Similarly, a small business can be assessed when it seeks credit: AI can look at information such as sales volumes, customer reviews and data gleaned from places including Facebook, LinkedIn, eBay and PayPal in weighing up whether a company is likely to pay its bills.
Getting more from existing customers
AI is now close to making mass-personalisation possible. It can bring together traditional data with detailed information about a customer’s behaviour gathered from sources such as online browsing, social media and wearables. This then allows the right product to be offered at the right time with right message. McKinsey has estimated that such mass- personalisation can lift sales by 10% or more.
With Amazon, for example, we already see machine learning exploited to make suggestions to existing customers: they are presented with a recommended “next product to buy”. These ideas are generated by looking at your demographic profile, what you have bought in the past and what has been bought by customers with a similar profile. At one point, some 30% of Amazon’s sales were prompted by its recommendation engine.
Business-to-business companies can adopt a similar strategy by mining past ordering patters and comparing them with the sales made to similar customers.
And don’t ignore AI’s potential for reducing churn and holding on to customers. Machine learning can look at such metrics as the frequency with which a customer logs on, their response rate to emails and how often they call service desks. Then it can estimate the likelihood that a customer will defect to a rival, and intervene to try to reduce the number of cancellations.
Take the example of Netflix. Customers are likely to lose interest if it takes them more than 60 to 90 seconds to find something they want to watch. By using personalisation and recommendations, the company reckons it saves more than $1 billion (£757 million) in revenues that it would otherwise have lost.
Getting the right price
There is no law of nature that says every customer should pay the same price for a given product or service – nor, indeed, why a single customer should pay the same price each time she or he buys a given thing.
Uber provides the best-known example of variable pricing: when demand is high in a particular area relative to the number of drivers available, the price of a ride will rise.
The same idea can be applied to a myriad of other industries. Mobile phone operators are experimenting with the use of machine learning to predict demand and gauge the price sensitivity of small cohorts of consumers. They can then use this information to decide the demand-price trade-off to maximise revenues and profitability.
And in the business-to-business arena, companies can use data science to pinpoint clusters of customers with similar buying patterns, identify similar deals and generate information about what prices are being paid. Arming salespeople with this information can help them negotiate the best possible prices without losing business.
Increasing sales productivity
Around two-thirds of a salesperson’s time is taken up with routine jobs such as making contact with potential clients, setting up appointments, taking orders and preparing contracts. Companies have begun automating activities such as these, freeing up salespeople to close deals, nurture relationships and manage deals that are out of the ordinary.
A company that can identify the leads that are most likely to lead to a sale will have an advantage over its competitors. Companies are using AI’s predictive capabilities to pinpoint the most promising leads and routing those to sales people who are best suited to close those deals based on their past sales history. Additionally, companies are recording, transcribing and analysing sales calls, demonstrations and meetings using AI. Looking at how the most successful sales representatives led conversations helps companies to coach other staff to up their game and secure more deals.
Making marketing spend more effective
Personalisation is now at the core of marketing. It can yield huge gains in the returns achieved on marketing spend.
But forward-looking marketers can now go further. Analysing customers’ behaviour is only the start. Companies can now target their marketing in a way that is attuned to a customer’s behaviour, preferences, and sentiment, creating emotionally intelligent personalised content using natural language processing (NLP) technology which could be deployed at scale across all marketing channels. Citi, the global bank, has employed Persado, a marketing language cloud company and has seen a 70% increase in its email “open rate” and a 114% increase in the click-through rate.
Abhijit Akerkar (MBA2008) is Head of Applied Sciences, Business Integration, Lloyds Banking Group
What every CEO needs to know about AI. Part two: return on capital