It's No Monkey Business

In the late 19th Century a railway signalman called James Edwin ‘Jumper’ Wide worked for the South African railways as a guard in a town called Uitenhage. Wide had lost his legs in an accident and hence got around on peg legs and a trolley. Around 1880, while visiting the Uitenhage market he saw an incredible sight; a young baboon driving an ox-wagon into the market. Wide convinced the baboon’s owner to sell him the animal. He named his new pet, Jack, and trained him to push his trolley and help with other household chores. Wide was a signalman, so as trains approached the rail switches at the Uitenhage train station they would toot their whistle a specific number of times to alert the signalman which tracks to change. By watching his owner, Jack the baboon picked up the pattern and started tugging the levers himself. Soon, Wide was able to kick back and relax as his primate helper did all the work of switching rails. The locals soon found out what was happening and many would come and watch Jack at work. The railway authorities too found out later and rather than firing Wide, they tested the baboon’s ability and came away impressed. Jack was reportedly given an official employment number and paid 20 cents a day and half a bottle of beer weekly till he passed away in 1890. Jack, the baboon, worked the rails for nine years without ever making a mistake and remained a fascination for curious onlookers.

While Jumper Wide employed a baboon, today, we employ technology to help us and to automate business operations. Not unlike the people of Uitenhage, fascinated by the sight of a baboon moving levers, many of us are fascinated with mentions of ‘artificial intelligence’ or ‘machine learning’. It’s as if such tools are imbued with some magical powers. Deploying AI and ML become a showcase project for many, even if it means putting the ‘cart before the horse’. Solutions using AI or ML seem to inspire a sense of confidence that you have a winner on your hands that needs no further investigation.

It's not go decry technologies like AI or ML, today they power the many things we use every day, from voice assistants to face ID, chatbots, rideshare apps and much more. While no Luddite is a marketer, it’s important to take a clear eyed and nuanced view of their strengths and weaknesses and know what your business actually needs. Here are a few things I’ve learnt.

*Don’t put the cart before the horse.

Have a clear use case and make sure that the technology solution picked is appropriate. Sometimes tools like AI, ML or blockchain may be part of pitches in desperate search of a problem to solve. The objective is not to deploy AI or ML but get the job done in the simplest, most effective and most efficient manner. One of my favourite examples is Indian Oil’s use of the humble telephone for ordering an LPG cylinder refill. To order a refill of your LPG cooking gas cylinder, all you need to do is give a ‘missed call’ from your registered mobile number, the LPG gas cylinder is automatically booked and the fulfilment process follows. The simplicity of using a ‘missed call’ caters to the wide spectrum of Indane consumers who may not be at ease using online booking tools.

*Ask the KKH question

The technology alphabet soup can be overwhelming, it’s good to define what you want to achieve – Karna Kya Hain – KKH? Answers to KKH may be different for your industry and organisation – it’s what you define as your goal. And It’s not static, as your business challenges evolve over time, so should your answer to the KKH question. Use this to guide your future course and not some fad or the desire to ‘keep up with the Joneses’.

*Be clear of the business benefit

Determine the business benefit the investment could deliver. For instance, if you are planning to invest in a chatbot to handle routine customer queries, start by measuring the volume of routine queries, understand your current spending on agents and determine the net savings bot deployment could deliver through replacing agents engaged in this routine task. While measuring ROI is important, it’s also useful to remember that not every experience and value delivered by your organisation is measurable.

*Understand the limits

We’ve all experienced embarrassing auto-corrects while texting someone and we may have read stories where chatbot auto responses simply didn’t understand the nuance in a user’s question or miss sarcasm in user comments. While the technology is constantly improving, it’s important to understand the limits of what you are planning to deploy – know what the system cannot do – be aware of the end-user experience. If cost savings to your business means subjecting a customer to inconvenient loops of trying to make an automated system understand a query, it’s simply not worth the effort. Your deployment can wait, till you can get the experience right.

*Get the foundations right

It’s possible to deploy ML, for instance, for a new use case by using existing open data sets. This can be used to train and test initial models and as the deployment matures, you can capture more data and improve the models using actual data. The value in such tools is generated with lots of data, without data, the algorithm will not be able to be trained. Hence, it’s important to check if your underlying systems and pipes delivering data are in order. RIRO, rubbish in rubbish out, still exists.

*It's not a sprint

Advanced tools once deployed need investments in monitoring, maintenance, investments in data scientists and other support. Your use case needs to justify the investments it calls for.

*When in doubt remember what Warren Buffet said

Some tools and technologies tech service providers pitch may seem very complex. It is vital for you to understand the solution at its conceptual level and what your consumers will experience. But if answers to your questions of ‘what the solution does’ seem hidden behind a thick haze of acronyms, highly technical responses and obfuscation, it may be useful to follow Warren Buffet’s advice – "Never invest in a business you cannot understand." (caveat, Buffet said it in a different context) 

*Be an open-minded sceptic
And finally, be an open-minded sceptic. As Carl Sagan describes it, aim for that exquisite balance between two conflicting needs; the most sceptical scrutiny of all hypotheses served and at the same time a great openness to new ideas.