Capping IT Off

Capping IT Off

Multi-domain MDM is a pointless question

In the world of MDM there appears to be an odd discussion going in around whether "multi-domain" MDM is required or not. What is multi-domain MDM? Well the first thing is to understand what product companies mean by multi-domain and what businesses mean by the same term MDM Entity DomainsThey mean in other words the different types of entities that you are managing. So do you just manage the customer entity, just the product entity or manage multiple types of entities. Sometimes this views suppliers and customers as being in different "domains" as they might have a different database structure. The point here really is that these are very much a technical view of a domain driven from a product perspective. The question the vendor is asking is whether your MDM programme is going to manage one or more tables in a database and managing multiple tables makes you multi-domain. So Multi-domain for product vendors is actually asking about the database schema and how many tables you want to manage. This question is driven very much from the type of MDM solution that a vendor has and linked very much to what their product can actually do from a database schema perspective rather than a business outcome view. So what about the business? Does this entity view match the business view of domains? No not really, in fact not at all.MDM Business Domains The business looks much more at the concept of business domains.

  1. Customer Centric - Focused on adding value and customer interactions
  2. Enterprise Centric - Focused on driving efficiencies and consistency internally
  3. Supply Centric - Focused on driving cost out of the supply chain
In each of these areas the product vendors question of "multi-domain" is irrelevant because quite clearly you can't do any of these things without considering the full trinity of Party, Asset and Location. You can't understand your customer if you don't know what products, and particularly what type of products, that they buy. You can't create more internal efficiencies through HR if you don't know the skills and locations of your employees and you certainly can't drive cost out of your supply chain if you don't know which products come from which suppliers. Multi-domain MDM therefore is a pointless question from a business perspective. You don't want to do supply-centric MDM which is focused on cost and at the same time drive it to deliver a single customer view from a business perspective. These are distinct business challenges, which may end up having a common technical solution but from a business perspective they are linked but distinct. The question of multi-domain therefore is pointless and wrong and asked because people are viewing business problems from a purely technical perspective. Technically MDM should always be looking at all three types of entity information and therefore should technically always be multi-domain. From a business perspective however you should always concentrate on a single business domain as that is the best way to achieve the right business outcome. Its an interesting question as to how on earth MDM technologies can even propose a single entity MDM solution to address these multi entity business problems. How to create a single view of a customers position if you are sure they have accounts in ten systems but you don't know if those are credit or debit accounts? How to negotiate between suppliers if you know who they are but you don't know which of the products are equivalent? So when you look at MDM remember that it is the business perspective which counts not the technical one. MDM isn't multi-domain, its important to concentrate on the single-domain business problem to deliver the most value in that area. You might then drive 3 MDM programmes which share a common MDM technology infrastructure which will support all three forms of master entity information, but the business really doesn't need to know that. Asking whether you need to technically do multi-domain MDM is like asking someone when they buy a manual* car whether they want all three pedals or just one from accelerator, break and clutch. So lets move on from technical questions and start focusing on the business outcomes.

About the author

Steve Jones
6 Comments Leave a comment
The original idea of MDM was to make sure that certain data, needed in more than one (set of) process, was always aligned and actual. So: get rid of all local copies, sync-processes and what have you.
If we are going to have different MDM solutions for different purposes, then we are back to square one, albeit in a different database.
Therefore multi-domain MDM isn't MDM; it's old-skool in a new skin.
stevejo's picture
Henk,
Its not about different _technical_ MDM solutions more about different MDM governance models. Technically they could all end up in a single platform but what it is saying is that the governance that you put on Customer onboarding is different from the governance you put on supplier onboarding. Some elements might be shared (e.g. Postal address validation).
The point I was trying to make was that MDM fundamentally has to be multi-(technical)domain as you can't be consistent at just an entity level you need the relationships to really present the managed business view.
However the governance, funding and other models are not related to the database definition of "party" or "asset" but to the business governance challenge which must be aligned to the business problem being solved. Technically driven MDM approaches help to improve data warehouses but tend not to deliver the operational improvements that deliver the real value.
This is an excellent article that aligns with my ideas on implementing MDM for E&U.
MDM is about mastering a information based on the operational/functional needs of the business. Not organizing db tables. It requires a technical component as a backbone or tool to support business drivers that are enforced through data governance.
Internal departments slant the same data all the time for many different reasons but, getting information from disparate data sources is one of the issues (local data sources). Integration is another. Silos are developed because its built into integration patterns. Business gets lost at this point. They don't want to hear technical jargon regarding large number of integration mappings, translations, or sequences. They want one place to see a customer, order, product, finance, asset, etc. MDM simplifies it for business when presented as multi-domain masters.
In my designs business does have a place in driving multiple domain mastering since they define customer, product, services, finance, and assets for one simple reason: a need for a common voice to its customer and partners that translates to revenue. Most companies don't do that today very well and it cost money.
My recommendation to business when discussing multi domain MDM is envision it as a shared information as a service (IaaS) component relative to the very domain required at the point of need (like up-selling, ordering, customer interactions, product development, etc). Extract those informational needs from a single source be it customer (party model based), product (dev-to-runtime based), financial (code based) and more. Who defines these requirements? Technical people? I think not.
MDM starts from a functional business need and cascades to IT portfolio. Just as most technical solutions do in today's business world. The maturity of an organization to handle MDM becomes a matter for the business to resolve. Not IT.
I do not think truly MDM will exist -- it is one of those kind of technologies that is in the portfolio of products (of a large organisation) because it is a new "trend" -- the idea is fabulous but the implementation is horrendously expensive and sometime even unrealistic -- multiple business domains (from a functional persective) it is very difficult to fund -- most of the transformation programmes are funding pieces of business domains -- not the e2e enterprise functional domain then it is very difficult to have an e2e information blueprint with multiple business domains relations -- it is the similar to the situation faced when designing an enterprise DW vs a particular business domain datamart -- the other concern is INTEGRATION -- like everybody who has been long enough implementing integration projects -- this multiple MDM sychronisation/integration of multiple data sources and/or master data (residing in multiple applications) is not a easy task and sometime performance and data latency is a BIG restriction
Then, yes is good to have a business driven thinking of modeling and conceptualising multiple business entities -- but it is extremely important to be technical realistic when comes to data movements and (near) real time requirements (specially for operational requirements).
I respectfully disagree. One of the key benefits of multi-domain is the maintenance process. By entering as well as storing a "thing" once instead of for each domain there is inherent maintenance value as changes occur which benefits the business avoiding multi-entry/maintenance of changes. Additionally from a reporting perspective the ability to see the multiple relationships have value. In a recent study of MDM implementations only 16% of current MDM implementations are single-domain.

Leave a comment

Your email address will not be published. Required fields are marked *.