Launching a Data Governance Initiative for the First Time: Interview with Emma Fortnum

How do you create the momentum and strategy to launch a data governance initiative for the first time?

That is one of the core themes of this interview with Emma Fortnum, an Information Architect at Mitchells and Butlers, a leading UK based operator of restaurants and pubs in the UK.

Emma has been involved in implementing data governance in one of the UK’s largest utility companies, and she was responsible for the enterprise modelling efforts for a multi-national insurance company.  Her goal is to put the information back into Information Technology, and she is a passionate advocate of putting data back into the hands of the business areas that own it.

Dylan Jones : How have you been trying to sell the business case for data governance and data quality in your organisation? What practical initiatives have been effective?

Emma Fortnum: I’m taking a two-pronged approach in this effort.

Firstly, I’m trying to generate a buzz around data quality and data governance – if I can get people Googling the terms, then that will be a win! At the moment, it seems to be the preserve of a few people.

I just want to get my company thinking that data quality isn’t something that other people do, it’s something that all of us have a role in playing.

I’m doing this by putting together a presentation per department that says what data governance is, why you’d want it, and what the consequences are of not doing it.

I’ve had good feedback so far, but I’ll admit I’ve started with IT, because I know best what arguments work on IT people! I know others will disagree that IT is the right place to start, but I’m working to the rule ‘Start with what you know’.

Secondly, I’m putting together a plan for the kinds of efforts and deliverables that would be needed in order to actually get a data governance effort off the ground. What I’m trying to avoid is a situation where I’ve generated a buzz and got people excited about it, but then have to turn around and say “Yes, but there’s no budget to do anything until 2015”. That would be counterproductive, to say the least.

So I’m working hard at selling the concept of a data quality review to at least the IT department, so that they can put some money aside for next financial year to back up the buzz I’ve hopefully generated this year!

Dylan Jones : How do you typically go about creating findings on the wasted effort through a lack of data quality or data governance?

Emma Fortnum : This is probably the hardest part of the job, because this is building the business case for data quality/data governance.  Basically it’s keeping my ear to the ground, but knowing where data quality problems often present themselves.  Integration, reporting, and migrations are always good for highlighting issues with data, so it’s getting to know the people responsible for those areas and following up with them any time something gets mentioned.

For example, recently I was in a team meeting, nothing to do with data, but someone happened to mention a data error that has arisen as a result of an organisational change.  So I went and talked to them about what was happening and what we might be able to do to fix it.

But it’s more than just having the conversations, it’s using them to build the business case, and try and quantify in pounds and pence just how much poor quality data or a lack of data governance is costing our business.  It’s by no means easy, because doing this legwork is often done in my own time.  It’s a bit chicken and egg – since I haven’t yet proven the case for data governance, there isn’t an initiative that can investigate – but I can’t prove the case without doing at least some of the investigation first!

Dylan Jones : You have a data architecture background, how has that helped with your data governance and data quality goals?

Emma Fortnum : Good question!

I think what my architecture background allows me to have is the ‘big picture’ approach and focus.

Data doesn’t respect operational boundaries, which is why I’m so keen on producing enterprise-level data efforts, as opposed to starting small and working your way up. Don’t get me wrong, I don’t think for a moment that I can tackle all data issues at once, but I think in terms of identifying and subsequently assigning data stewards and custodians, you need to have the entire picture first. Otherwise you run the risk that the decisions you make have to be overturned later on. If you can be as sure as possible that each small step you take is getting you towards the overall goal, then in my opinion that is the best approach to take.

I think what my architecture background also helps me with is the discipline and the logical thinking that is required when I’m creating data models and business rules. I understand how to build and maintain databases, therefore I’ve seen firsthand the kinds of problems that a poorly-designed system produces.

In my data modelling efforts, therefore, I call on this experience to get my business stakeholders to define constructs precisely, and to agree on terminology.

It can hinder, as well though – I have a technical background, and I understand systems. When it comes to the sales pitch to various stakeholders, that’s where I have to get support from my peers and colleagues, because it’s the part of the job I find the hardest. I am learning though!

Dylan Jones : Do you think that disciplines like data modelling are becoming a forgotten art in businesses or is it something that is still being well managed and maintained?

Emma Fortnum :  Ah, a pet subject of mine!  I definitely think that data modelling is becoming a forgotten art!  And since I have a data modelling background, this really upsets me, because you don’t produce a data model just for the sake of it.

The thing is, I see data modelling happening all around me – every time an IT person needs to explain a relatively complex concept to the business, they start sketching data models – which is fabulous.  The unfortunate part is that the vast majority of people who do this have no training in data modelling or data normalisation, so they take that ‘fag packet’ communication diagram and try to physically implement it, which leads to no end of problems!

I have found that drawing the ERD isn’t the hard part – the hard part is precisely defining the terminology that is used, and getting business agreement on it (actually, it’s getting the business agreement that is hard, time-consuming, and painful!)  This is the part that those amateur data modellers don’t take the time to do.

The whole point of producing data models is that it gives your entire business a common language and a common framework of business rules so that misunderstandings don’t occur as often, and data doesn’t get co-opted into representing something other than  the original concept.

Of course, the fact that you have a data model doesn’t guarantee success, but at least the exercise of creating one has forced your business and your stakeholders to think about how they are trying to represent their business concepts, and you have reference documentation that may or may not be used physically within system development.

From an IT perspective, an enterprise data model makes integration between disparate systems so much easier, but  there’s a lot of business benefit to having one as well.

Dylan Jones : Regarding your point about building a common framework of Business Rules – there is often a lot of discussion (and confusion) around the difference between business rules and data quality rules – how do you approach this?

Emma Fortnum : That’s a really good question.

Business rules are really easy to understand but data quality rules less so.  Simplistically, I believe a data quality rule to be the metric that your business owners have decided they wish to achieve in terms of a data quality score.

So if you wish to have 90% of your customer’s mobile phone numbers, then that is a data quality metric.  If you wish to say that a customer can only have one mobile phone number stored at any one point in time, that is a business rule.

You can’t have a data quality metric without a corresponding business rule, but you can have a business rule that you don’t wish to measure an associated data quality rule.

Dylan Jones : What does your year 1 plan look like for Data Governance and Data Quality – what goals are you looking to reach?

Emma Fortnum : For Year 1 I’m being quite realistic. If I can get my organisation interested in data quality and data governance, that’s my main goal. To back it up, I’m also – funding allowed – going to produce some foundational documentation to go with the buzz.

By this I mean produce architectural artefacts – data, process, application, and integration diagrams, and the mappings between them all, as well as some standards and guidelines for how data should be treated, both in BAU processing and project work.

I would much rather set up a data governance effort with the backup of these kinds of documents than try and manage data without having this in place.

I also don’t think it’s fair to assign stewardship and custodianship to business stakeholders but not tell them what it is they’re now responsible for. And despite whether it’s right or not, data modelling (and process and application modelling) is still largely an IT preserve.

Dylan Jones : How have you structured your data governance programs in the past? For example, how have you implemented stewardship and accountability?

Emma Fortnum : I’ve implemented data governance programs in both a structured and an informal way. The structured program was a large project that involved specialist consultancy effort, multiple workstreams and deliverables, and the end goal was a BAU data governance function within the business. That approach worked for that business, which was very large, had a relatively stable business model, and enough people and revenue to justify that kind of approach.

I’ve also implemented data governance through a more softly, softly approach.

Smaller companies often just don’t have the money (or necessarily the desire) to put in place a robust data governance effort all in one go. For my current company, whilst I think the business model is stable, and the size of the company may warrant it, their data isn’t seen to be causing them enough of a problem that an all-out data governance effort is needed. For my current company, I’m approaching it in a more organic effort.

I’ll also try and learn from the mistakes of past efforts – for example, choose your stewards and custodians carefully! Some individuals might seem like a logical choice, but if they have no interest in data, or governance, then even with the best benefits case in the world, they won’t do what they are supposed to!

Also, if you do decide to make someone a steward or a custodian, then bake this role into their job descriptions and their objectives. People tend to do what they are bonused on – leverage this! Now, to see whether I can convince HR of this…

Dylan Jones : Finally, if you think back to earlier data governance and data quality initiatives, what were the things that really made the difference on the project? What key pointers would you share with organisations looking to get started for the first time?

Emma Fortnum : The biggest thing for previous efforts was executive sponsorship.  You hear this all the time, and it really is true.  You can’t have an effective data governance effort without buy-in at the very top level.  Top tip – convince an exec.

Use whatever leverage you can to gain their buy-in, for example:

If your company is concerned about Information Security, then frame your arguments to show how your governance efforts will reduce risk – such as by implementing auditing, proving data provenance, or adhering to legislation.

If your company is concerned about Big Data, then frame your arguments to show how your efforts will increase the value of those types of analytics – such as by knowing your customers better, being able to integrate multiple datasets, or having a better handle on social media.

If your company is concerned about saving money, then frame your arguments to show how your efforts will contribute (eventually!) to the bottom line – such as saving money on project effort, pointing out how much time and effort is wasted on checking and double-checking reports because they’re not currently trusted, or avoiding fines.

However you do it, get executive sponsorship, because data governance efforts rarely seem to be a bottom up effort.

Dylan Jones : Thanks for your time today Emma, I’m sure our readers will find your insights valuable.

Emma Fortnum : No problem, it’s been a pleasure.

About Emma Fortnum

Emma Fortnum is an Information Architect at Mitchells and Butlers plc, in Birmingham, England.  She received her B.Sc. in Computing from Loughborough University in 1999, and started her career in Information Technology with the Hereford and Worcester Chamber of Commerce as a Management Information Officer.

Her career has spanned fifteen years and has always concentrated on data and information – everything from report writing to database design to application and data architecture.  She has been involved in implementing data governance in one of the UK’s largest utility companies, and she was responsible for the enterprise modelling efforts for a multi-national insurance company.  Her goal is to put the information back into Information Technology, and she is a passionate advocate of putting data back into the hands of the business areas that own it.

When not attempting to make business decision-making more data-focused, she enjoys quality time with her family and friends.

Her LinkedIn profile can be found at uk.linkedin.com/pub/ emma-fortnum/12/376/155‎.

Image rights : Creative Commons Flickr, Ewan-m

About the Author

I am the editor and founder of Data Quality Pro. 20+ years experience of data quality initiatives.

Leave a Reply 0 comments