Artificial Intelligence on eCommerce is a matter of the present, not the future; and it is necessary that the people who work in marketing understand how and when to approach the issue in their companies. But since I am digital strategist and not a technology person, my perspective is going to be practical and empirical.
Any expert in technology or data science who wants to contribute is more than welcome.
Artificial Intelligence versus Machine Learning versus Deep Learning
First I want to differentiate three terms that I see that many use them interchangeably, but even though they are closely related, are not the same thing. The concepts are:
Artificial Intelligence:
This is the ability of some machines to perform typical tasks of the human mind, such as planning, learning, recognizing written and spoken words, and solving problems for which they had not previously been programmed for.
Machine Learning:
It’s the way to train artificial intelligence to learn “how” to process data, so that – little by little – the system itself will gradually improve on how to present more accurate results.
Deep Learning:
It’s one of the multiple alternatives in Machine Learning that was inspired by the way in which the human brain works and how neurons interconnect to identify patterns, objects, sounds, etc.
Artificial Intelligence on eCommerce, the gifted Salesperson.
A.I. has the potential to analyze the millions of interactions that can occur between an ecommerce business and all its visitors, and from that analysis predict what visitors want to see; then optimize itself to present a personalized offer that maximizes the chances of converting.
Even though A.I. has applications in all areas of a company but I’m going to focus on three very specific ones that impact the effectiveness of Marketing efforts: (1) Recommendation Engines, (2) Chatbots and (3) Voice Recognition.
Recommendation Engines:
- Imagine that your company has the best Salesperson in the world, one that
- recognizes all the people who have visited your store
- remembers everything they bought and what they almost bought
- knows the price each visitor saw for the products they didn’t buy
- and is able to predict the price to which each visitor would be more likely to buy.
Based on this information, that gifted Salesperson will put in front of each visitor an offer made exclusively for him.
To give you an idea of the power of this predictive intelligence, 35% of Amazon’s sales come from product recommendations made by the Artificial Intelligence on eCommerce algorithm.
On average, of every 20 products that Amazon recommends, we buy 1, that is, 5% conversion. “Retargeting” further enhances the effectiveness of this gifted salesperson, because it can pursue prospects beyond the website itself.
Chatbots – Conversational Commerce
Chatbots are robots with Artificial Intelligence trained to interact in real time with a costumer. Through chatbots, consumers can interact in real time with the gifted salesperson described above on platforms such as Whastapp and Facebook Messenger, among others.
The mission of these chatbots is not to bombard the consumer with multiple offers, but to support and advise them during the purchase process by presenting optimal solutions to their need.
They -the chatbots- learn about the needs and concerns through the information they get from the conversation, and possibly further enriched with historical data from past interactions.
Companies such as H & M, Amtrak and 1800-flowers have reported a 150% increase in their engagement rates and 25% increase on the average purchase since the implementation of chatbots.
Voice Recognition – Voice Commerce
The interaction through voice commands is simply giving voice and ears to the gifted salesperson, so that he is able to listen, ask, respond and recommend, in the natural way in which humans communicate: talking.
There are two factors that had been preventing the voice commerce from spreading wider and faster, first, the error rate in the interpretation of voice commands which has improved to less than 4% in 2018 (from 10% in 2016).
The other factor, the collective concern that people can be spied on by voice enable devices.
Today voice searches represent around 20% of searches made from mobile devices, but it is projected that by 2020 50% of all online search and shopping processes will be done through voice.
Artificial Intelligence without a data strategy becomes artificial stupidity.
Now, all that potential of the Artificial Intelligence on eCommerce to improve the performance of a business, is just that: a potential. The way to begin to materialize this potential is by designing and orchestrating a data and technology strategy.
Learn how I can help you develop your Digital Marketing Plan.
The 4 questions that you should ask yourself to begin with are:
- Do I currently have the data I need to train the predictive model? If you do not have the data, you can have a couple of years of work ahead to collect it.
- If you have the data, do you trust in the source from where those data come, in that they have been captured in the correct context and stored as they should be?
- What are the risks of doing it wrong? How much it can cost the business, not only in technology and wasted efforts, but also in the loss of costumers’ confidence.
- How much human intervention will be necessary to support the proper performance of the AI? If the involvement is too high, the investment may not make sense.
If you are a marketer, then your job is to monitor the entire Digital Ecosystem to identify which tasks and initiatives of your marketing program can be made more efficiently by leveraging Artificial Intelligence on eCommerce.
Your job is not necessarily understanding how to develop de A.I. solution but you need to keep your eyes open for opportunities to drive Digital Disruption.
Any contribution to enrich this approach is appreciated and welcomed.
