What can a personalization engine do for you?

Let’s start with the basics, so what is a personalization engine (such as Frosmo)? According to the 2020 Gartner Personalization Magic Quadrant, it is a software that applies context about individual users to select, tailor and deliver messaging such as content, offers and other interactions. A classic example are dynamic product recommendations that become more and more targeted when a visitor browses through an ecommerce website.

How is it then different from a CDP (such as Custobar)? The first significant difference is that a personalization engine applies context, behavior, and AI-predictions to deliver relevant and timely messaging or recommendations to all website visitors. This works for everyone from the early funnel anonymous visitors to a returning customer. Customer Data Platform (CDP) is then again used to gather data and deliver consistent omnichannel messaging to known customers. Together, these two form a dream team that allows orchestrating customer journeys and delivering relevant experiences. Neither one requires third party cookies to operate on your site.

This article discusses how a personalization engine can be utilized to create relevant experiences for both new and returning site visitors. It also shows a sample journey explaining in detail how a personalization engine and a customer data platform work together to form a retail marketer’s dream team.

Let’s start with a new and totally unknown visitor that arrives on your site. The first objective is to make sure that they don’t bounce away. Here are three examples of what you can do to make the experience immediately relevant:

1. Use data about other visitors: “Others also viewed” is a classic example. If the new visitor lands on a product page, you can show what products other visitors viewed below the product information. This prevents the visitor from bouncing in case this product was not the one they were looking for. Since the visitor is new on the site, you can also highlight delivery or return policies!

1. Context is a big word and there’s plenty of contextual information that can be used to personalize the experience. Currently viewed page type is an example of contextual information which is always available. The most common page types in ecommerce are front, category, product, search and cart pages. Weather is also contextual information that you might want to use. When the first cold weather hits in late autumn, the demand for kids’ winter overalls skyrockets so retailers should be ready with their campaigns.
2. Content similarity algorithms can be used to find products or content items that are similar to the one being viewed at the moment. This takes into account the long tail of products that would not be picked up by a popularity algorithm.

As you saw there are plenty of options to counteract bouncing from the website using a personalization engine. A visitor’s first click on your site starts building up behavioral and preference data that can be utilized to drive conversions. Here are some samples of data dimensions to consider:

1. Journey: Visitors' needs and expectations are very different depending on the point on their journey. To start, Frosmo recommends using four steps: new visitor, discovery, about to buy, and has purchased as journey-based data points for personalization.
2. Segment: Website visitors are categorized into segments based on the behavior. This can be, for example, visitors arriving from a campaign and being placed in the campaign segment using utm parameters.
3. Affinity is the current favorite of most marketers. It is an AI-driven prediction created by a personalization engine that pertains to the visitor's main area of interest, which can be a brand, category or a specific product. For example, “Interested in gardening” as explained later on. Below is an image highlighting some product page personalization best practices such as complementary products, highlights of what others bought and AI predictions called “You might also like”.

Regardless of what recommendation strategies you choose to start with, continuous optimization should always be utilized to make small steps toward your bigger business goals. The are two basic methods available:

1. A/B-testing: Serve variation A to 50% of your traffic and variation B to the other 50% and check the results after statistical significance has been achieved.
2. MAB-testing (multi-armed bandit, bandit testing or “Earning while learning”): Multi-armed bandit experiments utilize machine learning and work by directing more traffic towards variations that have higher conversion rates. This is especially well-suited for campaigns where time is limited.

Now you know how a personalization engine works, so let’s take a look at how a personalization engine and a customer data platform help orchestrate a sample customer journey.

As an example, let’s use a retailer who sells garden tools among a variety of other products. How is the customer’s journey orchestrated by a personalization engine and customer data platform together?

1. A potential customer (let’s call them NN) searches for a garden tool set and lands on the product page of an ecommerce retailer. Unfortunately, the product is out of stock, which means a huge bounce risk for the retailer. Luckily, a personalization engine can use product similarity AI and recommend similar products. Customer goes through the suitable options and then leaves to compare prices at other sites. At this point, a personalization has identified that the main affinity of NN seems to be “Gardening”.
2. When NN has done the comparison and decides to return to the site, they come back to the front page. At this point, the personalization engine helps them pick up where they left off by offering a shortcut to recently viewed products. This helps NN to make the decision. When a personalization engine notes that NN has products in their basket, it immediately offers relevant complementary products such as gardening gloves which NN will eventually need but has forgotten about.
3. As NN makes the purchase, they also become a known customer in the CDP. Based on earlier behavior, they are placed in the “Gardening” segment.
4. Now the CDP can start communication with NN across relevant channels. NN will see personalized gardening ads on Facebook. The newsletter content is also directed to their interest and contains for example inspirational gardening articles.
5. Consistent communication across channels ensures that this retailer is the first option in NN’s mind when a new garden equipment is needed. This grows the loyal customer base of the retailer for both ecommerce and brick and mortar stores.

This clearly shows that both a personalization engine and a customer data platform have their place when it comes to creating smooth and friction-free customer experiences.