Without Data Management Standards – Anarchy!


When I posted a blog on Don’t Establish Data Management Standards (it was also on Information Management's website as Data Management Standards are a Barrier) I expected some resistance.  I mean, why post a blog and not have the courage to be provocative, right?  However, I have to say I was surprised at the level of resistance.  Although, I also have to point out that this blog was also one of the most syndicated and recommended I have had.  I will assume that there is a bit of an agreement with it as well as I didn't see any qualifiers in tweets that I was completely crazy.  Anyway, here are just a few dissenter comments:

“This article would be funny if it wasn't so sad...you can't do *anything* in IT (especially innovate) without standing on the shoulders of some standard.” – John O

“Show me data management without standards and good process to review and update them and I'll show you the mortgage crisis which developed during 2007.” – Jim F 

“This article is alarmingly naive, detrimental, and counterproductive. Let me count the ways…” – Cynthia H

"No control leads to caos... I would be amused to watch the reaction of the ISO engineer while reading this article :)." - Eduardo G  (I would too!)

After wiping the rotten tomatoes from my face from that, here are some points made that get to the nuance I was hoping to create a discussion on:

“I supposed more of us would agree that standards are not carved in stone, are subject to review and change thus the real barrier is the inflexibility in adapting standards to new requirements...” – Osama S

“…lack of a data management standard for what constitutes valid data for a customer, or a product - that is something that you ought to have defined as a standard, otherwise your consuming business processes are going to suffer, as well as analytical processes that depend on it.” – John E (I like this because it addresses aligning and possible adjusting standards to meet information use.)

To be clear, the intent of the blog was to surface that the standards we have in place today or the way we use standards today to manage data may be a barrier.  The level of resistance and horror voiced to the blog indicates to me the degree at which standards in fact do create barriers by being held up to such high (dare I say it) standards.  

What I think some missed (after getting excited over the blog title) is that you have to account for the fact that standards do and will change. For example:

  •  ICD-9 to ICD-10 diagnostic code changes
  • Waterfall to SCRUM development
  • Decentralized to centralized to hybrid data organization evolution
  • IT Cost control data strategy to business lead data strategy
  • Centralized to federated data management
  • Traditional RDBMS/ETL to Hadoop Solutions
  • CRM as the single source of truth for master data to linked master entity definitions across business units

The premise that you need standards to have consensus and consistency is not what I am arguing against.  In fact, it is often not that there are no standards in place for data management, but rather those that do exist don’t meet the needs of the business or data managers.  What I see all too often is data management and architecture approaches, decisions, and strategy changing only when the business has proven why data management should change (or failing that, the business goes around).  For example, in a client round table discussion one enterprise architect ask me if I was seeing the business standing up their own Big Data environments.  The response around the room from other enterprise architects was that they were definitely seeing this. 

If you aren't convinced yet, here is when you know existing standards are a barrier:

  • Expected data management cost will rise more than 5% this year
  • You can’t keep up with data requests
  • Your data organization can’t focus on strategy
  • You can’t measure data management to business value
  • Data governance is unsustainable or IT lead
  • The business is going around you
  • You are giving out login rights to feeder systems of warehouses
  • You expect the business to work within your process, not theirs

In the end, if there is only one standard that should always be in place – continuously assess your standards and change proactively, not reactively.  Have the courage to break from tradition when it is holding you back.

So, rather than talking in black and white terms, what standards are you finding don't meet current business needs?  Or, how do you adapt?


Some standards matter, and others.....

I can recall working on "Database Standards" several times in the past. Each time was interesting, perplexing and ended with some standards that surely everyone thought would make things better. In hindsight I think some made sense, while others were useless.

Let me explain. The naming of objects in an RDBMS for many objects is something I've done for a variety of objects. If the object is exposed and meant to be reused, like a column then a standard makes sense, unless the standard is ambiguous. More on that some other time. If the object is internalized and nobody sees it except the DBA then it doesn't matter. Think triggers, table spaces, rules, etc....

My point is simply that standards need not cover everything, just the things that are exposed, reused and have value.

Exposure, reuse, value

Mike - I like the perspective of focus and who needs the standard. Great example.

Something that I think needs

Something that I think needs clarifying is that data management standards aren't the same as database management standards. The prime purpose is managing the development of a formal 'language' that will be used to control and inform business (i.e. data). For example, how to define a new type or classification of customer about which information will be exchanged.

In the past such control was achieved using paper based systems of information, for which data management was just as essential. The management principles, rules and processes were developed and maintained by business design specialists rather than by technologists.

Data management is and should be promoted as a key business control concern. If business management are 'going around' then they haven't got it and aren't taking sufficent ownership.

Does the business "get it"

If I hear you correctly, I think what you are suggesting is that technologists take ownership of managing principles, rules and processes in order to enable in a system. Thus switching from spreadsheets and word docs to policy management systems, metadata management and business glossary tools, etc. Or, at least better management and reuse of rules and definitions within MDM, Data Quality, and ETL. What I see is that companies are still making the transition to improved creation of libraries within data management/integration tools (the latter example). Although, the challenge is that reuse may be falling short because they are developed and managed by IT to support integration and data process and often miss the context element. There is a shift to adoption of business glossaries and metadata management tools as companies have taken steps toward definitions and taxonomies. Although, I still see a lot of wiki based management. Data policy management solutions are still early and I have yet to see wide spread adoption here outside broader GRC solutions. Moving to a standard for collection and repeatable use of policies and rules in an automated fashion is absolutely the way to go. If IT can both facilitate and step up to manage this in these systems, the business will be appreciative.

To your point on the business doesn't get it or won't take ownership, I don't know that I agree with that. What I see on the business side is that they go around because they are taking ownership. What they don't want to take ownership of is the "technical" aspect of data management. They don't want to work in developer tools to manage data. They don't want to talk about tools and projects. They want to focus on getting value from their systems. They want to get value fast. When they go around, vendors are supporting this in the businesses mind (perception - right or wrong doesn't matter, the marketing and sales works). It will be a long hard lost battle to try and make the business "get it" and conform to IT standards and practices. Better to learn what vendors are doing to support outsourced data/information needs and collaborate with the business on better ways to align Business and IT processes.

Does the business 'get it'

Michele, over years of consulting in data management what I found is that when data is just seen as a kind of information system 'plumbing', management of it can become a technical inconvenience that gets in the way, including of some vendor's 'ideal' solution.

On the other hand when business management truly understands the role of data as effectively defining and controlling the world in which the business operates, the value of having a standard 'language' takes on a different meaning and they may then fully engage in data management.

This is enabled by business and IT specialists working as a united team, preferably using a common data management framework which will have some business and some technical aspects. It doesn't work very well in an us and them, business vs IT culture, which will often be divided and conquered by vendors.

When data matters

Ron - that makes perfect sense and I whole heartedly agree. We have evidence of this in research that if the business (executives in particular) value data strategically and competitively, business outcomes occur. It definitely breaks through us vs them scenarios and data strategy is driven by business strategy. Keep posted for our data governance report on this coming this quarter.

Depends on what sort of standards

I understand this clarification on the previous article and I agree with the content. Ron has a good point on what I would call the "bureaucratic" type of standard, that is present to comply with some forgotten strategy or procedure but that in fact doesn't add any value. Yes that is possible in an organisation and it happens quite often. A standard adds value when it integrates, enhances coherence, improves the quality and enriches data. Add value by "purifying" the resulting information hence the quality of the business decisions. Also improving communication between parties as a result of speaking the same language.