How digital recommender systems are changing personalisation

How digital recommender systems are changing personalisation

Over the last few years, the field of digital marketing has gone through significant change. As emerging technologies, artificial intelligence (AI) and data continue to drive business strategy, digital marketing has a limitless number of possibilities. However, the basic premise of marketing which is to attract new customers remains, but we now have a gluttony of channels to help us do so.

SEO, PPC, social advertising, retargeting, affiliate marketing, search advertising, email and SMS should be part of any standard digital marketing strategy. Those that are pushing ahead are taking advantage of the “new kids on the block” such as recommender systems. A recommender system is one of the latest additions to the list of marketing tools made available through technology. In simple terms, it is a digital script or algorithm that provides recommendations to a consumer similar to other choices they have made. Some of the most well-known recommender systems are Netflix, Spotify and Amazon. All these websites recommend products to the consumer rather than pushing them to make their own choice. In a fast-moving world where nobody wants to spend time browsing, recommender systems are having a huge impact on profit. In fact, Netflix have reported that as much as 80% of shows customers watch are those which have been recommended. Digital marketing teams can only dream of those sorts of rates via PPC, email or SMS.

Why have recommender systems changed digital marketing?

Customer segmentation and personalisation are not new to the world of digital marketing. For a long time, organisations have been sending emails and content based on what they believe are the customer preferences. For much of the time, these were based on what the consumer did previously. For example, you might get an email telling you about your recent orders and maybe some new products coming soon. These communications might use your name to feel more personal and occasionally some other data that the company has collected.

The recommender system can create a real-time digital experience. Going back to the fact that 80% of shows watched on Netflix have been recommended by an algorithm. There are thousands of shows to search from but eight times out of ten, subscribers would rather be told what to watch.

All of this is created through the behavioural data of the 100 million plus Netflix subscribers. Every time a user completes an action on Netflix it learns for itself. This means that everyone gets a tailored recommendation based on their habits and others with similar habits to them. This can even go as far as looking at what movies you stop watch half way through or making assumptions on shows you find too scary. The thumbnails users see for shows are always different based on their likelihood of clicking an image (check with a friend and you will have different images).

Where Netflix is tailored so precisely, the search menus almost become irrelevant and so does the use of keywords.

If we put that into a digital marketing context, so much focus over the last decade has been on keywords to improve SEO and drive PPC conversions. However, if 80% of consumers prefer recommendations over choice, does this mean the approach to digital marketing need a few focus? The goal is not to guess what a consumer might want to buy but more about what they are going to buy next.

Amazon achieves this quite perfectly by providing recommendations at both consumer and at product level. Like Netflix, they track all user behaviours and shopping patterns to create real-time personalised digital experiences. At least 35% of Amazon sales are generated via the recommendation engine which is available at all stages of the customer journey. This generates a vast amount of revenue that traditional keyword digital marketing would never achieve. In 2019, 90% of Amazon sales have come from consumers searching directly on the Amazon site rather than needing to go back to a search engine This shows the power of the recommendation engine and needs to be considering within a digital marketing strategy.

Limitations of recommendations

Whilst recommendation systems have shown to increase revenue, there are risks to over adoption in digital marketing. If consumers always stick to things they are recommended it could lead to them being stuck in a bubble. For example, consider somebody who likes buying shoes on Amazon. The likelihood is that the more shoes they buy, the more they will get recommended. A digital marketing strategy needs to think about how the recommender system can get consumers out of those bubbles, else run the risk of seeming too biased. One such strategy could be after X number of shoes, recommend other items similar behaviours and demographics have purchased.

To recommend or not to recommend….

Digital marketing is going through a period of change and businesses need to adapt to the consumer demand. New technology such as recommender systems are a big part of that and the results from Netflix and Amazon clearly demonstrate their popularity and impact on income