Inside Architecture

Notes on Enterprise Architecture, Business Alignment, Interesting Trends, and anything else that interests me this week...

December, 2012

Posts
  • Inside Architecture

    Steps To Create a Core Diagram

    • 0 Comments

    To be fair, these are steps to create a solid understanding of the architecture of a business, but the deliverable is a core diagram, so that’s the title of the post.  I first wrote about a method for creating core diagrams about a year ago, as I was preparing for a talk on the subject at the Open Group conference in San Francisco.  Now that I’m preparing for another Open Group conference, I find myself filling out some of the details from the previous effort.  Most of the text below is copied from an e-mail that I sent to a fellow business architect who was asking about how to create a core diagram.

    The text below describes a five step process

    1. Collect a list of your organization’s business models
    2. Create or leverage a taxonomy of capabilities that reflect differentiation in business processes
    3. Differentiate each capability on the basis of Ross’ operating model taxonomy (level of Information Integration and level of Process standardization)
    4. Make an educated guess about the operating model of the company
    5. Draw the core diagram and build understanding around it

     

    Understanding how to create a core diagram starts by collecting a list of the business models that your organization performs. Each business model is unique and different from the other ones. Each will require different capabilities and will often drive variations in those capabilities for the sake of market or product differentiation. You cannot create a core diagram effectively without the list of business models.

    So what is in a business model? I’ve blogged about that fairly well. A business model is a composition of elements that describes how and why a value proposition exists, who it is for, and what it drives in terms of internal and external requirements. The diagram is below. (click to enlarge)

    Metamodel for a Business Model

    Once you have the initial list of business models, you will want to engage with direct business stakeholders. Make sure that they understand the concept of a business model, and what makes a business model unique from other business models (e.g. selling the same product in the same way to the same people in another country is NOT a unique business model, but selling a product in three different ways to three different, potentially overlapping market segments within one country probably represents three business models). Engage. Build relationships. This is your first shot.

    Once you have that list fairly well baked, you have step two on your hands: a capability taxonomy that reflects process differentiation. In this case, it is a good idea to start with a high level process taxonomy like the ones made available for free from the APQC. I don’t know if there is one for financial services yet, but there should be. If not, you can start with a general one, but it will take some editing. You want your capability taxonomy to be worded in such a way that it represents “things that could be done” without reflecting the way in which they are done. For example, “customer identity management” is OK, but “customer deduplication” is not, because we want to make sure that customers have an appropriate identity but the organization may not want to remove duplication in order to do that.

    This requires some editing of a large list of items in a hierarchy. Excel is OK for this. There may be other tools as well… I haven’t experimented past Excel. This is the second point where it is good to be engaging with business stakeholders. Get their help to describe their business model to you in terms of capabilities, and make sure that all of their capabilities are included in your taxonomy somewhere (usually around the third level down in the tree).

    Step three is to differentiate each business capability on the dimensions suggested in the EA As Strategy book. (This can be done at a high level. If your taxonomy has more than 200 business capabilities in it, don’t use the most detailed level(s) of the taxonomy. No one has patience for the details in a core diagram.

    Draw out a grid like the one illustrated in the EA As Strategy book, only make it empty.

    Diagram illustrating the dimensions of Operating Models with Integration (low and high) on the Y axis, and Standardization (low and high) on the X axis

    In each one of the boxes, write in the capabilities that are well understood by a particular business stakeholder, then go to that stakeholder and validate your choices. Make sure that you have placed the correctly for that stakeholder’s particular business models. Note that very few stakeholders will have a valid opinion about capabilities that are NOT part of their particular business model, so don’t show capabilities that they don’t care about.

    You will quickly discover that most folks agree on some things and disagree on some things. Where a single capability shows up in multiple businesses, one stakeholder may say that it needs high standardization, while another will say that the capability needs low standardization (== high flexibility). Take note of these disagreements. THEY ARE THE VALUE POINTS FOR BUSINESS ARCHITECTURE.

    On everything you can get reasonable agreement on, go ahead and create a master table that has the capabilities differentiated in the manner above. That will probably be about 90-95% of your business capabilities in your taxonomy.

    Step four is to make an “educated guess” about the operating model that your organization has. It’s a guess because most organizations are difficult to read and no one person will be able to answer your question about what the company as a whole looks like. Most of the time, you can start with the generalizations that Jeannie Ross made when describing the four operating models in her book “Enterprise Architecture As Strategy.” If there are a large number of capabilities in the “High Integration, High Standardization” box, you can suggest that your organization is a “Centralization” model. If, on the other hand, there are a large number in the “High Integration, Low Standardization” box, you can suggest that the organization is a Coordination model. This is the educated guess part because there is no good formula for making the guess. By this point, you will know a great deal about the organization so your guess is as good as your stakeholders.

    Step five is to take a cut at your core diagram… Draw it out and then work with your stakeholders to get common understanding.

    For each of the four styles of models, there are different styles of core diagrams. The centralization model tends to break out capabilities by functional area since there is very little (intentional) duplication. So it will be a diagram with a series of functional areas as boxes with the capabilities for each function listed in the boxes. Good idea to put the name of the person accountable for that business function in the title of the box. Lines between the boxes represent flows of information or value between them.

    The Replication model is somewhat similar. There will be some functions that are owned by “corporate” while the rest are replicated into EACH of the operating units, so there will be two large “areas” on your diagram. The corporate area will have some functions with capabilities in them, and a single “replicated” area will have the remaining functions with capabilities in them. This is wildly valuable to business planners because they can get agreement among the leaders of each replicated unit about what each one of the is accountable to do and what they MUST depend on the corporate unit to do.

    The collaboration model tends to be “hub and spoke” with the hub being the most integrated capabilities and the spokes being unique to each of the business models (or in some cases, small groups of business models that share a lot of capabilities). The lines tend to be information flow, not value flow. The capabilities in the spokes are usually duplicated between the different business units but they (should be) the capabilities that each business unit needs in order to differentiate itself or its products in the marketplace.

    The diversification model is the most complex because the “corporate” unit tends to have a small number of core capabilities (often just financial ones) with each of the subsidiaries having a nearly complete and quite independent set of functions with massive duplication of capabilities across them.

    I hope this gives you a good start in creating your core diagram.

  • Inside Architecture

    Everything you’ve read about IT Project Failure is wrong

    • 8 Comments

    I did a scan around the web to figure out what many of the leading thinkers were saying about IT project failure and the root causes.  Numbers varied between 20% and 80% of projects failing to deliver on their business case.  The root cause analysis that follows from these failure numbers spends a lot of time looking at the IT project, but most forget to look at the causes of failure that are outside the project’s control.

    In the diagram below, the central blue box represents causes of IT project failure that are inside the control of the IT team.  As you can see, there are many more causes of IT project failure that are OUTSIDE that blue box than inside it.  Yet, countless articles have been written on the factors INSIDE the box.  I think it is time we take a slightly wider lens to the problem!

    image

    The factors outside the project are as important, or more important, than the ones inside the box.  I have worked on many projects over the years, and if I look back at the ones that ended up cancelled or scrapped, the reasons were not ones in the blue box.  They were usually ones from the top box: where the project should not have begun in the way that it did.  Let’s look at these six factors:

    1. Lack of accountability for success – Sometimes, in an organization, an interesting situation will arise.  I call this the “can’t lose” scenario.  In this scenario, the CIO will suggest that the business can benefit by doing some particular initiative, and his or her people will set out to find a business sponsor for the initiative.  In this case, the sponsor will reap the benefits of the initiative, not the CIO.  (Let’s say that the CIO has suggested a Business Intelligence initiative, and signs up the Chief Marketing Officer to sponsor it).  If the project succeeds, the sponsor wins.  If the project fails, the CIO (not the sponsor) loses credibility.  After all, it wasn’t the sponsor’s idea.  The CIO dreamed it up, after all.  In fact, the sponsor may not do any real “sponsoring.”  He may sit in an occasional status meeting and check his e-mail while the program manager he assigned from his team answers questions… and keeps him protected from any blowback.  The sponsor in this case is not accountable for success.  No one is.  The odds of this project succeeding are small.  (Note: a good IT Governance system prevents this condition). 
       
    2. Failure to get buy-in from above – Often, a senior executive will ask his operational leaders to suggest IT projects for the planning cycle.  The executive (let’s say Tina Smith, Vice President of Sales) assumes that the leaders (William, Ashok, and Lee: all three are General Managers) will ask for projects that align to their individual strategic needs.  One of those General Managers, William, decides to create an IT project to implement a SOA service bus.  He wants the bus because he is investing in replacing an application and his IT team tells him that a bus is a good idea.  After all, if everyone in the Sales group uses it, everyone will benefit.  But William doesn’t get Tina Smith’s buy in.  He implements his SOA service bus, but when subsequent IT funding requests come up to move other sales applications over to the bus, his peers Ashok and Lee are not interested.  Their priorities don’t include moving to the bus.  William bought a white elephant.  Because he didn’t get Tina’s buy in, none of his peers will reap the benefits.  Is the SOA bus a failure?  Well… it’s not a success.
       
    3. Misaligned prioritization – Lee James, General Manager of Sales for the Western region of Contoso owns a sales strategy for the enterprise.  The strategy is to make the sales force more productive by moving the company over to a new CRM system.  The CRM system has out-of-the-box support for a long list of reports, but his team tells him that he shouldn’t use those.  Instead, he should put in a Business Intelligence system in the cloud.  Lee wants to trust his team to pick the right solution.  They tell him that the cost of this system is $4 Million over two years.  The company is moving to a new strategy that de-emphasizes the direct sales force.  Instead, the company will be relying on social networking and direct eRetail to make sales.  To make the new strategy work, there are eight projects totally $8 Million that need funding.  There is a competition for dollars in the IT budget.  Lee goes to bat for his $4 Million commitment.  It’s important because “we are already here, so we have to complete the job.”  This is an example of misaligned priority.  The new BI project can be safely cancelled or scaled WAY back to support a strategy elsewhere, but that other strategy doesn’t belong to Lee.  The one that he owns needs money and he’s going to keep fighting to fund it because his boss, Tina Smith, isn’t aware of the tradeoff.  Lee fights, and Contoso loses.  Which project will fail?  Both of them.
       
    4. Slow decision making – This one is related to the other problems in that it is often, but not always, caused by a poor governance system or poor prioritization of the effort.  In this scenario, the team implementing the project needs the sponsor to make a decision, often one that requires consultation with his or her peers or counterparts in another part of the company.  The sponsor may be unwilling, unable, or incapable of having a difficult conversation.  On the other hand, they may be indecisive.  Regardless, this situation can cause serious delays in a project.  For example, Ashwin’s application sends EDI-formatted messages to a health insurance provider.  The provider had indicated that new fields are allowed in the messages that are very valuable to Ashwin’s line of business.  However, he has to upgrade his EDI translation software to get the new fields.  The EDI translation software is also used by the Finance group to send banking transactions.  Ashwin knows that the finance group will get a benefit out of the upgrade and wants them to help cover the costs.  This upgrade will add $25,000 to the cost of his project and he’s on a tight budget.  The software team offers to write a small component that will insert the fields into the data stream, but its complicated and fragile. Ashwin cannot decide if he should ask the finance team for funding or if he should upgrade the software.  The project team cannot proceed without a decision. 
       
    5. Lack of authority to drive change – This one is quite common.  Ashok reports to Tina Smith, VP of Sales for IBuySpy.  He has told her that he wants to take the lead on one of her strategies: to consolidate all of the eCommerce systems in the company to a single outsourced vendor.  Tina things that this will reduce costs all around, and Ashok is happy to take on the challenge.  His business analyst does a survey and it turns out that the services and support team offers customers the right to buy replacement parts online.  In addition, the “outside services” team allows customers to buy a service contract on their own area of the website.  The main sales site, of course, uses a high-volume low-cost vendor that Ashok is comfortable with.  Ashok sets up meetings with the owners of the other applications, but they choose not to provide the needed transaction cost information or the cost estimates of changing their internal systems.  His data is light but he strongly suspects that there is a good business case for switching.  Unfortunately, he doesn’t have the data to prove it.  Without data, the Tina won’t take the case to the CEO to require the other teams to jump onboard.  The strategy fails.  Neither Ashok, nor his manager Tina, have the authority to require their peers to adopt a lower cost system.
       
    6. Lack of influence to rally peers – In many organizations, the prevailing attitude is that the boundary between “centralized” services and “federated” services is intentionally fuzzy.  In these companies, there is no policy that requires a function to be centralized vs. federated.  For example, if a company has ten divisions, three may have their own financial processes, while the other seven share a common finance unit.  The decision to use the central finance unit, or to own a federated unit, can be made and unmade depending on who the leaders of the business are.  (Note: leaders change, and with them, these decisions change as well).  This kind of policy is good for flexibility, but creates both inefficiencies and often, confusion.  This is simply a tradeoff in the design of organizations.  It is neither right nor wrong.  In this environment, any business can create a “centralized” function by convincing more than one business leader to come together to create a shared resource.   The flaw comes when a group wants to create a “centralized” (or simply “shared”) function, but they lack the influence needed to get multiple businesses to join the initiative.  Difficult questions of shared funding can be demotivating, and the political need to create “alliances” can be difficult for junior executives that, in many ways, are competing with one another for the next rung on the career ladder.

     

    Each of these conditions has the potential to kill an IT project.  I would suggest that MANY IF NOT MOST of the failures of IT projects can be traced to one or more of these conditions, but these conditions rarely get counted in the statistics for “Causes of IT Project Failure.”  Why?  Because, in most cases, projects that suffer from these conditions are either never funded, or are reworked so that the political problem is simply avoided.  The project business case does not reflect the problem, so the criteria for failure (doesn’t meet the business case) is never met.  Efforts are made to avoid (but not address) the problem before the business case is written!

    This is the world of the Enterprise Architect.  These are the kinds of “failure” that fill the eyes and ears of an Enterprise Architect.  If an EA focuses on only these six causes, he or she will deliver real, tangible, and unique value to their enterprise without ever overlapping the roles and responsibilities of an IT architect, business analyst, or technologist.

Page 1 of 1 (2 items)