Skip to main content

Build better search personalization with control and visibility

 

Today, everybody agrees on the positive impact of personalization, for consumers and businesses alike. From simple demographic based website personalization to comprehensive behavioral based multichannel personalization, every step towards a more personalized experience is a step in the right direction.

omnichannel personalization

Having said that, how do you do personalization right? Which strategy is the right one for your audience and your business? What is the line between a creepy personalization and a personalization that really enhances the experience — and serves your bottomline?

Algolia offers Personalization to enable businesses of all sizes to deliver fast, tailored and personalized experiences across any channel. Today we’d like to double down on the importance of  total visibility and control of your personalization strategy.

Common personalization pitfalls

Personalization done wrong is worse than no personalization at all. Poor implementation has the potential to annul every uplift brought by personalization. Or even worse, damage the experience and drive your users and shoppers away from your brand. Here are key pitfalls to watch out for.

  • Improper data management. Let’s take data. A personalized experience is only as good as the data it relies on. But with this indispensable data come colossal responsibilities for businesses using it to drive their personalized experiences. Data collection should be transparent and based on consumers opt-in. Data storage and processing should comply with the highest security standards. Data should be used for very specific, clearly stated purposes. Handling data leaves no room for approximation: not only it can quickly be unlawful, but also demolish any trust consumers had in your company. 75% of consumers said that if they don’t trust a company to protect their data they won’t buy from them no matter how great their products are.

Unfortunately, the complexity doesn’t stop here.

  • Taking into account the wrong variables. A lot of variables can be taken into account to power a personalization strategy, however not all of them are relevant. Imagine Netflix-like use case. One of the parameters the personalization could use to personalize the content surfaced to users is the number of words in the title of the movies. Would it make sense and enhance the experience? Probably not, no matter what state-of-the-art personalization technology you use.
  • Making your personalization…a little too personal. Going further, when not well configured, personalization can become creepy. Personalization that follows consumers across their online activities, outside of a given brand’s properties, or that is too strong, will deter those consumers.  63% of them say they would stop purchasing products and services from companies that take “creepy” marketing too far. A personalization applied too strongly will not serve the experience, but rather constrain it. It will add friction for users to discover and enjoy the breadth of your offering.

As Jay Baer puts it, “[…] we cannot ever forget that our audiences—not us—are those that decide what is satisfying and what is creepy. They decide what is clever and what is stupid. They decide what is in bounds and what is out of bounds.”

Visibility and control are fundamental

You won’t be able to quickly assess the behavior of your personalization, nor optimize it, if you have no visibility on the logic behind it. Of course you will probably A/B test a personalized vs. non-personalized experience before rolling out your personalization to all your digital properties and all your traffic. But you’ll still be exposing a potentially broken experience to part of your traffic for several business cycles, and losing those users, as well as wasting this time before iterating.

Giving the keys of a core element of your user experience to a black box has its downsides even when it works well. You won’t be able to know why your personalization strategy is performing well, and to apply those learnings to other parts of your experience, or your business altogether. Visibility of the personalization logic is therefore critical.

Visibility unlocks even more value when you can act upon your learning, and define this personalization logic. Like in any conversation, context matters. Instead of blindly applying a personalization algorithm built for any business, control on the personalization logic allows you to adapt it to your business and your audience. If you know that a visit to a product page is less important than an “add to cart”, or that an interest in a brand is more meaningful than an interest in a color, you should be able to apply it to your personalization.

Personalization is key in every part of the user experience, and we believe it is especially true for search and discovery. Search and discovery are where consumers will most clearly state their intent. Answering them with irrelevant, non-personalized content is not an option.

Introducing Algolia personalization simulator: tale of algorithms, insights and human expertise

At Algolia, we believe that our customers should have full visibility and control over the search and discovery experience they build for their business, and their audience. We enable our customers to combine their knowledge, the data we surface to them, and our algorithms in order to achieve best results. And it is with this mindset that we’ve built our personalization.

Algolia Personalization Simulator is the human-facing side of our personalization engine.

personalization simulator

Control:

  • With this tool business stakeholders can quickly create personalization rules based on ingested customer behavioral data — such as click data and purchase data  — to tailor search results for customers. They can define what events to take into account (such as page views, add-to-carts, etc.) and how important they should be in the personalization strategy.
  • In parallel, they can define how important the affinity to each attribute of their products or pieces of content in the personalization logic is. So based on their knowledge of their business and their audience, they can for instance define if an affinity to a brand is more important to an affinity to a color.

Visibility:

  • The personalization simulator allows business teams to visualize the effect of changes to their personalization strategy in real time. They can quickly see how their configuration changes search results and impacts the end user experience. With the simulator, users can easily toggle across different user profiles to ensure their personalization strategy makes sense and actually enhances the experience for real life users.

Transparency:

  • A side but nonetheless extremely valuable effect of this personalization approach it that it allows businesses to expose the personalization profiles computed by Algolia to their users. As discussed above, consumers are becoming more and more demanding about their personal data, and are turned off by invasive personalization. Exposing them why they are shown specific products or content is key to gain their trust.

Use case: a large luxury fashion marketplace

Let’s look at a real life use case of this new tool. A large luxury fashion marketplace in Europe is using Algolia to power their search and discovery experience.

Given the breadth of their catalog (~1M items), it’s reasonable to think that all their shoppers aren’t looking for the same items. So,they decided to offer their shoppers a personalized experience — one that made sense for their audience, business and catalog.

Their business is unique which is why the ability to control the personalization logic matters hugely. For instance, people clicking on very high-end handbags might just want to have a look at them and not actually buy them. Therefore, the ability to tell Algolia that an add to cart or add to wishlist is a way more important event to consider than a simple click on an item is critical.

In addition, they’ve seen that shoppers tend to buy items from the same brand, while the color for instance isn’t that important. That’s why they decided to optimize the personalization logic to show their shoppers items of the brands they interact with the most in priority. A black box personalization tool that optimized for criteria that didn’t make sense for them wouldn’t have been productive.

Great user experiences delight users at every step of their journey

Personalization is only the cherry on the cake of a great experience. We have more tips and best practices for your multi-channel experiences.

And if you prefer a personal tour, we’ll gladly connect you to a search specialist.