In summer 2014, a journalist named Esther Honig asked a number of people around the world to Photoshop her picture, to make her "beautiful".
- Some users made her skin colour lighter; others made it darker.
- Some added business clothes; others added jewellery.
- Some made her hair longer; others covered it up.
The lesson learned is that without providing a definition of "beautiful", many outcomes are generated.
Whether you call it a data dictionary, a data directory, a business glossary or even a lexicon - there is a common aim of defining agreed terms for your organisation in such a way that everyone knows what they are referring to.
Yet, not every organisation has a business data dictionary. Fewer will have a complete one (whatever "complete" means). Fewer still will have their dictionary in active use.
Why is this?
Let us be in no doubt, creating a corporate data glossary (that's the sixth term I've used so far, I’ll stick with “data dictionary” for the rest of this piece) is difficult.
One approach is autocratic - a single definition is set from an ivory tower, and the glossary is complete. And unused.
Creating a data dictionary requires input from all your stakeholders. There will be debate. There may well be some who disagree with the final definition. It will take time.
We talk about it for twenty minutes and then we decide I was right. (Brian Clough)
Some decentralised organisations will need an online workflow to get things done; others may be able to set up a weekly group meeting which goes through little by little until completion.
As a data quality expert, your key challenges are to:
Get the project going, and keep it going
Ensure that the right people are in the conversation
Have sufficient content in the dictionary
Get the project going, and keep it going
The payoff from a good data dictionary is hard to isolate, as it combines with the benefits from wider data governance and quality activity. This can mean that an organisation seeking to reduce spend may pick on the data dictionary project as one that can be deferred or shelved.
So, even when you have got the project going, you must maintain a laser focus on seeing it through; or risk adding it to the pile of incomplete business analysis work.
Ensure that the right people are in the conversation
You obviously need subject matter experts. You also obviously need those who are senior enough to make decisions. You also need those who will be using the data dictionary.
If you did not capture every opinion when creating a definition, then changes and updates can always be made as part of your “business as usual” work.
If you do not generate sufficient enthusiasm amongst the business teams to use the dictionary, then what you have created is of much worth, but little value.
Have sufficient content in the dictionary
A common response to the data dictionary project will be that we "only need a few key items". If you were to do this, then you don’t have a data dictionary, you have a crib sheet. On the other hand, if you can identify the “few key items” for all your stakeholders, it will become apparent just how many key items there are.
And with that, you can have a data dictionary that is not only beautiful, but useful too.
Until next time,
Charles
A common response to the data dictionary project will be that we "only need a few key items". If you were to do this, then you don’t have a data dictionary, you have a crib sheet. On the other hand, if you can identify the “few key items” for all your stakeholders, it will become apparent just how many key items there are.
And with that, you can have a data dictionary that is not only beautiful, but useful too.