LIFE WITHOUT A DATA MODEL
Bill Inmon
Founder, Chairman, CEO, Best-Selling Author, University of Denver & Scalefree Advisory Board Member
LIFE WITHOUT A DATA MODEL
By W H Inmon
The easiest things to sell are those that are –
?? Immediately obvious
?? Inexpensive
?? Provide immediate payback
The hardest things to sell are those that have the opposite characteristics.
This simple conundrum explains why data models are so hard to sell in the corporation.
The truth is that data models have payback –
?? Strategically
?? On a long term basis
领英推荐
Furthermore, vendors don’t like data models because data models can become an obstacle to the vendor in making a sale. The vendor whispers into management’s ear that the data model is just a theoretical exercise and a waste of time.
So let’s consider what life is like without a data model.
Data starts to collect all over the corporation. Accounting has its data. Marketing has its data. Sales has its data. There is a deluge of data that is unintegrated, scattered, and stored in every conceivable location.
One day management asks – what is our projected revenue for this quarter? Sales gives one answer. Marketing gives another answer. And accounting gives yet another answer. So how is management in a position to make some critical decisions? The answer is that management is befuddled. Management has to use intuition, not facts. And that can lead to disastrous consequences.
A good data model would have helped resolve the different understandings of revenue. Sales is talking about projected revenue. Accounting is talking about booked revenue. Marketing is talking about possible revenue. A good data model would make these differences obvious and reconcilable.
So what happens when an analyst wants to find some data and there is no data model? The analyst has many rabbit holes to go down. And even diving into rabbit holes may or may not enable the analyst to find what is needed. And even if the analyst finds the data that is needed, how do the different forms of data need to be glued together in a meaningful manner to suit the needs of the analyst?
And what about the acquisition of new systems? How does the new system interact with existing systems? A good data model would act as a guidepost in order to answer this question. But without a good data model there is no way to easily and accurately predict how a new system is going to interact with existing systems.
And this is the short list of what life without a data model is like.
So if an organization doesn’t care about data integrity, if an organization doesn’t care about accurate data connectivity, if an organization doesn’t care about analysts easily and quickly finding accurate data on which to operate, then that organization really doesn’t need a data model.
The fact of the matter is that the usage of and dependence on a data model says a great deal about the professional maturity of an organization.
Bill Inmon lives in Denver with his wife and his two Scotty dogs – Jeb and Lena. Yesterday Lena cut a trim and a shampoo. Jeb asked this morning – when is it my turn? Jeb is jealous of his sister.
Data Platform Data Architect
10 个月Lack of a solid data model brings to data silos, as Bil said, too often to solve the conseguent data scattering problem fancy solutions wish to be put in place on top of this mess to quickly solve the problem. Unfortunately acting in this way they are only make the problem bigger. I't almost like an ancient motto: you want to close the enclosure when all the animal escaped.
Engineering Data with Passion: Making Data Dreams Come True!
11 个月Is it possible to have one data model for the organisation’s (like one big table) then slice it into smaller model to serve the purpose of different team in the org? Or would creating different data model for different purposes create data integrity issues? Thank you for the post and appreciate any information for my question.
Process Development Engineer II, Dgitalization and statistic analysis
11 个月We don't build parts, we build products, and if the data model is in parts (and duplicated in several places with its own interpretation) then the model will not reflect our value stream. Then it will be as you write, for the management to guess a lot of puzzle pieces that do not fit together. In my quest to convince my management of this, I took the liberty of adding a link to this post as they keep claiming that we have lots of data. Yes, I say, but there are lots of piles of data that you must dig into...
99% Retired - but really enjoyed Data Warehousing, Data Architecture, Data Archeology, and Problem Solving at all levels. It was always about the data; design, lineage efficiency and the people!
11 个月big, big, BIG part of having a data model is that it’s an enterprise central model. Tonnes of little models disconnect and duplicating bits and pieces on different tools scattered everywhere just don’t do the job and invite rabbit hunting. They have to all be part of the same puzzle but structured and organised. This then becomes a company asset, managed in the same context as you would with other inventory so to speak. Conceptual or physical is just another subplot of the same story. Designs for capture and usage are again another subplots. Of course all said this has deviations based on the twists and turns but high level data models are a must.
Chief Operating Officer at DataVaultAlliance Holdings and President of DataRebels
11 个月Bill Inmon Oh so true, Bill. We run into this all the time. "We don't need no stinkin' data model!" The same thing happens with trying to get across the critical and time saving discipline of building out an extended taxonomy and an extended ontology to inform the data model from the business perspective. Thank you for constantly heralding common sense thinking in the midst of the noise. Wishing you all the best!!