How AI puts an end to self-hosted webshops

In online retail, the forefront of development in artificial intelligence is held by major players such as Amazon, Alibaba and eBay. They have money, dedicated teams and data available allowing them to take big steps - and so they do. What does this mean to the position of medium-sized and small, often still independent, webshops?

Wat betekent de opkomst van AI voor de middelgrote, zelfstandige webwinkel?

Already now, the platforms and market places of these online retail giants are using a lot more AI than you would expect. For example categorizing products automatically based on textual data; predicting return of goods; checking reviews on authenticity; fraud detection with payments; predictive sales and personalized recommendations where the system itself can make the right connections

Thousands to hundreds of millions of Euros are invested here. Of course, smaller players do not have these budgets. But that does not mean the gap between dominant international market places and webshops, are unbridgeable.


My expectation is that the more technical e-commerce platform providers such as CommerceTools, Salesforce and SAP-Hybris (say all parties which provide a commerce platform, but do not run a marketplace or shop themselves) will use open source techniques such as Google's TensorFlow in order to develop an AI toolbox to support their connected webshops.

Some of these platforms already run proof-of-concepts of AI solutions. In order to be able to train algorithms, e-commerce' process data is required of course. On-premise package installations such as Hybris and Magento do not make AI smarter. As a result, platform providers want webshops to run in their own cloud so that they can start up the required data flow. They want to see what happens in their customers' shops and in turn become a better platform.

Thus, Artificial intelligence is the driving force behind the move-to-platforms in e-commerce. That means the stand-alone shop will largely disappear because these webshops do not generate enough data. These stores will not be smart enough to compete. Mind you, these upcoming AI toolboxes of tech providers will only cover generic tasks, such as detecting fraud and predicting return of goods. This does not mean however, that if you use AI your webshop you will soon have some form of distinctive character - the bar will simply lay higher and you will have to adjust your daily operation.

"Big players mainly focus on mass. This offers opportunities for smart individual merchants, as they can create special experiences using AI in webshops."

Friso Geerlings - Chief Technology Officer, ISAAC

Online equivalent of a specialty store

So where can you make the difference? AI could also play a role in market- and product specific challenges. If you handle that well, your webshop can become the online equivalent of a specialty store. Fortunately, the big players focus first on investing in the most popular and easy to automate product categories, such as for example books. Their focus is mainly mass. This offers opportunities for smart individual merchants who use AI in order to create special experiences for customers.

For example, think about hyper-personalizing the user experience, whereby navigation and product-display is really based on knowledge of the customer; or do something really smart using taste in products like food and scent. Products that are not only chosen on a visual base, but are also  chosen  based on other senses, can perfectly be sold online with the help of AI.

Visual Search and interpretation of style

In addition, there are interesting possibilities in the field of Visual Search and Style Interpretation, particularly important for fashion- and lifestyle webshops. Visual Search helps the customer to look for similar products, offers interesting alternatives or just matches items. This is a huge growth market. It is the online version of personal advice and the way products are presented in a physical store, but also on advice from friends.

However, this is not easy. Turning raw data into those needed insights, requires a very specific infrastructure and a deeper knowledge of AI and integration. Being a merchant you will therefore need to look for that knowledge, or purchase it from a specialist. But if you do this well, you'll surely increase your service-level and conversion in all kinds of lifestyle market segments. AI enables you to offer online, correct and scalable recommendations for a customer who is looking for 'something summery' or 'something nice' after which you upload moderate smartphone photo.

Battle of the smartest?

Therefore, online shopping becomes customer driven rather than product driven. The webshop visitor receives advice that matches his preferences and context (what does he already have, what does he like, what is his age, what does he need it for, et cetera). The user does not have to plow through loads of information or set all kinds of filters. The e-commerce platform itself shows what the user needs or wants.

According to Gartner, by 2020, 85 percent of customer interactions will be managed by AI. In other words: 85 percent of your webshop's daily operations will depend on constantly improving and more complex AI, which in its turn requires more and more training data. This is largely the battlefield of e-commerce platform providers, who will increasingly be selected based on offered AI functions, rather than front-ends. They need to improve their cloud proposition, add AI and buy smart start-ups which are already well on their way. It will turn into quite a battlefield, and not all 'Gartner quadrant' players will survive.

The next generations' real online winners are companies which cleverly deal with a two-stage rocket. With an e-commerce platform in the cloud and not only using generic AI toolboxes from a platform party, they have their basic technology in order. Plus, they convert certain market- and product knowledge into information which is useful for customers,  aided by specifically trained AI. 
For example, an online lamp store will show the effect of a certain lamp light in your living room. A shoe store can project the fit of a sneaker on your foot. A fashion store shows if, and how an item matches into your existing wardrobe.

Due to move-to-platforms, self-hosted webshops will slowly come to an end. They will disappear or move to e-commerce platforms in the cloud which integrate AI functions. That will finally give the go-ahead for online specialty stores that can match or even surpass the physical shopping experience.