently by Data Stewards It allows them to capture their rules simply, quickly, and with minimal effort It is more flexible and adaptable to change This is much easier to sell internally The legacy MDM Vendors have to rewrite their current match-merge technology to leverage machine learning
- 29. Confidential – Tamr, Inc. ML vs. Procedural MDM Effort Time Train Operate MLP
- 30. Confidential – Tamr, Inc. Data Stewards - Spend More Time Analyzing/Strategizing Before: Data Scientists spent months & 100% of energy preparing data. Today: ML can do 80% of data mastering lift... …. Enabling Data Scientists to put final touches on the last 20%. 30
- 31. Confidential – Tamr, Inc. Why Hasn’t Your MDM Vendor Modernized? Blunder #9: Succumbing to the Innovator’s Dilemma. In his classic book The Innovator’s Dilemma, Harvard Business School professor Clayton Christiansen suggests that when technology changes, and you are a vendor that is selling the “old stuff”, it is very difficult to pivot to the new stuff, without losing significant market share in the process. As a business, you have to be willing to change and evolve when it is needed. It’s possible—and even likely—that a reinvention will hurt your business in the short term, but it’s absolutely necessary to stay in business for the long run. There are plenty of examples of this in practice. One that most people are familiar with is the emergence of ridesharing companies like Lyft and Uber, and the negative consequences for legacy taxi companies. Today the cost of a taxi licence in the City of Cambridge has dropped from $700K to $10K. - Dr. Michael Stonebreaker https://www.tamr.com/blog/avoid-10-big-data-analytics-blunders/
- 32. Confidential – Tamr, Inc. So Who is Enabling The Data Steward? We made the case ML is important Are any vendors doing Next-Gen(Now) ML?
- 33. Confidential – Tamr, Inc. The TAMR Agile Approach to Data Mastering Mastered data OLD WAY Rules-based Source data Mastered data Time Quality Months to years 60%–80% Accuracy Modify rules, create exceptions Months 1–4 Months 5–12+ Iterate Machine-driven NEW WAY Days to weeks 90%+ accuracy Source data Weeks 1–12 Iterate with human- guided machine learning Identify developers Get business input Write rules Review with business Unified data Rules
- 34. Confidential – Tamr, Inc. Selecting The Right Solution
- 35. Confidential – Tamr, Inc. My Guidelines Run from Vendors ... ○ that highlight technology features over proving value through example ○ that can only support a one or two domains ○ that can’t offer strong references ○ that offer ‘fully-integrated’ packages (ETL, Catalogue, MDM…) ○ that have not modernized their approach to MDM (still crazy after all these years) Run towards Vendors ... ○ that focus on business outcomes ○ that leverage modern approaches to MDM ○ that simplify the life of the data steward ○ that make it easy to operate the system
- 36. Confidential – Tamr, Inc. What is Important? A B C D A — Critical Capabilities ● Data Steward Trains The Model - Procedural vs. ML ● Data Steward Collaboration ● Multi-domain ● Cloud Enabled B — Important ● Schema Mastering ● Data Quality ● Metadata Management ● Hierarchy Management ● Match/Merge/Golden Master ● Core - Data Catalogue ● Reference Data Management ● Data Modeling ● Analytics C — Nice to Have ● Master Data Governance ● Data Integration - ETL/ELT/Pub-Sub/API… ● Data Sharing Exchange ● Data Syndication D — Considerations ● Graph, Big Data, RDBMS, Sparq, In-Memory, Hadoop… ● Organizational Change Management Required
- 37. Confidential – Tamr, Inc. What Do Customers Say? https://www.gartner.com/reviews/market/master-data- management-solutions
- 38. Confidential – Tamr, Inc. So What Makes a Great MDM Vendor? Great MDM is powered by people working together with software that efficiently transfers their expertise into the mastering of data so business can have confidence in the information that drives decisions They respect and enable these Data Champions so that their knowledge and time is leveraged as efficiently as possible Great MDM Vendors provide smart ways for Organizations to have conversations on live data, and implement corrections without needing IT
- 39. Confidential – Tamr, Inc. Thank You!
- 40. Confidential – Tamr, Inc. Q&A For a consultation on your data strategy: info@tamr.com pbalas@303computing.com http://www.303computing.com/
- 41. Confidential – Tamr, Inc. Be On The ‘Right-Side of The Fence’ Most vendors have moved towards ‘Quick-Winners’ strategies and it’s shaping their offerings
Editor's Notes
- #2: Add in my war stories instead of just relying
- #4: As a technologist i’ve implemented many mdm solutions and want to share with you the state of the industry and where it’s going so that you can make informed decisions in your next successful mdm project.
- #9: In the past, there was Customer MDM. It was built to solve match/merge for customer data. Over time, it evolved to leverage more advanced capabilities based on NLP libraries, but was essentially just ‘Customer MDM’. It was rule-based, hard-coded logic, and had little focus on enabling Data Stewards. Sales and Marketing loved it. It was called Customer 360 Product Information Management (PIM) was built to help companies with large product catalogues and inventory where they had no clue as to what their spend was. Larger companies would buy both Customer MDM and PIM. PIM was even more complex to implement than Customer MDM. Very few companies got the benefit of their investments. Data was suspect.
- #19: Clear up the last sentence with data stewards - in the business not it
- #23: Tighten up
- #24: Think through flow 22-23
- #26: Connected data - with accellerator based solutions, you still need a way to connect your information. I’ll tell you that connecting information is not hard, once you have mastered entities...
- #29: Gartner - By 2022, 60% of organizations will leverage machine-learning-enabled data quality technology for suggestions to reduce manual tasks for data quality improvement. As artificial intelligence (AI) technologies mature and become more widely adopted, many data quality vendors have started incorporating them into their solutions. In building augmented capabilities, they are driving better automation in areas that have traditionally relied on intensive manual tasks such as data matching, cleansing and transformation. Augmented data quality extends conventional data quality features to reduce manual tasks with automatic recommendations on “next best actions.”
- #30: With Machine Learning-based Data Stewardship, your stewards transfer their IP of how your organizations data works into the framework by simply training the model. Very quickly, the model ‘learns’ how to handle the ambiguity of data through example. The effort to manage the data becomes trivial in days to weeks. With the old Procedural approach, the burden on the Data Steward is significantly higher, and the cycle times of stewards telling IT how to code the rules slows results down. As you add new data sources into the system, the difference in approaches becomes even clearer. Finally, using Procedural code, it becomes almost impossible to keep all the rules straight, and you end up with unintended consequences. ML does away with all that proving to be a faster and more cost-effective solution.
- #31: Data Stewards don’t want to spend time telling IT how to fix data, nor do Analysts want to be buried in spreadsheets and queries. They want to analyze the data. With TAMR, Analysts spend more time analyzing and less time cleansing data.
- #36: Picking vendor - add in some more stuff