Approaching Proof of Concept like Sun Tzu, A Military Strategist and Philosopher

When I first read The Art of War, I struggled to understand why Sun Tzu was invaluable to the art of business. Two decades later, with a few scars in battle, I now recognise the pearls of wisdom within this classic and believe it should be mandatory business reading especially to technologists with entrepreneurial dreams. There are many lessons in the text, which is a beautiful read for those of us who have an open mind, a desire to improve and become a better version of ourselves.

One lesson I use probably every day is from the formation chapter (Thomas Clearly translation):

“A victorious army first wins and then seeks battle; a defeated army first battles and then seeks victory.”

I learnt this lesson the hard way. Many moons ago, as a co-founder, to win over investors we desperately needed more mobile operator clients despite having signed one of the largest mobile carriers in Asia. My Norwegian colleague found a mobile operator in Russia who was interested in our software. One month later I was in Moscow presenting. It was a strange meeting, falling into my top 5 most peculiar meetings category. The VP was pointing a toy gun at me throughout the meeting and said he was interested in evaluating our software. He had the authority and budget to move forward.

Six months later my CTO and I were very stressed. We had many meetings and flew out considerable technical resource onsite. The integration between both platforms seemed like a nightmare. It was a precarious scenario, as it was a self-funded start-up. We realised eventually, we were fighting a losing battle on the technical front and decided to pull out.

We should have spent more time upfront technically qualifying before expending resource, but the excitement of securing another client blinded us. The qualification process is fundamental for the buyer and seller of Enterprise software and for me it is a continuous process throughout the engagement lifecycle. The burden of technical qualification typically weighs heavier on the buyer’s side in an enterprise, primarily due to the fact, the seller just wants your Tubmans.

One way of qualifying a technology is the Proof of Concept (PoC). Evaluating whether the technology is fit for purpose, this approach is a useful engagement for both parties once the business has taken an interest in the conception of the idea. I recommend a fast track PoC, one to three months. If the vendor wants to prolong the process that should raise your vapourware alarm bells unless it is a pilot being integrated with internal or external systems. Another tip - jointly define the functional scope with the vendor; otherwise the seller will try to align the PoC scope with what is easy for them and when it comes to production time, you, the buyer will enter a whole world of pain.

Before embarking on the journey of the PoC, what can you do to qualify further it will not be a useless exercise? Let’s use the technology domain Cognitive computing to explore this further (skip to the last paragraph if this is not your domain of technology). I believe different approaches are required for each technology area, so this advice is only relevant for areas of Cognitive systems:

1.     Data - Check whether you have the data and access rights to the data. Some organisations have intentional Chinese walls for governance reasons. You may find half way through the PoC you are not allowed to access the data.

2.     Build or Buy? – Define your requirements and conduct a gap analysis on the functional scope helping to determine what can truly be done in house, then develop a strategic roadmap of how the system may evolve. If you find over 60% can be done in-house at a cheaper cost than procuring a system and there will be no strategic change over the next few years, it makes absolute sense to build in-house. Bear in mind building a Cognitive system can be rocket science and hiring the right folks with those skills is not cheap.

3.     Digitalisation of the Data – Your department has access rights to the information but can you do anything with the data? The data may be voice calls or stacks of PDFs - if both data sets contains either strong Glaswegian accents or handwritten doctor’s notes, the accuracy of conversion could be limited. Try Google’s Tesseract to test the quality of the conversion of PDFs. Also, consider you may just need part of the document or individual paragraphs as inputs - this will help the quality of the conversion process.

4.     Secure Business Buy-in – Going renegade just won’t work for a Cognitive system as they require business buy-in because it is a change programme and to make your vision into reality you require business buy-in and budget.

5.     Right Software for the Business Challenge – Speak to a few different vendors and have an open dialogue with them. I’m confident their software can fix all the world’s problems, however, present the most complicated part of the business process and request their solution approach and indications of TCO (Total Cost of Ownership) of the overall system.  

6.     Resource to Manage the System – A Cognitive system is no different to an Enterprise system. Somebody from the IT team will be required to manage the system but also the business rules need to be managed, and there is always discussion about where this piece of work should reside.  

7.     Find and define the Use case – Probably the most important! Capture the use case in as much detail as possible. Engage the business to validate your thinking. Capture it as an “As is” and “To be” use case to illustrate how the use case would impact the business.    

This list of points is not prescriptive nor indeed exhaustive, but it should furnish you with a sense of how to qualify from a technology perspective. Also, there are contractual tricks in your armoury to push the risk on the seller’s side, but that is another blog later. To finish, I leave you with another quote from Sun Tzu:

Therefore the victories of good warriors are not noted for cleverness or bravery. Therefore, their victories in battle are not flukes. Their victories are not flukes because they position themselves where they will surely win, prevailing over those who have already lost.

Quotes from The Art of War Sun Tzu, 1998 Translated by Thomas Cleary, Shambhala Dragon Editions

Picture: Google Image