Catalogs are rapidly expanding and customers are moving faster through their daily life than ever before. The implementation of recommendations has become essential for the customer experience of anyone selling online. Why? Brands need to reduce the complexity of making decisions for their customers while continuing to drive value.
Well designed recommendation systems must be a simple exercise in matchmaking. That doesn’t cut it anymore. Instead you need to guide your customers on a self discovery journey by building experiences around recommendations.
Continue reading for some ways that you can create a great and innovative customer experience. However, first we need to look at some background information to create context.
In a 2004 Wired magazine article about how traditional brick and mortar stores would no longer be constrained by the limits of shelf space and for retailers that provided an opportunity.
What was the opportunity?
Traditional retailers favored the most popular products over niche (now called long-tail) products with lower sales volume. However, the article argues that with the advent of ecommerce, the long-tail products would have the potential to outsell the the more popular products because they cater to individual or unique tastes.
15 tears later, it’s 2019 and the “tail” is very long.
- Spotify has over 50 million songs in its catalog.
- Etsy has over 60 million listed items.
- Netflix offers up to 6000 titles in its library.
These numbers will only grow.
The challenge is satisfying the demand side. Supply is easy today. This is because there is a strong need to know the customer and provide recommendations that cater to the customer’s unique needs and interests. To do this the brand management needs to think about the many customer touchpoints of the customer journey where recommendations and personalization can add to the customer experience to provide value.
Now for the tips and examples!
Reduce the Complexities of Decision-Making
A huge variety of options in any catalog can be paralyzing. Or at least lead to infinite scrolling and previous time has passed before you know it and you gotta move on without having made a decision. The “paradox of choice” is well documented. The paradox of choice describes an experience when stress levels increase due to a large selection without any guidance. Cue the recommendation system.
When a customer visits a website or loads an app, there are several signals that can be captured and processed to aid in the decision-making process.
Let’s look at Foodpanda
Let’s take a look at the Foodpanda app. When a user opens an app for the first time, there’s very little known about the user. Which makes sense, right?
Yet the app still makes use of contextual signals. These contextual signals are:
- The user’s location filters the number of recommendations based on distance and delivery time.
- The time of day to further narrow down the options.
- Special deals such as free delivery or a free gift food with orders over a set amount.
Additionally, users visiting intent is communicated using the app through:
- Search queries
- The specific cuisine types visited
- Visits to specific restaurant pages
These are simply yet very powerful signals that can provide recommendations and can save a lot of time for the customers.
You can see that the core user experience employs several different widgets that cater to different user personas.
- The Order Again widget lets users quickly scroll through recent orders and pick something that the user would like to order again.
- The app also shows Featured restaurants that might encourage some more curious customers to place an order.
- Fast Delivery options are also shown. Users who might be short on time or are simply starving can see the food that will reach them the fastest.
- The app also has Recommended For You selections that take in both implicit and explicit user data to provide a tailored restaurant selection.
Personalization certainly includes the explicit data provided by a user. This can mean data from past orders, favorites, reviews, or how a user filters their search results within the app itself. Users of the Foodpanda app can filter selections by price range, special offer types, various cuisines, and particular attributes such as halal, BBQ, tapas, waffles, and many more to select for filtering.
Create “Connected Yet Unexpected”
It’s easy to recommend products of the exact same type as a previous purchase but that can come with its own customer experience outcomes that are undesirable for your customers and your business.
The image above shows the example of a “filter bubble”. A filter bubble occurs when a user is repeatedly shown recommendations based upon data from a small amount of categories.
You’ll find filter bubbles are the most evident on social media sites. For example on YouTube when a user watches a particular category of videos the recommendations will be the same type of video ad infinitum. The filter bubble and its effect on people who use sites like Facebook is regularly in the news because targeted ads were used to influence voters during the 2016 American presidential race.
Spotify Does it Right
Some brands have been able to overcome the filter bubble. Often this is done through the intervention of an editor and “featured” or “editor picks” recommendations encourage diversity as an explicit goal of the system. Another option is to offer a path to encourage independent “discovery” of new content or items.
Spotify’s recommendations for 3 Cats Labs founder Shane Hebzynski.
Spotify does this through a combination of daily mixes, curated content such as artist curated playlists, categories such as New Music Friday and Your Release Radar.
Twitter allows users to choose between an algorithm generated feed or a reverse chronological feed. Users can also choose to see personalized trends or non-personalized trend recommendations. Options like these allow users to take the initiative and move outside of the filter bubble.
There’s More Than the Core Experience
You’ll find that on multiple occasions visitors are interrupted in their journey to make a purchase within a single session.
Viewing your website’s analytics will show you where users are dropping off and leaving the website. These users who don’t complete the journey to a final purchase provides you with an opportunity to move beyond the core experience. This is where you’ll find opportunities to extend into email or push notification campaigns. These give yoru giant a gentle nudge forward.
Another example is when users search for content on your website. However, sometimes the content isn’t available. Often this can be true with Netflix. If you search for something not in the Netflix library, the search results will show similar alternatives.
The search results for Netflix might show similarly themed shows, shows with the same actors as the original search query. This gives a user plenty of alternative viewing options.
Downside Evaluation & Feedback Mechanism Creation
It’s important to analyze the cost of an incorrect recommendation. This is of paramount importance in industries like healthcare where an error in a recommendation has the potential to cause harm.
The way to overcome this and solve the problem is to hold group discussions with diverse participants and integrate feedback mechanisms into the product in an effort to mitigate potential downsides.
Some of the most common approaches to mitigating inaccurate or incorrect recommendations is to include downvotes, “see less” option for social feeds, hide recommendation buttons, or a full reporting module for greater depth of feedback.
What It Means for Customer Experience
Recommendation systems can be extremely personalized or follow general popularity. They can be context unaware and context aware.
Well designed recommendation systems can generate a great deal of value in terms of frequency of use, cart value, and retention. For customers, recommendation systems decrease decision-making complexity and open up moments to customer delight.
It’s important to understand the limitations of recommendation systems in terms of the filter bubble and cost of inaccurate recommendations. When designing a system there needs to be sufficient feedback loops and checks. The feedback loops and checks protect customers from harmful or divisive content.
Let’s Talk About Your Brand
shane@3catslabs.com | Call +65-3159-4231
Share
