I write this post on the eve of my 34th birthday. It seems an apt time to reflect on my journey towards Convert Dial for two reasons. Firstly, it's my birthday and isn't that as good a time as any to reflect? And two, I have just deployed the final edit on the AI engine for Convert Dial. It's done, it works and now starts the journey to take it beyond my study and out to the wide world looking for that holy grail of startup dreams: product-market fit.
The story of Convert Dial goes back about 16 years to when I was a wide-eyed engineering student who had never even heard of 'artificial intelligence'.
Learning about artificial intelligence
My bachelor's degree was in Operations Research from the University of Auckland and artificial intelligence was one of my favourite subjects. My most memorable projects were ones that used artificial intelligence to solve problems such as
how to stack containers on a ship such that the centre of gravity was maintained?
how to schedule paint jobs such that the least amount of time was spend cleaning out the machine between jobs?
Where to locate ambulances so that the time to travel to the next emergency is minimised?
That is where I started to develop interests in artificial intelligence and data mining. I realised that I really loved how technology could be used to solve problems! In the final year of my undergraduate degree, I did an internship at a company that used artificial intelligence to predict water leaks and I decided that I wanted to pursue a career in data analytics and artificial intelligence.
My artificial intelligence journey continued when I joined an international consulting firm where I worked on projects involving artificial intelligence in industries such as banking, telecommunications, retail and insurance... but more about that another time.
Learning to use artificial intelligence
When you are a beginner, you realise that artificial intelligence is just a method of trying to find patterns in data. It can be used for e-commerce or any industry where data is abundant. Fast forward to now and I have been working on artificial intelligence applications for retail and e-commerce businesses for ten years!
Some of the artificial intelligence applications I have developed are:
Customer segmentation using the shopping behaviour of customers to predict new product demand
Demand forecasting using artificial intelligence algorithms such as decision trees or random forests to predict future sales and stock levels
Range optimisation to decide which products should be in store at any time
Why Convert Dial?
About a year ago I realised I had been working for 10+ years and had amassed a wide berth of diverse experience. It was also around when the COVID-19 pandemic hit and it made me reassess my future. Paid employment no longer felt stable, although my job was mostly unaffected, something I am grateful for. But more importantly, COVID-19 was also a sign of change across the world in every industry. I decided that I would use this opportunity to launch a new path in my career.
I had always dreamt of creating something myself, something new and innovative so I starting thinking, what can I do with my skills that hasn't been done before? Naturally, data science, analytics, machine learning and AI were the things I knew best.
First I thought of starting a consulting business, I chose the name Arche Analytics after much deliberation because 'Arche' means 'from the beginning' in Greek! But I soon realised that consulting is a tough market to crack and it didn't feel like the right option. By the way, I am a big believer in my gut instinct, it never lets me down! My gut said that consulting wasn't the path for me.
I then turned to the idea of a SaaS platform. Something that seamlessly integrates data, does ML and AI and creates insights that can grow a business. I knew that many tools in the market are tens of thousands of dollars a year for custom solutions. I was confident that I could build something simpler, faster, cheaper and automated. I knew Python and R programming and I knew my way around machine learning algorithms and their applications.
Slowly, a viable idea started taking shape
It was around this time that I came across the Western Sydney University accelerator program. I applied and was accepted based on my idea to create an application that used first-party data to optimise conversions in e-commerce. To date I don't know what they saw in me! I had one slide and I suppose some credibility but they saw promise and they offered me a spot. I credit this program for what comes next.
First, a slight detour into the world of AI
What is artificial intelligence?
Artificial intelligence is about teaching computers to find patterns in data. Machine Learning is a field of artificial intelligence and ML algorithms are used to predict the future behaviour of customers.
To do this, we must first find patterns in historical data. Once artificial intelligence has found these patterns in historical data, it can then make predictions about what will happen in the future based on these patterns.
This means that artificial intelligence can be used to predict customer behaviour, which is what I have done with Convert Dial
How is artificial intelligence used in e-commerce and retail
The most common artificial intelligence application for e-commerce and retail is sales forecasting. Sales forecasting is an artificial intelligence technique that predicts future stock levels and sales based on past buying patterns. Sales forecasting can be combined with demand forecasting (also known as range optimisation) to predict whether a product should be ranged or not.
Customer segmentation is also an artificial intelligence application. This application tries to predict the future shopping behaviour of customers. This can then be used to generate targeted marketing messages for individual customers
I have learnt through my career that artificial intelligence is very useful for retail businesses where there is an abundance of data that can be used to predict customer behaviour. For my next artificial intelligence project I decided to focus on e-commerce businesses for three main reasons :
e-commerce businesses have a lot of data that is underutilised, artificial intelligence can be used to extract value from this data
machine-learning can be used to solve many problems that e-commerce businesses face such as - what customers are interested in purchasing, which customers are the most valuable, what products to stock and recommend, when to reorder stock, when to discount sales & much more
While artificial intelligence is not a new technology, it is not widely implemented in e-commerce
Whilst artificial intelligence will not solve all of the problems in retail, it can help e-commerce retailers make more informed decisions at scale and reduce costs. I wanted artificial intelligence to be at the core of e-commerce businesses as a tool for predicting customer behaviour - and thus was born the idea for Convert Dial.
The Convert Dial journey begins
My artificial intelligence project grew from here and I decided that I wanted AI to be at the core of my solution for e-commerce businesses. As you might have noticed from looking at the homepage, artificial intelligence is the base of Convert Dial.
Now that I had decided that I want to help e-commerce businesses to predict customer behaviour, I quickly realized that there are three key challenges to using artificial intelligence in eCommerce that I would have to overcome
Firstly, artificial intelligence can only create a predictive model with the available data
To predict customer behaviour in e-commerce businesses we need to know what customers are likely to do next (e.g. purchase) - and these events will never be 100% accurate because only humans will buy something for the first time. However, artificial intelligence can predict customer behaviour in a given time frame.
Secondly, artificial intelligence is only used as a tool for predictive models, not to act upon the predictions of these models.
An artificial intelligence model might say that a customer will spend $80 in one week but will it automatically convert them to a sale? No, the business has to take action on the predictions to ultimately make the sale happen.
I realised that not only would I need to ensure that businesses have enough data to build accurate models to predict customers' behaviour but I would also have to help e-commerce businesses to act upon the predictions.
In the next post, I will share how I tackled these challenges and the evolution of Convert Dial from an idea on a slide to a SaaS platform