3. Business Architecture - 1980s Paradigm
Business Data Model
represents stable and
unchanging view of
“what the business
knows”
Business Function Model
represents stable and
unchanging view of
“what the business does”.
Every company in an
industry should have the
same basic models.
The basic models remain
stable over time.
Business Modelling Industry Model
4. Repertory Grid technique
Grid
Component
Definition Example
Topic Subject or situation Special Offer
Element Three contrasting
instances
Bread, Baked Beans, Chocolate
Construct Something that
differentiates two
instance from the third
Shelf-Life - easy for customer to store
Necessity – Luxury
Rating A way of characterizing
the elements in terms
of the construct
Bread is a necessity with a fairly stable demand and a
relatively short shelf-life.
Customers are not likely to increase their consumption
of bread, and are therefore not likely to be attracted by
BOGOF offers
In the 1980s, this technique
was used to build a conceptual
model showing entities and
their attributes. We expected
this model to be pretty stable.
Veryard 1984
5. Retail Industry Event-Response Model
Before Loyalty Cards
Undifferentiated
mass of customers
NOTE
Customer
Buys Stuff
Stock
Check
Stock Low
Stock High
Sell
Customer
Stuff
Restock
Store
BOGOF
to all
Customers
CBDI Journal Feb 2006
6. Retail Industry Event-Response Model
With Loyalty Cards
Ability to identify
customers
Can differentiate business
by customer
NOTE
Customer
Buys Stuff
BOGOF to
specific
customers
Stock Low
Sell
Customer
Stuff
Analyze
Customer
Behavior
Restock
Store
CBDI Journal Feb 2006
7. Extending the scope of the analysis
Include customer’s
process as well.
Change scope of model
On what basis does
customer differentiate?
Use repertory grid
NOTE
Sell
Customer
Stuff
Analyze
Customer
Behavior
Create
BOGOF
Restock
Kitchen
Buy
Stuff
Evaluate
BOGOF
Customer Store
CBDI Journal Feb 2006
8. Retail Industry Information Model
Before Loyalty Cards
Product
Transaction
Outbound
Transaction
(Till)
Stock
Check
Inbound
Transaction
(Supply)
Supplier
*
*
CBDI Journal Feb 2006
9. Retail Industry Information Model
With Point-of-Sale (POS) and Loyalty Cards
Customer now appears as
a key entity.
Stock check ceases to be
a primary transaction.
NOTE
Product
Transaction
Outbound
Transaction
(POS)
Inbound
Transaction
(Supply)
Supplier
*
*
Customer
*
CBDI Journal Feb 2006
10. Abstract Metamodel of Business (extract)
What the business knows
information
intelligence
(what is going on)
What the business does
capability
response
event
CBDI Journal Feb 2006
Note how business intelligence
and analytics can now trigger
additional business operations.
What the business knows follows
from what it does and what the
environment does.
With loyalty card and other
technologies, the business now
knows more than before.
11. Issues for Business Systems and Processes
Differentiation
• Cost-effective
differentiation,
based on a
rigorous
analysis of
differentiation
requirements.
Process integration
• Retention of
rich information
from one
process step to
the next.
Real-time enterprise
• Convergence
between
information and
intelligence
13. Alternative
Model of
Operational
Capabilities
Customer
Management
Product
Management
Store
Management
Staffing &
Resources
Customer Purchase
Segment
Brand
Supplier
Local
Community
Location
Facilities
Layout
Employee
Role
Shift
Promotion
Capability Areas Domains
Context
Career Path
Supply Chain
Management
Stock Item
Location
Warehouse
Transport
How do we choose
between alternative ways
of decomposing a retail
operation into discrete
chunks? One way is to
consider how independent
is each chunk, and how
much coordination is
required. Any given
decomposition raises
cross-cutting concerns.
14. In a joined-up
business, these
capabilities need
just enough
coordination.
Customer
Management
Product
Management
Store
Management
Staffing &
Resources
Consumer
Supplier
Customer
Employee
Coordination
This line denotes the
dependency between
capabilities. There may
be different degrees
of dependency.
15. Intelligence from Differentiation
(Requisite Variety)
We may differentiate
customers (and their
behaviour and intentions)
according to a number of
factors, with greater or lesser
granularity.
And we may also wish to
differentiate suppliers,
locations, and any other
significant business concept.
Each degree of differentiation
increases the complexity of the
process.
Under the right conditions,
increased differentiation may
produce better outcomes for the
organization and its customers.
Under the wrong conditions,
differentiation merely adds
complication, increases cost and
risk, and may produce worse
outcomes.
Differentiation Consequences
16. Operational Differentiation
Zero variation.
No differentiation between customers.
One size fits all.
Fixed segmentation.
The retailer identifies a number of (fixed )
market segments, and assigns each
customer to the appropriate segment.
Dynamic deconstruction .
Differentiation based on the detailed
actions and inferred intentions and
context of customers.
Degrees of Differentiation Retail Example
Customer
Management
Customer Purchase
Segment Context
17. Fixed segmentation: many industries have a
simple way of classifying customers
Strategic Information Classification – Banking Example
CUSTOMER
COMPANY INDIVIDUAL
LARGE CORPORATE
SME
SOLE TRADER
HIGH NET WORTH
ORDINARY
LOW NET WORTH
18. Customer service with fixed segmentation
A different set of services is
provided to each category of
customer.
Correct categorization of
customer allows efficient and
low-risk differentiation of
services.
BUT customer category
may become functional silo.
NOTE
CATEGORY
SERVICE CUSTOMER
19. The retail industry is leading the way to
smarter differentiation
Anonymous Customer
Customer Identified at
Checkout (Loyalty Card)
Customer Behaviour
Tracked Inside Store
(RFID)
?
Retail Example Your Industry
20. First let’s look at differentiation within one
capability area
Focus on the most relevant differentiators.
Sufficient range of responses to differentiators.
Coordination between variety of perceived
differentiation and variety of response.
Feedback loops to improve relevance and
accuracy of differentiation.
Feedback loops to refine responses.
Progressive elimination of unnecessary or
irrelevant complication, along with exploration
of new opportunities
Success Factors of Effective
Differentiation
Customer
Management
customer
operation
customer
intelligence
attenuation
amplification
21. But each capability may
have some intelligence
customer
operation
customer
intelligence
attenuation
amplification
product
operation
product
intelligence
attenuation
amplification
store
operation
store
intelligence
attenuation
amplification
staffing
operation
staffing
intelligence
attenuation
amplification
intelligence
coordination
operation
coordination
So how do we
join all this up?
23. Using Repertory Grids to
Identify Integration Pathways
Topic Subject or situation Buying
Element Three contrasting instances Food, Wine, Coffee
Construct Something that differentiates two
instance from the third
Responsibility
Rating A way of characterizing the
elements in terms of the
construct
With coffee, the supplier/importer may be
handling some of this complexity rather than
the retailer, and hiding it inside product
packaging that says "product of more than
one country".
Topic Subject or situation Buying
Element Three contrasting instances Food, Wine, Clothing
Construct Something that differentiates two
instance from the third
Seasonal Patterns
Rating A way of characterizing the
elements in terms of the
construct
Wine has a buying process similar to clothing
in that you buy a seasonal batch, often a long
way in advance, to a specific "design" but you
can't go back for more of exactly the same
product.
24. Modelling Challenges
Which business events are
we going to respond to?
Which process/capability is
response-able?
What variety of response
are we going to support?
Aligning variety of response
with variety of demand
(“requisite variety”).
Coordinating responses.
Aligning complexity of
processes / capabilities
with the capacities of the
organization.
Monitoring outcomes to
ensure effectiveness.
Developing capacity for
progressive improvements
over time.
25. Practical Questions
How do formal management information systems
support differentiation and integration?
• Within a single capability area
• Across multiple capability areas
How do informal communication mechanisms
(including social networking and interpersonal
trust) support differentiation and integration?
• Within a single capability area
• Across multiple capability areas
How does the structure of the organizational and
systems environment affect the “intelligence” of
the enterprise?
How can CIOs, enterprise architects and others
help the enterprise to leverage its intelligence?
How can we communicate differentiation
and integration to the enterprise
Where these differentiations and
integrations are, or need to be.
How to measure the differentiation and
integrations
How can we demonstrate and monitor the
change?
Do we have an opportunity to derive
measurement of quite free flowing
collaboration through BPM and social
networking capabilities?
27. Four Types of Customer Behaviour
Comparison
• Choosing between
different solutions to
the same demand
Destination
• There is only one
place for the
customer to go.
Cost Convenience
• Choosing between
similar standardized
offerings
Custom
• Supplier adapting the
offering to the
customer’s
requirement
SupplyInfrastructure
Capability Requirement
MicrosoftArchitectureJournal
AsymmetricDesign
28. Narrow Focus or Broad Focus?
Big Data Smart
Retail
Store Design?
Customer Service?
Customer Attitudes?
Hi. My name is Richard Veryard and I’m going to be talking to you this afternoon about Business Architecture. In particular I’m going to explore some of the changes that have taken place in the retail sector over the past 25 years or so and look at the implications for other industries.We’ve got quite a few people on the webinar this afternoon, so we’re going to mute everyone while I’m talking. If you want to ask a question, you have a button to raise your hand and a panel to type your question. We’ll probably take most of the questions at the end, but I’ve got a guy from Unicom here who will nudge me sharply in the ribs if he thinks I need to stop and answer something straightaway.
In this webinar, I’m going to talk about a fictional retail company and its development over the past twenty years. Any resemblance to any real retail company in the UK is entirely in your own imagination. There are some huge simplifications in this story, especially if we think of the enormous range of challenges facing any real retail company, howeverlarge or small, but I hope it will give us a rough idea of some areas where it may be helpful for business architects and others to develop new ways of thinking structurallyabout business.There are a few things I want to explore here in particular.Firstly, I want to look at how the retail business has changed. Much of this change has gone along with a growing sophistication in the use of Information Technology, but IT alone doesn’t explain some of the particular things that have happened in the retail sector.Secondly, I want to look at how these changes have introduced greater complexity into the retail business. Some architects seem to think that complexity is a bad thing, and that the agenda of architects is to eliminate all signs of complexity. But I think that ignores the benefits of complexity - giving the organization more power to operate effectively in a complex demand environment. So the agenda here is not to eliminate complexity but to respect it.There are some important structural issues here, that I don’t think are particularly well represented by the traditional methods used by business architects. So we shall need to look at business structure from a range of different viewpoints.And thirdly, I want to look at what this implies for the intelligence of the retail organization – both operational intelligence and strategic intelligence. It is all very well to have an extremely refined intelligence in one area of business operations, but that’s not so good if it means other areas are being neglected.Notes:http://www.guardian.co.uk/money/blog/2009/aug/17/tesco-clubcard-reward-schemes
There is a view of business modelling that was heavily promoted by the centralizing methodologies of the 1980s (such as Information Engineering) and isstill accepted uncritically by many business architects. According to this view, there is a general function model that represents what the business does, and a general data model that represents what the business knows. These models are unaffected by the kinds of change experienced by the business, including technology change and organization change.Conventional approaches to information analysis need to alter to support the characteristic features of SOA, including support for event-driven and context-sensitive services.Systematic approach to discovering context and opportunity areas for differentiation will allow the business to make informed choices on the level of support they need today and in the future. Informed choices drive practical levels of generalization. Remember SOA is a framework for flexibility, it is not necessarily delivering unlimited flexibility. So better information analysis can facilitate business change. In the third part of this series we will examine how we use the knowledge of context and differentiation based on a better understanding of information in order to support inherently flexible processes.
Let’s start by dividing a retail operation into a number of discrete capability areas, each focusing on a discrete domain of interest.We wish to define capability areas that each contribute to the viability of the whole enterprise by having some value exchange with the environment.Each capability area can then be managed semi-autonomously.
We may compare alternative ways of decomposing the retail operation according to this criterion: how independent is each chunk, and how much coordination is required. Any given decomposition leaves some cross-cutting concerns. For example, does Supply Chain Management count as an autonomous capability area, or is it a critical cross-cutting concern?Many architects skip this step, and base their models uncritically on the most obvious decomposition, without exploring alternatives.
What counts as “environment” depends how we draw the boundaries of the system. Thus for the purposes of this model, we may regard staff as an external stakeholder class – the retail organization creates value by establishing job opportunities and career development. The long-term viability of this capability area depends on things like being able to develop store managers and buyers without poaching from competitors.Each capability area may focus on a different class of primary stakeholders. Thus each capability area can be regarded as operating a single-sided market; the retail operation as a whole is modelled as a multi-sided market.In order to make this model work, we may need to introduce a conceptual separation between consumer (e.g. the child who eats the breakfast cereal) and the customer (e.g. the parent who buys the packet).Remember: This is just a modelThe Map is not the Territory
A simplistic model of operational capabilities regards the stakeholder as part of the environment.A more sophisticated model recognizes that the stakeholder may be an active participant in the requisite capability. This is particularly obvious in the case of supply chain management, although it applies to all capabilities to some extent.As we shall see, the management of the supply chain raises some interesting questions about shared intelligence - how are supply chain requirements and problems resolved, how do the supply chain partners collaborate around planning, problem-solving and innovation, and what are the knowledge flows and communication mechanisms that support these collaborations.Does the fact that Supply Chain Management is a distributed capability / responsibility weigh against our modelling it as an autonomous capability area?
All other capability areas are dependent on coordination capability. We need to understand these coordination requirements.Within each capability area, we have a potential trade-off between the economics of scale and the economics of scope.For example, simple economics of scale within product management suggest stocking large quantities of a very small number of products from a small number of suppliers to obtain maximum discounts and minimum transport costs. However, this conflicts with the need within product management to offer a broad range of products (economics of scope). There are also trade-offs between capability areas. For example, a simplistic approach to the economics of scale within product management will conflict with a simplistic approach to the economics of scale within store management.In practice, it is impossible to optimize all these economic factors simultaneously. So we need a coordination mechanism that allows for a reasonable accommodation between these semi-autonomous areas, as well as a sense of the economic and organizational cost of this coordination.
We may differentiate customers (and their behaviour and intentions) according to a number of factors, with greater or lesser granularity. Each degree of differentiation increases the complexity of the process. Under the right conditions, increased differentiation may produce better outcomes for the organization and its customers. Under the wrong conditions, differentiation merely adds complication, increases cost and risk, and may produce worse outcomes.There are arbitrarily many degrees of differentiation – these are just typical points on the curve.
Let’s look at differentiation within a single capability area – Customer Management. We may differentiate customers (and their behaviour and intentions) according to a number of factors, with greater or lesser granularity. Each degree of differentiation increases the complexity of the process. Under the right conditions, increased differentiation may produce better outcomes for the organization and its customers. Under the wrong conditions, differentiation merely adds complication, increases cost and risk, and may produce worse outcomes.There are arbitrarily many degrees of differentiation – these are just typical points on the curve.
With business intelligence, we have the opportunity to select the most relevant forms of differentiation, based on statistical analysis of characteristic features. In many situations, what the business really wants is to use the past to predict the future. We only want to differentiate customers based on their past buying patterns if this provides a good predictor of their future buying patterns.When I lived in Central London, I sometimes used to visit a hairdresser on the King’s Road, The King’s Road is full of hairdressers, and the staff often stand in the street handing out leaflets to likely customers. When I was walking towards my hairdresser, I would walk past several of these touts, but I wouldn’t be recognized as a potential customer because I didn’t have a smart haircut. After my hair was cut, I would be given loads of leaflets because I then fitted their preconception of what a customer looked like – but of course I didn’t then need a second haircut. This is the kind of stupidity trap that many backward looking sales operations fall into – trying to sell lawnmowers to people that already have lawnmowers.Business intelligence helps us find alternative differentiators. What are the characteristic features of someone who needs a haircut, or might buy a lawnmower? We can then differentiate the services based not on a fixed set of differentiators, but on a current (and periodically updated) set of differentiators, and we constantly review the predictive power of these differentiators. This means that the underlying business type model now needs to be constructed at the meta level, with DIFFERENTIATOR and CHARACTERISTIC FEATURE as business types.
The introduction of the loyalty card represented a radical strategic shift for the large retail chains. Stores now had a formal basis for recognizing a customer as the same again. They can identify customers, and collect and analyze data about the behavior of specific customers. And they can use this analysis to differentiate the response to different customers. For example, different customers may receive different special offers. Amazon is of course well-known for its pioneering work in providing targeted information and deals based on a customer’s browsing and buying history, and creating new forms of associative information which may be reflected back to the customer.Obviously if the retailer can identify the customer as she enters the store, then this differentiation can be done as the customer browses, rather than only when the customer comes to pay. This is relatively easy to implement with online shopping (for example through the use of cookies); and there are various mechanisms that might achieve the same result in a physical store – perhaps face-recognition software in the store camera, or loyalty cards with RFID chips – if the obvious privacy concerns can be allayed.And if there are RFID chips on the goods and RFID scanners on the shelves, then the customer can be presented with information based on the stuff that is already in the shopping basket. Again this capability is very easy to implement for online shopping, and this stimulates retailers to build an equivalent capability for physical shopping. (This is an instance of an increasingly common generalization pattern: take a simple innovation from one channel and look for ways of implementing it in other channels. We then need to define a general structure for the data from different but equivalent sources, and render the data through a common service interface, to enable aggregation and comparison of data across channels.)
Effective differentiation is a function of the intelligence embedded within the customer management capability. The greater the “quantity” of intelligence, the greater the capacity to differentiate effectively.For the sake of analysis , we may regard “Intelligence” and “Operation” as separate sub-capabilities of Customer Management.We can then identify the signals that pass between “Intelligence” and “Operation” , and the attenuation and amplification mechanisms that govern these signals.
What are the events and information flows that help to join up the retail operation as a whole?Where is the strategic knowledge of the enterprise located, and how is it continuously development and effectively used? What are the mechanisms to support innovation and organizational learning?
There are many different pathways for joining-up the business.How does a jar of organic baby food get managed alongside other jars (pasta sauce)?alongside other baby goods?alongside other organic produce?Wrong answer #1 – pick one of the above. Adopt hierarchical managementWrong answer #2 – pick two of the above. Adopt matrix management. Wrong answer #3 – swing erratically between different alternatives (oscillation)Correct answer – manage and coordinate all the relevant pathways. Adopt complexity management.Note – each of these pathways may be associated with a different capability area, and mobilizes a different kind of intelligence.
Alignment ChallengeHow is the differentiation aligned across different capability areas? For example customer differentiation x product differentiation?HypothesisThe first task is to develop intelligence in each autonomous unit.Only then can we start to worry about the intelligence of the whole enterprise.Thus intelligence is an emergent collaborative bottom-up development rather than a directed top-down development.Lawrence, P., and Lorsch, J., "Differentiation and Integration in Complex Organizations" Administrative Science Quarterly 12, (1967), 1-30.In this paper Lawrence and Lorsch develop an open systems theory of how organizations and organizational sub-units adapt to best meet the demands of their immediate environment. Organizations must balance differentiation and integration to be successful. Those companies who manage to achieve high sub-unit differentiation and yet still maintain high integration between sub-units seem to be best equipped to adapt to environmental changes.Groups that are organized to perform simpler, more certain tasks (e.g., production groups) usually have more formal structure than groups focusing on more uncertain tasks (e.g., research and development).The time orientation of sub-groups is primarily dependent on the immediacy of feedback from their actions. Thus sales and production groups have shorter time orientations than R&D.The goal orientation of sub-units is based relative to the part of the environment that affects them the most.http://faculty.babson.edu/krollag/org_site/org_theory/Scott_articles/lawren_lorsch_cont.html
What is the status of these models? Are they supposed to be a simplified description of the current state of the enterprise (AS-IS)? Or an idealized blueprint of some future state of the enterprise (TO-BE)?Enterprise architects typically produce both AS-IS and TO-BE models, and then produce transition plans to get from AS-IS to TO-BE. This is therefore an exercise in requirements engineering at the enterprise levelThese models can therefore be regarded as a TO-BE architecture to some extent, but that is not their only purpose. There is a collectivemanagement need to make sense of those existing conversations and collaborations that need to become more systematic and focused, and so the model is a sense-making tool as well as an engineering tool.Obviously we can install more sophisticated mechanisms if that's justified, but often it's about making better use of the existing mechanisms. I see this as part of the job of enterprise architect – not just fixing the mechanisms but understanding and advising on their effective use.Thus my purpose here goes beyond conventional EA practice. I want to improve the manageability and intelligence of the whole organization as a complex sociotechnical system of systems. I am therefore interested in the management system - the information flows and coordination mechanisms and feedback loops - rather than the operational system.The operational model is therefore a map to support this kind of investigation, as well as a sense-making map to be used within the management system itself, rather than just a blueprint for reengineering the enterprise.
Determining which business events each process/capability is going to respond to, and at what level of granularity.Deciding what variety of responses each process can support. (This is not just a question of the logical design of the process but also the capacity of the organization to carry out the process.)Aligning the variety of response with the variety of the target events. It is clearly a waste to collect information that you can never make effective use of, and it is also a waste to build variety into a process unless you are capable of detecting the events that require this variety.Coordinating the responses of different processes across the enterprise to the same event (or connected events).Implementing these processes into the organization , so that the collective cognitive capacities of the people (for example the quantity of detailed information that they can utilize effectively) are fully exercised (stretched, developed) but not overloaded.Continuously monitoring business outcomes to verify that the variety build into the process is justified by business results.Evolving a portfolio of responsive processes over time, allowing for progressive improvements in differentiation and integration. (The target events themselves may be fairly stable, but the technology for detecting these events may get better and cheaper.)Allowing a loosely coupled organization of autonomous sociotechnical units to emerge, with open and democratic (?) governance.
The traditional retailer acts as a hub in the food supply chain, aggregating food supply from fields and factories, and distributing food to workshops and private kitchens. This is essentially a positional strategy: the retailer seeks to establish and maintain a strategic position within a value chain, as the bottleneck/hinge point between upstream and downstream. Within the positional strategy, the business drivers are understood in terms of the economics of scale and the economics of scope.But if we shift from a value-chain perspective to a service-oriented perspective (effects-ladder), we can see that the retailer is providing a service (=delivering value) downwards as well as upwards – it is a food distribution platform for farmers and manufacturers as well as a food supply platform for consumers and catering companies.So instead of drawing the merchant in the middle, we can draw the merchant as a new kind of platform providing various kinds of market interaction.This takes us from a positional strategy to a relational strategy. No longer just focused on the economies of scale and scope, the relational strategy emphasizes how economies of governance are generated in relation to two kinds of demand context. The big question for a company such as Wal-Mart is how to balance the exploitation of each of these forms of asymmetric advantage.
The names for two of these four ways came from research on shopping behaviour by Gary Davies in which he distinguished ‘cost’ convenience (choosing between similar standardised offerings) and ‘comparison’ behaviours (choosing between different solutions to the same demand). The other two separated out circumstances where there was a demand particular to the customer, distinguishing offerings in which there was only one place for the customer to go ( ‘destination’), or in which the supplier would adapt the offering to the customer’s requirement ( ‘custom’) within their context-of-use. The point being made, however, was that the ‘destination’ form of offering required asymmetric governance because of the need to hold power at the edge of the organisation. What distinguishes this position in the cycle?If we think of the original development of pc-based spreadsheet programs in-house (destination), they soon became a limited number of alternative branded solutions (comparison) that in turn became dominated by the one offering all the others’ features in one package (cost). From here we have seen an increasing ability to customise the ways it can be used (custom) to the point where we are now looking at a new cycle of web-based solutions that we can build one-by-one (destination).http://www.asymmetricdesign.com/2006/04/managing-over-the-whole-governance-cycle/