“Here we must run as fast as we can, just to stay in place. And if you wish to go anywhere you must run twice as fast as that.” – Lewis Carrol, Alice in Wonderland.
Providing any kind of customer service, you regularly have to deal with the challenge of how to become even better to keep up with the times.
Today, the winners are those who respond to customers first, who show personal interest in every client, and who offer the best price compared to others. Personalization is what the modern customer desires from any product or service but, to provide this, companies have to learn the demand of their audience as precisely as possible. While many administrative processes have already been automated with the help of sophisticated systems like CRM, CMS, ERP, and other tools, these still do not provide any analytics or answers on how to personalize the approach for each individual client.
The main hurdles that companies have to overcome are:
- Inability to process collected data
- Lack of personnel dedicated to personalization
- Legacy technologies
In a recent survey by BCG, almost 50 percent of companies confess: “We have the data, but integrating and using it — that’s the hard part.”
That’s why leading companies and p2p marketplaces have already implemented artificial intelligence (AI) into their systems to automatically analyze multiple factors coming from collected data. AI-based solutions have already found their way in such spheres as eCommerce, financial services, healthcare, tourism, transportation, and others.Machines learn from the context they have and become more and more precise in suggested decisions. As we write this, businesses can use the power of artificial intelligence and machine learning to enhance their customer service by:
- Predicting user preferences
- Optimizing prices
- Detecting fraud
- Personalizing communication
Let’s explore how companies use AI to retain their customers.
Finding the key to each customer’s heart is possible only when you start treating him or her as an individual by providing a very special custom experience. World-famous companies such as Airbnb, Uber, Pinterest, Starbucks, and others actively use AI technologies to provide personalized suggestions.
Case Study: Airbnb’s Personalized Search
Airbnb, a leader among p2p residential real estate marketplaces, with more than 200 million guests in total and 3 million unique listings, focuses their efforts on the enhancement of the tailored search experience.
It is not an easy task to compare a large amount of listings, and to select only a dozen of those to match exact user preferences. That is why, starting in 2014, the engineering team built an AI model that analyzed more than a hundred signals at once, to personalize the search in real time.
Previously, search results were retrieved according to predetermined software rules, taking into account only a few factors such as price and number of bedrooms.
“We match guests not just with hosts and listings that align with their preferences, but also with neighborhoods and experiences that meet their needs and interests, which makes for a better consumer experience, through and through,” explained Surabhi Gupta, Director of Engineering at Airbnb.
Their ML algorithm, named Embedding Listings, looks at the collected data from the user’s previous search history and wishlist pins to combine it with desired listing features that can include location, availability, amenities, and other options. The system ranks the results in a Real-Time Personalization Search and a Similar Listings Carousel to show the most suitable variants.
The success of such approach was proved during the first tests of the AI-fueled search mechanism. CTR and booking percentages increased by 21 percent and 4.9 percent, respectively.
How do you stay profitable without raising prices? This is the question that touches the service hosts of the p2p marketplace, ride-hail drivers, and online retailers. Analysis of multiple signals to suggest the right price here and now – this is the task most often dedicated to AI technologies.
Case study: Airbnb’s Smart Pricing
As you know, platforms like Airbnb deal with two groups of customers – hosts and tenants. Thus, putting “cultural ideology” and affordable pricing at the head of the table, Airbnb managed to win the love of travelers looking for lodging; however, Airbnb has to struggle with the no less important tasks of retaining property hosts and attracting new ones.
Airbnb’s analytics always explore their customers progress and they quickly found out that the biggest challenge for hosts was price setting. When monitoring user activities, analysts noticed that many new hosts left the website on the step of the price estimating. Try to do it yourself and you’ll understand that this is a really difficult task. You need to set a price that is competitive among similar listings, but that also provides you with enough compensation, so you don’t have to worry about negative earnings.
This is why Airbnb’s engineers implemented an ML model that provides daily suggestions on price optimization, taking into account multiple indicators like location, seasonality, amenities, prices of similar lodgings, current listing prices, and availability. Hosts are not obliged to follow the recommendations, but the study says that “when a host selects a price that’s within 5 percent of their tip, they’re nearly 4 times more likely to get booked, when compared to hosts whose prices are more than 5 percent away from their tip.”
It doesn’t take too much surfing on the Internet to see that there is a lot of negative feedback from Airbnb hosts about this service. The most evident reason why people say that Smart Pricing “sucks” is that the algorithm is built with an aim to increase bookings, which means that lower pricing is almost always proposed. If you are normally used to having larger revenues come from your lodging, it may seem silly to lower the price, even if it means having more bookings. Of course, we can be assured that Airbnb, given its place in today’s market, is aware of the problem and will definitely be coming up with ideas in the near future, in order to fix it.
One of the most exciting benefits of AI technologies is how they work with cybersecurity. Nobody knows when it can happen, but it is extremely important to react to unauthorized actions in real-time and try to prevent any repetition.
Case Study: PayPal
PayPal is one of the biggest electronic payment services, having processed 7.6 billion transactions with 227 million registered customers in 2017 – with revenues of more than $13M. No wonder this is a desirable platform for hackers.
PayPal engineers merged AI with their security system to analyze user behavior across transactions and identify unusual activity. If a behavior is reported as fraud, the system flags it and files as a “feature.” If the pattern is repeated, the fraudulent transaction will be immediately identified and prevented.
“We now process thousands of ‘features’ in our system, compared to hundreds when the system was first put to use in 2013,” says Hui Wang, PayPal’s Senior Director of Global Risk Sciences.
Also, AI solutions allow differentiating suspected frauds from actual breaches of security and, thus, decreasing false alarms.
Most of the services we use today can be accessed online – from booking a taxi to buying a new dress. People are sociable beings, however, and they often need to ask advice or follow guidelines to complete a task. It can be quite redundant to read long-winded instructions, desperately search for trustful reviews, or hang on while an operator on the phone finishes up with another customer before getting to you. Fortunately, a way out is possible with fast learning programming algorithms that allow for the creation of conversational UI that will patiently and politely respond to all client needs.
Case Study: The North Face
Buying clothes for extreme sports and hiking may be a real challenge. Will this jacket be okay for snowboarding at a resort you’re heading to next season? Who do you ask for professional advice if you are buying online?
The North Face seems to understand their customers’ pains and has enabled a conversational UI driven by IBM’s Watson for a smooth personalization experience. The bot asks you when and where are you going and, retrieving weather conditions and peculiarities of the landscape, provides you with the best matching items.
Case Study: Sephora Virtual Artist
Selecting a makeup product from a large variety of brands and color pallets is always an experiment which can be very costly. Luckily, you can now try the new online “Sephora Virtual Artist” to pick out a new style and suitable products in an interactive environment. Sephora, one of the most advanced – and growing – retailers in the beauty sphere, strives to understand their audience better with this solution.
“They are creating dialogues with customers, not monologues. And those dialogues – whether it be in store, on the app, or online – are what helps Sephora understand their customers better, and then deliver the kinds of experiences that not only meet but exceed customer expectations.” – said Brendan Witcher, Forrester analyst.
AI algorithms underlying the application scan your photo and determines where your eyes, lips, and cheeks are. A number of settings allow you to play with colors and styles, selecting among 20,000 products sold at Sephora.
As a result, over nine million customers have already tried the “virtual assistant” feature since its launch. The organic revenue in the selective retail of LVMH, which Sephora is part of, increased by 9 percent in the first quarter of 2018 thanks to AI implementation.
As you can see, from unification we inevitably run to personalization in all spheres of our life. For years, we collected data that is now stored permanently in Excel files, CRM and databases tables – even though nobody realized it when it was happening.
Modern customers literally tell us “You know everything about me, why don’t you give me what I want?” This is where Artificial Intelligence comes to help. There is no ready-to-use solution in AI. Any system should be adjusted individually to your specifics. Leading companies have already dived in this pool of limitless opportunities to personalize customer experience in searches, pricing, security, and conversation.
Studying the experience of the pioneers who dared to implement AI and ML in their business, you can learn how to improve your services.