How primary school science helped me decide to join nPlan

This is my story of joining nPlan — a seed stage, London-based start up applying machine learning to construction programmes. We are helping contractors, their clients and other stakeholders better understand risk and uncertainty.

If I leave half an orange on the kitchen bench it will develop mould faster than the half I leave in the fridge.

This was the hypothesis of a science experiment I did in primary school. What does this have to do with choosing to work at nPlan? On the surface, nothing!

In 2017, I began an MBA at London Business School. Returning to school has reintroduced me to the principles of experimentation I learned in primary school science. Back then, we would develop a hypothesis then follow a method to test that hypothesis. I didn’t see much space for this method outside the science lab.

In my job as a project manager at Lendlease the stakes were high, so we tended to avoid experimentation. On reflection, my part of the organisation was necessarily organised to execute, not organised to learn. So, I have been surprised to learn how widely used this hypothesis led thinking approach is in business contexts. Consultants warn against “boiling the ocean”, startups subscribe to the “lean startup” methodology and product teams run “design sprints”.

Job searches are always hard, but especially at a career inflection point like your first job, returning from parental leave or after an MBA. Should I start fresh in a new industry, stick with what I know or take an adjacent step? Towards the end of my MBA, and after rediscovering hypothesis led thinking, I decided this would be a good way to choose amongst the options. This is how the mouldy orange relates to my decision to work for nPlan.

My hypothesis about nPlan:

nPlan is a high potential startup, operating in an industry that is growing and ready for technological disruption. The team is strong and a culture is forming that will help me and the company succeed. The product is technically superior to the competitors and the timing is right.

Once I had my hypothesis, I set about testing it with the resources at my disposal. The quality and availability of data varied across the key dimensions. In some cases, I had primary data and in others I relied on secondary resources or other heuristics like social proof. The nPlan “experiment” is still to be concluded, but testing the key elements gives me confidence in my decision.

High potential startup

It is relatively easy to test this element. McKinsey Global Institute has written a series of papers on the construction industry that have been widely shared online. The key insight is that to deliver the growing global pipeline of construction projects, innovation is desperately needed to ensure projects are delivered on time and to budget. There is evidence that companies throughout the value chain are adopting innovative technologies and venture capital finance has followed in turn.

Team

The team element of my hypothesis is a bit more difficult to test. I relied on social proof to test the aptitude and potential of the founding team. This was a quick tick, as Dev and Alan are graduates of Entrepreneur First and have been backed by top Venture Capital and Angel investors.

Culture and fit is even more difficult to gauge but I was impressed by the frankness of our discussions and was welcomed by everyone on the team during our early meetings. When we shared a beer in my interview, I was pretty sure we would get on!

Product

This element was most difficult for me to test given my non-technical background. From my own research, I established that for Artificial Intelligence companies need both a predictive model and a good data set to succeed. One of nPlan’s earliest backers was Demis Hassabis, founder of Deep Mind and one of the world leading Artificial Intelligence entrepreneurs. His authority in this space gave me confidence in the quality and potential of the model.

Ash Fontana of Zetta Venture Partners, an AI investor, describes an elegant method for assessing ML data sets in this podcast which influenced my thinking on the nPlan data set. His team uses criteria like difficulty to acquire and model, value of the prediction, perishability, network effects and dimensionality. My research about nPlan and subsequent discussions with the team gave me confidence about the strength of the data set against these dimensions.

Timing

Testing my readiness to join an early stage startup required introspection. The financial trade-off when joining an early stage startup like nPlan is between income today and the opportunity for wealth creation in the future. More importantly, it is also a chance to shape the future for your team, company and industry. After much deliberation with family and friends, I decided that this decision works for me at this point in my life.

The timing was also right for nPlan. There was a small gap in the team for a commercial skill set with industry experience and network that I have joined to fill.

Conclusion

Unsurprisingly, my school science experiment hypothesis proved to be correct — the orange left on the kitchen bench went mouldy and the one in the fridge did not. Leading with a hypothesis in that situation seem like overkill but the practice of applying hypothesis led thinking is an important skill. So far, the process has provided a rubric for career decision making but this is just the start — I look forward to applying it to the challenges and opportunities ahead with nPlan.

We’re always looking for exceptional talent so if nPlan sounds like your thing check out our current openings or reach out to me on LinkedIn or email [email protected]