Visit page
Skip to content

Online Groceries and Artificial Intelligence: 7 tips for a sales increase of up to 30% in the online supermarket

Buying fashion accessories or books online – that has become the norm for most customers. However, online shopping behaviour has remained more cautious in the online grocery industry. However, forecasts are beginning to suggest that strong growth can be expected in this area over the next few years; in Germany, it is set to rise by around 5 percent, which corresponds to a sales volume of 6 to 8 billion Euros.

The prime advantages for customers are an end to the reliance on opening times, home delivery and enormous time savings. According to a Bitkom study, every one in four online shoppers has purchased food online at least once.

However, an online supermarket must differentiate itself in many respects from online shops in other industries – speed and simplicity are especially decisive in e-commerce for online groceries. Customers normally know in advance which products, brands and quantities they are intending to purchase. They simply want to put “their” butter and “their” pasta in the shopping basket – together with other articles that they require on a regular basis – and they want to do that as quickly as possible.

The online grocery sector has very special requirements in terms of user guidance and product search. This article summarizes the most important tips and tricks for increasing sales in your online shop.

1. The Predictive Basket: AI handles food shopping nearly automatically

Everyone has to buy groceries, but it definitely doesn’t top the list of fun things to do. In that respect, the Predictive Basket is a much welcome help for online shoppers. The Predictive Basket is a new solution from FACT-Finder, powered by Artificial Intelligence. Put simply, it predicts what the customer wants to purchase in the current session. The AI calculates the suggestions based on the behaviour of each individual customer and on the behaviour of all other customers in the database. In this way, it gets to know the purchasing patterns of the customer and can improve how it adjusts to these.

The Predictive Basket is already successfully being used at the Austrian food wholesaler Kastner, predicting more than 80% of what their customers are going to buy. The tool reminds shoppers about items they may have forgotten, but which they probably should have on their shopping list based on their past shopping habits. The personalised proposals make the customer feel understood and treated as an individual – a completely new level of customer focus.

Buy more, forget less. The Predictive Basket tells customers in advance what they probably need to buy today, increasing the speed of the buying process by up to two-thirds.
Infopaper Predictive Basket

2. Personalise search results – because tastes differ

Customers have their own individual preferences when it comes to food. Brand, quality and price are just some of the product properties that people take into account when shopping. The more an offer matches a customer’s tastes, the more likely it is to be purchased. The objective should therefore be to offer individual shopping experiences to all customers – dependent on their personal preferences.

Tracking data can be used in e-commerce for effective implementation of personalisation concepts. A solution that effectively personalises the ranking of search results has proven most effective; for example, organic products will be ranked higher than conventional products for a customer who often clicks on organic products to place them in the shopping cart. The same is true for gluten-free or vegan products, or in terms of brand preferences. Individual ranking improves the shopping experience and increases the likelihood of a purchase being made, whilst also boosting customer loyalty and satisfaction.

Your customer has already placed organic potatoes in their shopping cart; your online shop can respond by showing organic products first when this customer next makes a purchase.

3. Use the suggest menu as a sales arena

Some customers are in a rush with their food shopping, especially those on mobile devices. It must be possible to place the desired products in the shopping cart as quickly as possible. Instead of just using it to show completed search terms, you can turn your suggest menu into a sales arena. You can do this by integrating product images with a shopping cart function. This allows the selection and ordering of products after just a few letters have been entered.

Quick and easy: While the customer is entering “tomatoes” in the search field, the system is already generating proposals that can be placed in the shopping cart with just one click.

4. Use antonyms to increase the relevance of search results

You can make the purchasing process significantly easier for customers by excluding superfluous hits and irrelevant products from the results list. The easiest way to do this is to define antonyms for the search. These are terms with a meaning that is totally different to a specific search term. One use case would be the antonyms “hair tonic” or “watermelon” for the search term “tonic water”: this ensures that neither hair tonic nor watermelons appear in the results for “tonic water”.

There’s water, and then there’s water. It is fair to assume that a customer searching for tonic water doesn’t want to find hair tonic, watermelons or water bottles.

5. Guided Selling supports customers in making selections

Customers can be spoilt for choice; the customer support that makes retail stores so special is often missing in e-commerce. However, guided selling functions in the search system allow you to map interactive advice online, too. After specific search terms have been entered in the search field, product-specific questions then appear that lead the customer to the right product.

For example, a question-answer process for the search query “wine” can find out which types of wines the customer would like and then filter the results after each answer, leading to only the most relevant results for that particular customer. Simulating a real consulting situation allows a large range of products to be filtered quickly, according to individual requirements. The effect: customers feel that they are buying just the right product, which can increase the conversion rate and customer satisfaction.

Red or white? Little questions can help you to help your customers find the right product quicker.

6. Recommendation Engine – recommend suitable additional products for your customers

Spaghetti and tomato sauce, flour and eggs, bread and cold cuts – use combinations like this to recommend suitable additional products for your customers. The Recommendation Engine, a tool that investigates which products are often purchased together, can help with this. Based on this, sensible additional products can be recommended to the customer. An important factor to bear in mind is that these recommendations need to be individual, as they are adjusted to suit all of the contents of the shopping cart.

For example, recommendations can offer basil as a suitable additional product for tomatoes and mozzarella. As with the search results, it must also be possible to personalise your recommendations. Depending on the customer, offering the premium detergent or the discount variety, or whole-wheat bread as opposed to standard wheat bread, can really make a difference. Suitable recommendations increase the average order value significantly.

Customers searching for eggs might also need bread, ham or cheese.

7. Location-based search – the must-have for omnichannel strategies

A London resident searching for “beer” in an online grocery shop normally wants to a buy a different brand to the one a shopper from Belfast is looking for. Take local differences into account and offer suitable products to your customers – someone from London is more likely to put London Pride in the shopping cart than Guinness.

A location-based search is especially important for the click-and-collect concept (order online and pick up at the local store) in particular. The search must recognise the customer’s location and adjust itself to take the assortments, local products and different prices into account. This allows the online shop to drive sales for the physical retail store.

When customers in London search for “beer,” they see the proposal “London Pride.” However, customers in Belfast see the proposal “Guinness.”

Conclusion: make the most of your online potential!

The aforementioned tools and tips all help to increase sales and boost customer loyalty. However, there is one feature of retail stores that even the best online grocery shop cannot replicate: the feel. That is simply because the sensory part of the internet has yet to be invented. You can compensate for this – to a degree, at least: by offering relevant content in the form of recipe ideas, detailed product specifications or information about the manufacturing process, for example. Information like this is useful for the customer and helps with the decision process. Ensure that this valuable and costly content is not just hidden in the navigation, but can also be found with the search.

You also need to provide product data that is as complete as possible and maintained regularly. Great product data means that your key functions like search, filter navigation and personalisation can work even better, which in turn improves your sales figures.

About the author


Stuart Patterson

is the Sales and Operations Director at FACT-Finder for UK and Ireland. Stuart brings a wealth of ecommerce experience having worked in both UK and Irish markets extensively. He has worked with small entrepreneurial start-ups to Global brands such as BMW, Vodafone and Mindshareworld. Stuart brings a fresh outside the box type of approach based research and best practices. He has 10 years ecommerce experience across all verticals. As a previous business owner he knows how to create bespoke ecommerce solutions tailored to specific requirements and objectives.

More interesting topics

    Leave a Reply

    Your email address will not be published.

    This site uses Akismet to reduce spam. Learn how your comment data is processed.