Diagram-driven and data-driven: The art and science of architecture

By combining diagram and data-driven architecture, architects can move faster, offer more cost-efficient programs and most importantly track and measure success, writes Dr Tim O’Neill, founder of Avolution

In the movie Moneyball, Brad Pitt’s character breaks with baseball convention by relying on data science instead of traditional scouting to choose a team he hopes will stop the Oakland A’s losing streak. His analysts use probability to see past biases and identify both champions and affordable players that everyone else has overlooked.

As enterprise architects, we’ve always been comfortable with data. But the addition to our toolkits of swifter integrations, APIs and the availability of data from all corners of the business has meant huge leaps in the volume of data we can use to understand operational and technical infrastructure and guide decisions.

The way we store data has also improved our ability to access and analyse it. We’ve written before about the benefits of graph databases in managing large data volumes and organisations are also benefiting from modern analytic techniques such as no-code algorithms and interactive data visualisations.

Enterprise architecture is quickly becoming data-driven – and needs to be. Data is used more rigorously and regularly, but architecture needs data to stay relevant. What of diagrams, then, the “art” of enterprise architecture? Are they now old-school?

Visual modelling

Most business strategists – not just enterprise architects – use diagrams or visual models to do design. We all do this instinctively. The first thing people reach for when planning, creating a strategy, or conceptualizing a new line of business is not a spreadsheet: It’s a whiteboard and a marker. Later they might “draw” plans in PowerPoint, Visio or another drawing tool with a native canvas-and-stencil functionality.

However, diagrams which aren’t anchored in strong data can leave us with the uneasy feeling that organisations are relying too much on gut decisions or pretty pictures when making important business decisions. Ad-hoc “colouring” and “tagging” doesn’t provide the quantitative rigour needed to steer today’s super-tanker enterprises.

If a diagram isn’t interactive, dynamic and most importantly connected to fresh, real-time data, there is a significant risk that it becomes a dusty artefact, quickly losing relevance. It will be out of date before that cloud migration is complete, before the M&A strategy is executed, before the shiny new applications come online.

Strategy & Execution

The true value of enterprise architecture always lies in capturing strategy and guiding the execution of that strategy against defined business outcomes, not in the style. Diagrams are useful at various stages of the architecture process but alone they are little more than ‘wallpaper’ for the conference room.

The challenge for architects in organisations is to take the pulse of different facets of each line of business, and make decisions to rebuild parts of it whilst still ensuring it functions — a challenge similar to fixing the suspension of the car whilst it’s still travelling beneath you.

To do this really well, and safely, you’ll want plenty of operational data, and you’ll want a powerful analytics layer on top of that: Properly set up analytics allows you to move faster.

Anchoring your architecture and diagrams with data means your organisation will be more effective at understanding the strengths and weaknesses of their existing infrastructure and technology against familiar patterns, calculating what impact external forces and internal changes will have, running useful scenario and gap analysis, and keeping a multitude of projects happily on track.

Diagram-driven architecture works best when diagrams are: Customisable to your business, automatically updated; connected to data; and designed to support communication or decision making.

Data-driven architecture works best when it is: Integrated with enterprise data-sources (e.g. using an API); focused by key business metrics or outcomes; provides both diagnostic and predictive analytics; and “explorable” through interactive data visualisations.

Modelling success

Nowadays, architects and digital business managers can benefit from a full toolkit of styles that go hand-in-hand with data science techniques, including familiar reference models and hand-crafted diagrams sketched out using canvas-and-stencil, that can be animated by data from financial and technical algorithms.

By maintaining a future and business-driven mindset, architects can move faster, offer more cost-efficient programs and most importantly track and measure success, putting their business in the position to excel and win.