Why designing a simple solution is often more valuable than engineering a complex solution

Analytics Engines Lead Data Scientist Liam Brannigan discusses why a simple solution is often more valuable than a complex solution, showcasing how this approach has helped deliver value for our clients.

When clients approach us, their central objective is to drive value from their data. Organisations can drive value in a variety of ways, a number of which we explored in our recent Value of Data eBook.

Simplicity or Complexity?

When beginning a data science project, the question that we ask ourselves is “What is the simplest way that we can solve this problem?”. Whether it’s developing a Machine Learning model to monitor Food Safety, using Natural Language Processing to identify potentially fraudulent behaviour or using Knowledge Graphs to show how various entities within a variety of disparate datasets are connected, our ethos remains the same.

This approach has benefited our clients in a variety of ways.

• Time to Value
  The work needed to develop a more complex solution often far exceeds the work required to develop a simple solution. This additional work can often erode the value created by developing a solution in the first place.When developing our Farsight Horizon Scanner solution, we encountered this very challenge. By choosing to develop the solution using less complex Machine Learning techniques, we were able to drive meaningful value within the project much faster.
• Evidence-based Development
  As illustrated in a number of our blogposts, there is a growing demand among businesses for data analytics services. Despite this growing demand, organisations are often hesitant to commit to largescale data analytics projects.A simple solution allows organisations to adopt data analytics on an iterative basis and to evolve, adapt, and enhance the solution as their needs and requirements grow.For example, our solution for Coriolis Technologies has followed this principle. The project began with a modest objective in mind and has incrementally developed as the needs of the client evolved.Also known as the Fail Fast approach, finding a simple solution also allows organisations to experiment and refine their project requirements, ensuring that the end solution more closely aligns with their core business objectives.
• Adaptable and cost-effective
  By their very nature, a simple solution is easier to implement than a complex solution. Moreover, they are generally faster to calculate and as a result, have much lower ongoing computational costs. Simple solutions are also more easily maintained and adapted.For example, solutions may often encounter scenarios that they were not initially programmed to deal with. With a simple solution, we can generally identify what has changed and develop a solution more quickly. With a complex solution, it can be more difficult to identify how the model should be changed in order to adapt to this new challenge.

Conclusion

A complex approach does not guarantee the best outcomes. Oftentimes, such an approach can erode value, create bottlenecks for broader understanding, and limit the long-term capability of a solution.

Adopting a simple, iterative approach allows organisations to realise value more quickly, react to challenges more effectively and to adapt the solution more easily as their needs evolve.

As famed product designer Dieter Rams says, “Good design is as little as possible.”

Find out more

To learn more about how a data analytics solution can support your organisation, download our eBook or contact us using the form below.