by Andy Hayler

Hand master data decisions back to the business

Sep 26, 20114 mins
IT LeadershipIT StrategyTelecommunications Industry

In its latest research The Information Difference found that the master data management (MDM) market grew by 31 per cent over the past year.

This is much more than a rebound from the depressed conditions of 2009, and reflects my own experiences in the market in the last year or so.

I have been getting enquiries from relatively conservative companies about how to go about starting MDM, while also seeing a number of companies bite the bullet and convert their initial pilot projects into broader enterprise deployments.

A number of factors are at play here.

There has been a realisation that despite all the investments in ERP, most enterprises struggle to answer basic questions about how profitable their customers, products and channels are, and the level of risk exposure they have to suppliers or counter-parties in trading.

Core to this problem is the diversity of competing master data around customer, supplier and product: a typical large enterprise has nine competing sources of product data and six of customer data, and many firms are in even worse shape.

In order to get to the heart of the master data problem it is critical to get the business to reclaim ownership of its data assets and to stop delegating this responsibility to the IT function, which does not have the authority (or possibly the knowledge) to sort it out.

This reclaiming of ownership has resulted in a dramatic rise in the level of interest in data governance, with joint business and IT data governance groups assigning responsible business individuals to decide which version of critical shared data is actually the ‘master’, and hammering out the processes and authorities required to get something done about it.

Data governance, crucially, includes data quality within its scope. This is vital because poor data quality is another key barrier to effective business insight: just having a consistent set of data definitions is of limited use if the actual data is wrong, incomplete and out of date.

Once data governance processes have been put in place then the foundation for a successful MDM project is laid. Next it is crucial to properly document the state of key master data around the enterprise: what are the source systems that currently generate this data, which ones are in fact the most trustworthy, and what is the state of data quality in these respective systems?

It is important when setting the scope of an MDM initiative not to bite off more than you can chew, so pick off a few critical areas of shared data: maybe customer or product, perhaps supplier of location or asset.

Ideally start with the ones that are giving the business the most pain right now, as making an improvement in these areas will provide tangible benefit. Similarly, pick a limited geographical or business line scope for a pilot. Agree the key shared data in the pilot, document it and choose a technology (including data quality tools as well as an MDM hub) to try out.

These days there are plenty of good competing MDM technologies, from vendors large and small, at a variety of price points.

Put enough resources into the pilot to make sure that the project is successful. It is vital to show early success in such initiatives. If clear business benefit is demonstrated then other parts of the business will be keen to adopt the ideas and approach.

Take the time to review the lessons from the pilot, make sure that the approach in terms of project management, process and technology is working, and then extend the pilot gradually into other areas of the business, prioritising ideally the areas suffering the most from inconsistent and poor data.

This gradual phased approach is the one that I have seen succeed in MDM projects. I witnessed one ‘big bang’ MDM project with 150 full-time external consultants grind to a halt under the sheer weight of its ambition. Proceeding in a phased manner, with small mini-projects gradually building on earlier success does take time.

For a large enterprise it may well take two or three years to roll out and fully implement a global MDM initiative, but if the project is delivering incremental business benefit along the way then this is not necessarily a problem. MDM is a journey rather than a destination: on-going processes have to be put in place rather than quick, one-off fixes applied.

The recent rapid growth in the market suggests that MDM, which as an approach to data management barely existed a decade ago, is finally moving from its pioneers and early adopters into the mainstream.

Andy Hayler is founder of research company The Information Difference. Previously, he founded data management firm Kalido after commercialising an in-house project at Shell