If you have a product recommendation tool on your website, visitors are 22% more likely to make a purchase. This is how you can create your own recommendation engine quickly and easily.
Giant E-Commerce companies like Amazon and Sephora spend millions on creating recommendation engines to personalize their product selections for each person who visits their websites. They have complex algorithms and programmers who do this for them, but there is an easier way to recommend products than spending millions and hiring developers. Today I want to walk you through an example from Murdoch's Ranch & Home Supply because they have a really cool snow blower recommendation tool on their website that we can all learn from when it comes to E-Commerce personalization.
Before diving into this I just want to clarify what I'm talking about when I say "Recommendation engine" what I mean is a tool that asks people who visit your site to tell you about themselves and their needs, and then gives back a recommendation for a product or products that are the best match for who they are. Basically asking people what they're looking for and then showing them what they need.
I want to take you through a really old school and down to earth product recommendation tool to show that this works with any kind of business.
The tool they created helps you find the right snow blower based on your needs, this is something that might happen in-person if a store associate was helping you pick out a snow blower, but since the interaction is happening online it is being done on their website.
Now I want to pause for a second and bring things back to reality. I'm throwing around the term "Recommendation Engine" as if it's some hyper-technological magic machine that will solve the world's problems, but what I'm really talking about is making sure someone gets the right product they need.
If you've ever made online purchases, you know the frustration that comes when the product arrives but it isn't quite right, so you have to package it back up, return it, and try again. To date there really isn't a great way to know if you're choosing the right product, most stores have some selection criteria but you don't really know what you're looking for every time and self-selecting often leads to choosing the wrong thing.
From the stores persepective every single one of these mistakes results in a costly transaction. The status quo is now that you not only accept every return but also pay for shipping, and often-times you can't re-sell the product that was returned so you have to eat that cost as well.
One of the ways you can work to fix the issue of getting the wrong product is using a recommendation engine, which in actuality is just an online tool that asks the same questions you would in real life to someone who was choosing between your products.
In this post I want to show the details of Murdoch's recommendation engine and hopefully it sparks in you ideas for how you can use one on your own E-Commerce store.
I'm going to picture the questions of the Murdoch's tool to the right and talk through each step as we go.
The second question on their recommendation tool is asking how large of an area the snow blower is going to be used for. This makes sense as the follow-up to the first question about how deep the snow is.
I used to work in landscaping (where I lived there was no snow, so dirt was the closest substitute for snow). But when we'd do estimates for dirt removal the first two questions were always "how deep is the sand?" and "how big is the area?" because you need to know both before giving a cost estimate.
These snow blower questions are the same, you need to know the details of the area before going with an estimate on which tool you need.
The third question the recommendation engine asks about surface type. Now again I don't know anything about snow blowers but I'm assuming that a rough surface needs a different blower than a smooth surface because the rough surface might have gravel that you don't want to be picking up.
Pro tip: If you want your recommendation engine to be taken seriously and for people to actually trust whatever products you show at the end of it, then make sure your questions ask the things that really matter. For example, if this question was off-topic like "What color is your favorite?" (sorry, that's a stupid example, but you get the idea), then the quiz wouldn't be taken seriously because that obviously doesn't matter in the context of finding a good snow blower.
The next question brings up a really important point about the use of graphics in a recommendation engine. Using pictures and illustrations like the ones pictured to the right is a perfect way to make your tool more useful and have people actually answer the questions.
Graphics can mean the difference between a tool that is helpful and one that is confusing because often times you just need some pictures to put things in context.
Once someone finishes the quiz on Murdoch's website they are automatically redirected to a product page that says "Your results" with two snow blower options at different price points for this individual.
How does the scoring system work?
If you haven't already wondered, by now you're probably thinking "How does this work in the background?" "What connects the answer choices to how products are recommended?" those are good questions and I have good answers for you. There are really two ways that a recommendation engine connects up someone's answer choices to their results.
This where is the products are the "Personalities" and as you go through the questions of the tool you are being assigned points towards the various personalities based on which answer choices you select within the question. As you go through the recommendation engine you are getting "+1's" for different products based on your answer selections and whichever product or category of products has the most "+1's" assigned to it at the end of the questions is the one you see as your result.
With a branching logic quiz you can ask people different follow-up questions based on their initial responses. For example, if someone chooses "Deep snow" then you can show them a different second question than somoene who chooses "Light Snow" This allows you to create "branches" off of each question and lead people down very different paths based on how they are selecting answers.
Recommendation engines are powerful, there's no doubt about it. Research from Amazon and similar sites shows that recommended products on a home page convert 22% better than generic selections. Hopefully this guide sparked some ideas on how you can create a recommendation tool within your own online store, click below to start creating a tool right now.
Share this post