Each variation is automatically crafted for the individual consumer, and decisions are made based on the consumer’s unique tastes and interests

The first time Amazon showed me the perfect book for me via their recommendation engine, I was truly amazed. The idea that a retailer could not only recognise me as a return visitor, but also learn their interests and alter their experience accordingly, felt like magic. The secret behind this seemingly magical shopping experience? Dynamic content.

The big data revolution has transformed how marketers approach every aspect of their work, from how they design campaigns to how they measure success. At the heart of this revolution is the goal of creating genuinely personalised experiences for consumers and also measuring their results more precisely than ever before. Dynamic content is one of the primary ways companies tailor experiences for their consumers. Combined with machine learning, dynamic content enables companies of any size to create genuinely personalised experiences that are fresh, relevant, and cohesive.

What is dynamic content?

Dynamic content is a piece of content that changes based on a user’s interests, behaviour or demographic data. It is used in emails, websites, landing pages, ad units, and more. Dynamic content is a powerful way to engage with your audience meaningfully. As a business practice, it has quickly becoming table stakes.

Dynamic content can be as simple as inserting a consumer’s first name into an email subject line. For example, “Hi Jane, check out our latest products!” Even this small bit of personalisation has been shown to increase open and click-through rates.

Dynamic content through machine learning algorithms can change every part of a user’s experience. Each variation is automatically crafted for the individual consumer, and decisions are made based on the consumer’s unique tastes and interests. This type of dynamic content works best for brands with lots of customers and lots of products — E-commerce sites typically personalise this way.

All of these techniques are designed to tailor the consumer’s experience to that person’s specific interests and make it personally relevant to them, whether they’re existing customers, prospects, or first-time visitors.

Also read: Indonesian AI platform for natural language processing Bahasa.ai gets seed funding from East Ventures

The two elements of relevance

To ensure content is super-relevant for each consumer, you need two things: an understanding of (1) each consumer and (2) each product:

1. Understanding each consumer

In the Amazon era, it is essential to understand each consumer’s unique tastes and interests. How?  By creating an ‘interest graph’ for each consumer. An interest graph incorporates the full spectrum of the customer’s behaviour, including search queries, browsing activity, and purchases, instantly adjusting to the customer’s actions while incorporating prior history.

For example, someone may be searching for a party dress and a clutch handbag – an interest graph will automatically infer they may be going to a party and can recommend the perfect pair of evening shoes. The shoes may be quirky and unusual: shoes that most people wouldn’t appreciate – but they’re perfect for this specific person. An interest graph enables each customer to get personally relevant recommendations- the kind you’d get from a personal shopper who really knows your tastes, and the kind that drives real engagement.

Spotify, Pandora, and Netflix are best-of-breed examples of brands that use interest graphs.  Each builds an interest graph for every consumer, enabling them to serve a curated stream of personalised entertainment recommendations.

2. Understanding each product

The problem with existing product recommendation technologies is they don’t fully understand product attributes. They either rely on existing metatags, or there may be some simple machine learning involved, which can really miss the mark. The solution? Natural Language Processing. Natural Language Processing (NLP) is a technology that deconstructs product descriptions into tokens. In retailing, NLP deconstructs product descriptions into tokens like ‘Tommy Hilfiger’, ‘slim fitting’, ‘flare shape’, ‘shimmering’, or ‘work to evening’.

NLP is widely used across the web and powers the grammar checks used by Google Docs and Microsoft Office, as well as more sophisticated applications like Google Translate.

To show how it works, let’s use the sentence “The dog began to bark.” The word ‘bark’ can either refer to (a) the skin of a tree or (b) the sound of a dog. In that sentence, Natural Language Processing can see the word ‘bark’ is preceded by the word ‘to’ and is therefore probably a verb, and the word ‘dog’ is two words away. Based on these datapoints, NLP infers that ‘bark’ is the sound of a dog, rather than the skin of a tree.  The goal is to look beyond the words in the product description to see the actual meaning.

Why is NLP so Important to deliver dynamic content?

If you can understand the product’s attributes and the user behaviour at a deep level, this allows you to know what product attributes are valuable for each consumer, and therefore deliver a genuinely personal shopping experience. Read more about NLP here: ‘Creating Serendipity via Natural Language Processing’.

Creating dynamic content

Now that we’ve seen how dynamic content works, what does it look like in practice?

  • Individualised communications. Whether as emails, ads, or some other channel the ability to communicate with each consumer on a 1:1 basis is a great way to boost revenues and engagement.
  • Targeted recommendations. For retailers and e-commerce brands, this can mean offering recommendations based on a combination of personal tastes and observed behaviour.
  • Customising landing pages. You can create curated landing pages based on the keywords and ads that brought the consumer to your site. Did someone reach your website by searching for “evening dress”? If you know them already, you can make sure the first thing they reach is a custom landing page that showcases dresses that matches their unique tastes and interests within the occasion wear genre. A system that offers 1:1personalisation can allow you to create billions of landing page permutations.

Already, 35% of what consumers buy on Amazon and 75% of what they watch on Netflix comes from product recommendations based on personalisation algorithms.

Also read: 4 ways artificial intelligence is innovating e-commerce

Conclusion

As retailers adapt to the ever-changing technology landscape the need to adopt new models has become really clear. To stay ahead of the curve – and perhaps to stay in business – retailers must not only pay attention to evolving consumer preferences but anticipate them and ‘surprise and delight’ with an infinite stream of fresh and relevant content.

The trends that will most affect the retail industry’s future are evident, and the requirements are clear. The time to act is now. Retailers that act now will be the winners when the next chapter of retailing history is written.

—-

This article was originally published on the Mercanto blog.

e27 publishes relevant guest contributions from the community. Share your honest opinions and expert knowledge by submitting your content here.

Photo by rawpixel on Unsplash

The post Dynamic content in the era of machine learning appeared first on e27.