Enterprise Architecture in Startups: Is it relevant?
Practitioners of Customer Development, Lean Startup and Enterprise Architecture can all learn from each other. But they shouldn’t enforce their views on each other as there are some incompatibilities. Let’s see how enterprise architecture in startups can exist.
The Startup culture and methods have largely been defined by Steve Blank who wrote The Startup Owners Handbook and later, by Eric Ries who wrote Lean Startup. Both of these consider how newly-created companies can grow quickly and in the right direction for their founders and customers. Many authors and speakers followed, but for this article, we’ll mainly focus on these two.
Enterprise Architecture (EA) functions can be found in many large, mature organisations that have a need to get a grip on their ICT* landscape. The practice of EA allows companies to understand how they perform currently, using which tools, what standards, which customers, which locations, etc. EA then allows for rationalization or consolidation towards a more considered approach. One of the frameworks, TOGAF emerged as a relatively open* standards for defining enterprise architecture within organisations. Most of the organisations using TOGAF will have a complex collection of legacy, COTS (commercial-off-the-shelf) and bespoke systems across multiple locations.
From the very brief descriptions above, you’d think that one has nothing to do with the other. In most cases, you’d be right. I don’t see Enterprise Architecture taught in entrepreneurial programmes. Nor do I see Lean Startup taught or practiced in many large companies. Fortunately, some MBAs are now addressing entrepreneurship, including Lean Startup. But that’s a small section of the university and college courses in the UK. Plus more companies are introducing innovation, sometimes in the form of innovation labs or partnering with smaller, nimbler companies.
Startups and Enterprise Architecture can benefit from each other. To help dissect this, we’ll use Mintzberg’s 5Ps of Strategy. It will provide some link between the two that’s usually missing. First some more background about each:
Let’s start with a definition so that we don’t treat all new companies or organisations as a startup. Steve Blank defined a startup as “an organization formed to search for a repeatable and scalable business model”. It’s important to realize that the startup only exists to find a business model. It has no legacy, other than any legacy the founders bring with them. It will not have a large, multi-tiered operational workforce, nor a mass of machinery, existing protocols and procedures, nor multiple networks of out-dated computers. In fact, the reason it’s a startup is to be free of these constraints.
However in searching for the business model, the founders have to choose an operating model from day one.
This may not be the operating model that will survive into the future. Instead it’s an operating model for how the founders will find the right business model for their startup. That may recursive or a circular argument at first glance, but it’s actually sequentially. Unfortunately the first model is often ignored and the founders just walk into it.
Let’s look at that in a bit more detail.
The 1st operating is the one that’s in place while the founders are looking for the business model that they will implement. This operating model will include roles, tasks, deliverables, partnerships, computers, software. These are all the elements that can be placed on a Business Model Canvas (Osterwalder). However I’ve only ever seen startup founders create a canvas for the 2nd or 3rd operating model, never the first.
Founders share the activities between them and their employee/interns if they have them.
Enterprise Architecture is not just about ICT but that’s usually the initial and most powerful focus. In parallel, to the architecture of buildings in the real world, Enterprise Architecture relies on understanding the current landscape and then setting the foundations for the future. Then the building blocks are put in place.
Implementation of an Enterprise Architecture provides an organisation with the insight necessary to reduce duplication and to understand the impact of proposed changes. For instance, imagine changing the operating hours of a core service. What other services would be impacted? And what other services need to change in order to support that operating hour change?
Enterprise Architecture functions in large organisations can be larger than the total number of people working in a successful, slightly mature startup. The function can generate a lot of documentation and analysis about the current organisation and potential futures.
Typically, there are several strands to Enterprise Architecture:
- ICT Architecture – dealing with the ICT systems, methods, standards. Sometimes networks are separated out to another strand.
- Business Architecture – dealing with organisational structure, partnerships, processes, process standards, policies and procedures
- Data Architecture – Dealing with standards, access to data, data lifecycle, information governance standards and policy.
Process Architecture is often included within Business Architecture.
Mintzberg describing the concept that strategy can be developed in 5 different ways; the 5Ps. The 5Ps present different lenses that strategy can be viewed from.
- Strategy as Planning – large planning exercises, defining the future of the organisation
- Strategy as a Ploy – to act to influence a competitor or market
- Strategy as a Position – to act to take a chosen place in the chosen market
- Strategy as Pattern – the strategy that has evolved over time
- Strategy as a Perspective – basing the strategy on cultural values or similar non-tangible concepts.
The concepts within Lean Startup mean that it doesn’t fit well with Strategy as Planning. Startup founders do not have time to undergo lengthy strategic planning activities. In face, I’d argue that activities of this type should be limited, even in larger organisations. Enterprise Architecture fits well with Strategy as Planning even if only to submit information and to assess the impact of the strategic plans once known (or as they’re being evaluated).
Lean Startup fits better with Strategy as a position, e.g. to be the best high-margin, low-volume in your sector. It provides a focus for the startup. Enterprise Architecture is useful for evaluating how to achieve that position.
Lean Startup also fits well with Strategy as a ploy, e.g. to disrupt a market by introducing a different business model. Again Enterprise Architecture is useful for evaluating how to achieve that ploy, but also for assessing the ability to execute
Startups in general fit very well with Strategy as a perspective, e.g. we are an innovative company, not afraid to take risks. However those practicing Lean Startup concepts may have a harder time until they define validated learning in cultural and/or behavioural terms rather than relying on typical business development metrics. Or instead, the metrics may be business development (e.g. number of clients retained who actively use the product, etc.), but the changes that you need to learn from could relate to your startup’s culture.
I think that Strategy as a pattern occurs more often than we’d like to think, even in startups. However, I can’t imagine a successful startup then breaking through to becoming a global name using Strategy as a pattern. It would be interesting to see evidence on this. Whatever the case, Strategy as a pattern does gel well with Lean Startups (who should have decided their strategies and applied them through Build-Measure-Learn). Enterprise Architecture is useful for Strategy as a pattern for helping define information about the direction to take.
In fact, most Lean Startups aren’t thinking about Strategy in the way that large companies do. The strategy that startup founders think about is how their service/product will operate, not how they will operate before they get there. In fact, most (if not all) startups would fail a capability maturity assessment and that is usually one of the first steps in implementing a Enterprise Architecture (especially if implementing TOGAF through ADM). First you understand if the organisation is capable of defining and adopting its target architecture, otherwise a lot of the effort will be wasted.
While formal tools for Enterprise Architecture are not required, some thought is required into how the company operates now and how it will operate in future.
From the Startup Perspective
The first issue for startup founders is that we don’t have the time to develop Enterprise Architecture in its entirety. We may perform some of the activities but not in the way expected within EA practices. So you may see founders creating canvases, maybe even multiple canvases. Perhaps they’ll have some rudimentary reporting in place, e.g. following Pirate Metrics.
The second issue for startup founders is that the company could be changing rapidly to accommodate the change in scale from acquiring more customers each month. This means that many of the concepts involved in ICT Architecture would have to be rewritten every couple of months. This would take time away from building in order to build, measure and learn and startups aren’t renowned for their documentation. Additionally, even if the EA documents were written, who would read them? And I’m strongly against data wells. So don’t create anything that won’t be read. From that, you can see that implementing Enterprise Architecture in startups is going to be a struggle.
Some founders will be thinking about the here and now but in combination with the near future, just over the horizon. Hence they may think about questions such as “if I build it this way now, am I creating dead-ends later?” The issue of technical-debt is one example of how working for the present is great but it’s a trade-off against what you’ll have to pay later to fix any shortcuts.
The large electronic repositories beloved of Enterprise Architects just aren’t applicable to lean startups. The licenses cost too much, they’re complex requiring specialist skills and knowledge and they don’t provide the results necessary for startups. Instead startup founders are more likely to use more well-known tools to get the job done, such as the Business Model Canvas and Pirate Metrics mentioned above. Adding several of these more accessible tools together and you’ll end up with a simplified Business Architecture. That’s the key here: there are tools and techniques that can get startup founders most of the way there.
The following table provides a summary of the analysis:
|Planning||✖ Too cumbersome||✔ Great at providing info to the planning process. And assessing impact of strategic plans|
|Position||✔ Ideal if you have the ability to execute||✔ Great at assessing impact of achieving that position.|
|Ploy||✔ Ideal if you have the ability to execute otherwise its an idle threat||✔ Great at assessing ability to execute.|
|Pattern||✖ This shouldn’t be the case for a lean startup. If it is, change to Position, Ploy or Perspective||✔ Great at providing info for deciding what strategic direction to take.|
|Perspective||✔ Ideal, adds scope for changing products but may need to be combined with Ploy or Position.||? More related to the cultural components of Business Architecture|
It’s interesting to see that Lean Startup and Enterprise Architecture work well with Strategy Position and Ploy. The also both work well, but EA to a lesser extent with Perspective. Lean Startup doesn’t gel well with Strategy as Planning or Pattern.
What can we apply from Enterprise Architecture in Startups?
There are two things we can do in startups to improve:
- Develop at least 2 canvases. One for where you are now, one for where you want to be in future (when you have a place in the market) and perhaps one for an interim stage (e.g. first launch). They could be 3 separate canvases or could be one canvas with colour or shape-coding to highlight the differences.
- Decide how you develop strategy, e.g. Strategy as Position, Ploy, Perspective or a combination of 2 or 3 of them. Note that Pattern and Planning are omitted from the list of options as not being suitable for startups.
*It’s open in that there’s a foundation created to maintain it, rather than it belonging to a company that has other interests. But it isn’t free, it’s licensable.