John Owens provides Business Systems Modelling advice from his Integrated Modelling Method
Business Systems Modelling : Introduction to Business Functions
with John Owens
Master practitioner John Owens, creator of the Integrated Modelling Method, is our guest expert for this webinar discussing techniques for business modelling and its key role in ensuring long-term data quality.
In this initial session John dispels the myth that processes create data and answers the question if processes do not create data, what does?
The answer is Business Functions and John goes on to explain the difference between the terms:
Poor business function design is one of the principal causes of poor quality data today. Learn how to reverse the trend in your organisation and create high quality data through effective Business Systems Modelling.
Dylan Jones: Hello and thanks for joining us for another Data Quality Pro webinar.
Today I’m really excited because we’re actually kicking off a series of webinars that I feel are long overdue in the data quality profession.
As a data quality practitioner, one of the recurring issues I discovered when meeting companies of all sizes, across all sectors is that in most cases there was a huge disconnect between how their business functioned and how this translated to the design of their systems, workflows and data models.
This is why on data migration projects you would find scores of hidden spreadsheets and databases, squirreled away on workers desktops because the central system just didn’t have the procedures and functions the workers required to complete their daily tasks.
It’s why users over time develop their own data “hacks” to store the facts about equipment, people, locations and orders that the original system design no longer caters for.
It’s why over time the data quality gradually degrades, creating increased cost of maintenance, lost opportunities, inability to compete and scale, the list is endless.
But how should we define these functions, procedures, models and business requirements correctly so that we get it right and build systems and data stores that reflect how we want to do business?
How can we carry out gap analysis between that new system IT want to procure and the functions and data that drive our business now and in the future.
Well, tonight’s guest is going to help us address these and some other issues.
John Owens is an expert in Business Systems and Data Modelling and creator of the Integrated Modelling Method or IMM, a unique modelling approach that totally integrates all business models into a unified and usable framework.
This series will be broken down into 3 sessions as follows:
Session 1: Exploring the different terminologies, looking at the difference between processes and functions, procedures and mechanisms
Session 2: How do we derive data requirements from Business Functions so that we build on a solid foundation from the start
Session 3: Logical data model – what is it, how do we create it and why it is pivotal to eliminating data defects.
To frame session 1 this arose from one of our members talking about how their processes were creating data and your view John was that this is fundamentally flawed.
John Owens: This was a conversation in a blog postIsaw where one of the practitioners in data quality was talking about how people ought to be aware of the data created by business processes.
This made me go… Ah… Ah… because business processes do not create data and people will say everywhere but they do “it’s the process that creates all of our data” and this is one of the main problems, there is a complete misunderstanding in many areas of the difference between a what is a process, everybody will use the word every day and you say what is a process and they’ll say “that’s what we do” and that’s not true either.
And it’s because the different terminologies, and it is worthwhile knowing about them, the same way as a builder will know all the different names of the tools and the different facets that go to build a house. In business and data modelling we ought to be really strict about what we use.
And processes do not create data, what creates data is functions. So I think it’s probably worthwhile having a discussion about these basic terminologies because if they are misunderstood then there is no way you are going to end up with good data quality when we are starting from the position of using the wrong terminologies, we’re using the wrong foundations, obviously to try and build quality so I thought it would be very useful to have that discussion.
Dylan Jones: Absolutely I think if we can all agree on a common definition around these things which is often very hard to do in the data quality profession but I think this is an area which I don’t think is covered very well at all really.
So I think if you could start by explaining some of these different terminologies and definitions that will be a great starting point.
John Owens: Ok, I think we should start off with the four things we should look at.
The first is function then mechanism, process and procedure.
Now function, everything starts with function and the business is all about function. And what is a function?
Well a business function is really an activity or a coherent set of activities that a business must perform in order to meet its objectives and continue in existence.
So it’s pretty fundamental and all other things everything that a business does on a day to day basis is a function.
It could be something like, to actually receive an application or receive an order from a customer, this is a function. To dispatch an order, to bill a customer, these are functions.
The other thing that goes along with function is mechanism.
Now most of what we do in a business is doing the business functions but when people talk about them this is where the initial confusion comes about, they will talk to you about how they do them and how you do functions are called mechanisms.
Now I will explain the difference there. For example suppose you were receiving an application for employment then you could do that in various ways and the various ways will be a mechanism you could take a phone call and that would be the function you would be performing would be to receive an application for employment.
You could receive a fax; you could receive the application over the internet. So all of these different mechanisms and yet one function.
People have asked me in the past how do you tell the difference [between a Function and a Mechanism], the way you tell the difference is by asking what is the objective of the activity.
So if you say the activity is filling in an employee form, an employment application form, the objective isn’t to fill in the form, the objective is to receive an application for employment.
What’s the objective of taking the telephone call, well it is in fact to receive an application for employment.
So that’s one of the basic places we start, knowing the difference between functions and the mechanisms.
Another important definition of functions (and this is really the one that actually accelerates growth in the business) is a business function, when you are modelling them, they are a definition of what it is the business ought to be doing.
So it’s nothing to do with, if you going out to model business functions do not go out a model what the business does now. Model what the business ought to be doing.
And this is again we will come back later on when we will be talking about business processes, and the as-is and to-be, this has got nothing to do with that.
Basically we’re saying if you are going to model what the business is about then you ought to model what it ought to be about.
If it’s about building houses or ought to be about building houses and it’s in factpapering walls you don’t model back if it’s doing the wrong thing you don’t model it because it will never get you doing the right thing.
So [that was] business function, the next one is the business process. Now these are two that are interchangeable. People will talk about business functions and business processes interchangeably but one thing I will stop at.
Business functions, this is something in fact very widely misused. When people talk about business function they actually mean, they incorrectly refer to a department in the organization, they talk about the finance function when they mean the finance department or the personnel function or the HR function and this is wrong.
If you mean the personnel department, the HR department say the personnel department or the HR department. When you mean the finance department say the finance department.
Function is all about action. If you mean when you say the finance function if you mean all of the functions needed to be carried out in order to look after the finance of the company then you are talking about the finance and it wouldn’t be singular, it will be functions.
So then what is the difference between a function and a process?
Well, a process is really the definition of the sequence in which functions need to be carried out.
So in response to a business trigger, what is the order in which the business functions need to be carried out in order to arrive at a preferred outcome in the business? That is a process.
A process is just the order of execution of business functions. Every step in a process is a business function. And it is these steps in the business process which is based on functions that create or transform data.
So all data in an organization is created or transformed by functions, not by processes, because processes actually don’t do anything, processes are just sequence.
It is the functions that “do the doing”.
And we finally come to procedure, so what is procedure?
Well procedure is to process what mechanism is to function. So when you are modelling the procedure you will in fact be modelling a sequence of mechanisms by which the functions in the process are carried out.
So again in the procedure you will have things like fill in the application form, file this into the HR system.
Dylan Jones: Absolutely, going back to your point before about that you believe when we talk about functions we should talk about what the business should be doing as opposed to what it is currently doing.
And I guess this is something we see time and time again with the data cleansing for example on data migration projects, integration projects where they are just continuously cleansing the defective data but two months down the line the same defects are going to keep coming through.
Is this part of the reason you think why we have so many data quality problems because we are continuously looking at things as they currently are?
John Owens: Absolutely, it is quite crazy when you think so many years on from the inception of computing because you know many years ago when I first started in this thing in this industry when it was in its infancy and there were simple rules when you would actually say before you make sure the code is right, make sure it is the right code.
And so that really means that you can go in, you can look into a piece of code and you can make it much more efficient to make it better and better.
But if it’s actually doing the wrong thing no matter how efficient you make it you will never get it to do the right thing.
And this is very much the same with data quality.
People are going to the data that they currently have in the enterprise and then making it in quotes “better and better” but all they are doing it is making it different.
Is it taking it any closer to what the business requires? The answer is they don’t know because without a function model and without a logical data model they don’t know what the business requires.
So it’s changing it but it’s just changing it from what it is to something else.
And as long as they do that as long as they don’t know what it ought to be, they can never get it there.
Dylan Jones: I see and going back to the point before about mechanisms and processes and functions how does it relate to for example business rules.
When you are listening to say data quality vendors or talk about their product, they talk about how they can encapsulate business rules in their products or translate it into data quality rules, when we talk about business rules how does that relate to those terminologies you just been discussing mechanisms, functions, process and procedure.
John Owens: Well all other business rules need to be .. they are part of the function.
Business rules belong to functions because that’s the only place you should be defining them, the only place it make sense.
Because the business function is in definition of what is the business ought to be doing and in that will be embedded the business rules and in that in a weird way there is no such thing as a data rule.
It will be a business rule based on function and that will result in their being standards or constraints or allowable values for data.
It isn’t an abstract data rule you could say in a way there is no such thing as a data rule.
The rule only applies because the data is needed to support a business function and that business function will have rules.
Dylan Jones: And its part of the problem here then when we talk about fit for purpose and fit for intended use, something’s about because we don’t specify what the purposes are and what the intended use is it’s inevitable that we are going to get data quality problems.
When you look at some of the classifications what data quality is there is so many disagreements about what it is what it isn’t but one of the recurring ones is fit for use.
So is this kind of part of the big problem we have them is that we just not following the ________ [15:52] like IMM and deep function that we just don’t have the knowledge, we don’t have it documented in USO it’s inevitable that we are going to have data quality issues.
John Owens: It is because everybody is saying fit for use but nobody has ever defined what that use is and the only use is to support the business functions of the enterprise.
Again you get back to some of these rules are so simple they frighten people.
The only reason data exists in an enterprise is to support the business functions.
The only data needed in an enterprise is that require to support the business functions nothing more nothing less and there are no exceptions to that.
But people would say no no no that’s too simple and so they have lots of data that’s out there, there is going to be claims to support something or be fit for some purpose and the fact is it is an abstract its illusion, know what the business functions are then you know what the fitness for purpose is.
Dylan Jones: Ok. So whatever we do these webinars and perhaps someone doesn’t really understand data quality management, don’t understand modelling perhaps a business manager and they are thinking hey I’ve never seen any of our team doing anything like this, I don’t know of any groups of the ideal business side who are doing any of these kind of modelling what is the benefit for me as a business sponsor or as a business leader to start adopting some of these best practices.
John Owens: There is one very simple one, until you do you will never going to arrive at data quality.
It’s as simple as that. So one of the benefits is you are going to save millions and millions of dollars.
Otherwise don’t do this you are going to actually create in your business in your enterprise a whole separate industry called data quality.
So at the front end you can be doing lots of things which by virtue of the fact that they’ve never been designed to do what they are ought to be doing, nobody has ever defined what they ought to be doing are or actually going to create data which in fact is not of much use and then you have to create another industry in your enterprise called data quality.
And you are ensuring the continuance of data quality forever.
And in fact in some parts of the data quality industry _____[18:47] but sadly it is true have a vested interest in not solving the quality problems specially upstream because their tools all about putting right what has gone wrong and if enterprises don’t have data going wrong there would be no need for such tools.
So again the emphasis is at the wrong end so major benefit but it might seem to abstract is that to in fact going to do a waive with your data quality issues because you should going to get it right first time but there are all sorts of other benefits on the way.
There is going to be one of this major simplicity you are going to simplify everything, you are going to remove the complexity.
Far too many people think that they have to be clever in this but people back in my days as a programmer people used to come to me and tell me that they have written this amazingly complex program and I would say well keep practicing, one day you will be able to write a simple one.
Because that really is what it’s all about, its bringing simplicity to the business and my mantra is power through simplicity and by building the function model you’ll in fact gain an understanding of your business that you can gain in no other way.
A business function hierarchy when you put it up in front of people allows the business for the first time to really understand their business.
Now business processes can’t do this because they are sort of just slices through the business.
When I am trying to understand a business looking at business processes is the same as trying to understand the structure of a country by looking at motor ways.
All the business process has is in fact a beginning and bits on the middle and an end.
You don’t know what lies outside it so people are trying to understand their business by looking at business processes, and to me it’s like just giving somebody a list of roots across the country you start to run and you end in a Carlisle now you understand what the UK is.
Or when you start at San Francisco and look at Chicago, there are great deal that lies outside that which is your business.
Dylan Jones: We kicked this off by talking about today’s quality management by how doing modelling correctly by adopting best practice we can actually help get of these perpetual wheel of the sort of cleansing, reconciling data, fixing cleansing consulting and going round and round.
But it also sounds as though If I am into this if I am a business sponsor or business leader it sounds almost like you are talking about framework really where I can open up my business for the first time and really understand because you are looking at future processes looking at how it should be functioning it’s almost like a business, almost like a measurement tool or an evaluation tool so that it gives you a clarity that you probably haven’t had before.
So if your business isn’t performing as you would hope it to is this a technique that can actually help that?
John Owens: It is. You can actually you know, one of the things which you talked about when you mentioned that there was gap analysis and you can actually, just take slides there back one of the things that popped into mind when your business is going out and spending out millions of dollars on a new developments, bringing in new systems and they get to bring in this systems in the firm belief that they are going to actually answer all their problems but they don’t know how because they have not able to do a gap analysis.
Without a function model you can’t somebody is going to spend tens and hundreds of million of dollars on an ERP and they are going to say that solves our problem but what is it that it’s meant to do?
Well it does the same for everybody else. So we will just use it but what is it your business ought to be doing and will this application, will the system support it?
Without a function model you can’t do that gap analysis.
From the function model you will in fact build a logical data model and we will talk about that in a later seminar without the logical data model, you can’t do gap analysis on the data structures.
You can’t say will the data structure implemented by this system support our business.
So it is hugely, it’s all about getting it right first time.
When we are talking about data quality, sadly data quality is still in the stage where total quality management or quality was a long time ago when it was all about quality control when you used to have people standing at the bottom end of the back end of production lines and spotting all the defective things coming down the line at the end of a very good day they felt really pleased because they’d found so many defects in the manufactured products and they felt proud to themselves.
And then people started asking a question why would you spend all this money building defects into your products and then all that money trying to find them.
And that’s what we are doing in business at the moment spending millions of dollars implementing huge systems that are creating defective data and then spending millions of more dollars trying to find those defects and put them right.
When you put it like that it sounds a bit insane but in fact the thing is how do you stop it.
Will you do what you ought to be doing which is you actually model what it is the business ought to be doing and once you know that then in fact you can define what data and what structure that data ought to have to support what the business ought to be doing.
Dylan Jones: That’s fantastic, Thank you John for some great descriptions there and some great tips as well so that concludes Webinar 1.
Webinar 2 will be shortly coming online on data quality pro and we will be focussing on business functions looking at how business functions can drive correct data requirements and then final session we will be looking at logical data model.
If you have any questions on this session there is a link below the webinar view on your page and you can post some questions into the forum there.
So even after webinar has gone live feel free to post questions there and that makes sure that John sees them and that can respond to any of issues or questions arising from today’s webinar.
Thanks again and I’ll see everyone on the next session.
Image credit: Flickr