The retail landscape has changed massively in the last two decades. It’s been over 20 years since Amazon disrupted the online retail space and has since adapted to a personalised, data-driven shopping experience.
“Most retailers today use data to get people to the store; Amazon uses the data to keep them there.”
I mean that companies spend significant marketing budgets to bring traffic to the store, i.e., marketers spend most of their marketing dollars on Google, Social media marketing, and remarketing to entice customers into the store. At the same time, very few focus on engaging and keeping people in the store once they land on their website. The real opportunity is to increase sales by using data and Artificial Intelligence to engage customers more deeply in the store and protect them from competitors.
As with anything new and exciting, there can be a lot of confusion, and Artificial Intelligence is no exception. As a retailer, finding the right technology can be extremely tricky and time-consuming. If you’re a small retailer, it is easier to fall into the trap of using incompetent systems at the cost of providing a poor customer experience for your website visitor.
In this blog, I wanted to give a bit of background to the different approaches that use data to help customers and lay out the pros and cons of each.
LEVEL 1 - SEARCH - DRIVEN CX
Simple tools like search and filtering work well for customers in their final stage of the buying journey. However, it’s important to remember that most customers arriving at your store are still in the ‘consideration’ phase of their shopping journey. By not providing more than a simple search, you could be losing the chance to engage them in your store more deeply.
- They are the most basic requirement for an e-commerce site. You have to have them!
- It makes it easier to get to specific products quickly.
- Customers don’t stumble upon products they hadn’t considered.
- Products can get excluded if they’re not tagged well.
LEVEL 2 - BRAND-DRIVEN CX
The next step is to provide a level of suggestions created by merchandisers and marketers. There are various problems with this method, but the two big ones are:
- It is impossible to provide an individualized customer experience with a system that puts your customers in broad categories. Moreover, it simply focuses on what ‘you’ want to sell rather than what the consumer wants to buy. The post-pandemic shoppers demand a store-like online shopping experience, and retailers need to re-visit their e-commerce strategy.
- It consumes valuable people. To do it well, retailers are dedicating valuable staff to managing rules in these systems. It’s not sustainable, causes significant bottlenecks for launching new products, and is very costly.
- Works well for product variants, for example, different colours.
- Works well where there are only a few SKUs and a low level of new products.
- Consume resources (people) who could be better employed.
- Hugely biased towards the opinion of the merchandiser.
- Takes no account of customer preferences.
LEVEL 3 - ‘Data-Driven’ Recommenders
Today, various e-commerce systems provide ‘recommendation’ capabilities based on the products viewed/bought by other customers. Although this appears to help, it only works well when all the products have good ‘click data’. For example, new products have no click data, while older products would have had more clicks and views, which makes these recommendation systems biased towards older products.
So if you’re an e-commerce store with a fixed product catalogue, these recommendation systems work reasonably well. But if, like most retailers you’re adding new products regularly, these data-driven systems are highly flawed, and won’t work in your best interest.
- They have customer context.
- They adapt to the customer, usually at a segment level.
- Products need data to compete, making them biased toward best sellers.
- Difficult to manage new products which won’t have any historical data.
- Very biased towards the last product a customer clicked on and did not take the entire customer journey into account.
LEVEL 4 - AI Customer Experience in Real-Time
These systems understand ‘how’ people shop instead of just focussing on what they’re looking for. They can create very complex associations to match customers with products. AI systems can understand how different products work together, whether it’s new or old products.
True Artificial Intelligence systems are fully automated and don’t require merchandisers or dedicated people to overlook shopping behaviours. They learn from customers and adapt to how they change from day to day and season to season.
Unlike recommendation platforms that assist customers using historical data, AI Shopping Assistants learn from customers in real-time. They start learning about the customer from the very first click and engage with them while they’re still in your store, nurturing them to become loyal customers.
- The best combination of customer relevance and showing a wide range of products
- If done correctly, works with new products and products with no data
- Removes all bias as the customer is listened to
- In the past retailers have struggled with the cost of these systems, the skills required, and the timeline for the projects. They have prevented retailers from deploying such systems, but these barriers have been removed by companies like Shopbox.
True Artificial Intelligence systems were out of reach for most retailers in the past. They required a significant investment in both people and systems. Because of that, retailers invested in systems that a merchandiser could control but added little value. With the emergence of “drop-in” AI systems for retail, such as Shopbox, retailers no longer need to compromise.
If you’d like to hear more, let us know.