I love Netflix, Pandora, Last.fm, and, naturally, Amazon. They all do a fantastic job in finding what I want, and providing me with relevant product recommendations. They manage to dig in deep into their products and “inflate” the goodies that I really like. I am a loyal customer. Loyalty is brought by understanding customers and delivering to them what they want or value.
The wet dream of every marketer in becoming the next Pandora, Netflix, or Amazon, but providing personalized content recommendations as amazingly as they do is hard to achieve: systems need to speak with one another, there’s a need to find meaningful connection between content and the need to thoroughly go through semantics as well as consumer behavior. That being said, the cost of getting all this done can add up very quickly and become very expensive. Once this reality sets in, most managers would start excusing themselves from this undertaking along the lines of “I’m already busy now and have good engagement from my customers anyway. Besides, my IT team has plenty to do without me breathing down their neck…”
Climbing a mountain always looks like a big task, but it always begins with a first step and that deep breath of courage. Let’s discuss quickly what product recommendations engines are, what they can do for you, and how relatively simple and inexpensive (surprisingly!) they are to implement.
What Are Product Recommendation Engines?
Product recommendation engines or systems are in charge of finding relations between products/services based on the inherent complementary nature of items (because you bought this, we recommend you that), based on what clients with similar behavior did/bought (people also bought/viewed), and according to the crowd popularity (trending now… top sellers).
How Do They Work?
The recommendations are calculated by algorithms that make sense of all the information that is harvested from the website or app (which not only acts as product or service inventories but an actionable marketing database), and from transactional data. Sounds pretty straightforward, right? I believe so as it’s basically letting your current infrastructure and setup work for you; you just have to add software that can compute those algorithms, and there are quite a few in the market at fairly reasonable prices.
The Benefits of Product Recommendations
So, here are a few ideas and cases where marketers can benefit from personalized content recommendations. (By the way, such engines are capable of increasing sales and conversions consistently by 10 percent to 30 percent, which is why you want to have it!)
First and foremost these systems recommend products. It creates a feed that can be used in your website, in email campaigns, and other channels.
Enriching Subscribers’ Profile/Web Analytics
Now that you know which books and music I like, we can move on to subscribers data-mining.
It is hard, really hard, to receive great quality data from subscribers. You can mine data from social CRM systems as well, but in this case I am referring here to websites or apps. Integration between these and recommendation engines harvests real-time and historical data from your clients such as frequency of visits to the web pages, which pages they have visited, duration and time spent visiting specific pages, product(s) clicked on, purchases made, etc. Everything is and can be collected. Your customers will now have much more information in their profile without them needing to populate and manually fill in forms, and you can start understanding who they are and how to serve them better.
The majority of marketers (globally and not just in Asia) use Google Analytics. While I think the product is good, there’s almost nothing you can do with the data collected – it doesn’t belong to you, and it is very hard if at all possible to integrate this data and link it to the recipient level.
Recommendation engines do not replace web analytics, but they let you collect a lot more information of your users’ behavior and transactions with your business, and systematically store it in your users’ profiles for future queries. Call it actionable information.
Ad Targeting (Retargeting)
With integrations between recommendation engines and ad exchanges, you can target your clients on other websites with products they’ve liked on your website. You have the true ability to “meet” your customers during their trail on the web.
You can increase revenues by simple actions. For example,
- A purchase confirmation with recommendations of matching products and services.
- Abandoned shopping carts.
- What customers are buying now (this can come in email or posted on landing pages).
- Other customers who viewed/purchased (once again you can implement this in different channels).
- Recommended for you.
The rich information harvested from the web allows triggering emails based on online behavior, e.g., you can send an email to user who viewed five pages of a particular product with a discount code, or offer more information on the product. You can also send triggers based on what the user didn’t do, e.g., introduce products the user has not visited on the web.
In most cases the integration itself with the engines requires your web development team to put scripts in the website and product database. That’s it. The integration should be very fast and may take only hours to accomplish.
As I said, the mountain is not that high, and the first step to having great product recommendations for your customers is really just having the courage to dive into better conversions. And remember – the only way to truly engage with your customers is to communicate with each one as an individual, which is impossible for mere mortals.
Learn how Emarsys’ smart personalization engine lets you individualize your content for each user.