Why You Need Better Product Recommendations
Have you ever been to a restaurant and had trouble deciding what to order?
The server comes over and asks if you’re ready, and, on a whim, you ask them, “What do you prefer, the holubtsi or the deruni?” (You’re at a Ukrainian cafe.)
When you do this, what you’re really doing is asking for is a product recommendation.
And the response—whether you’re scratching that pierogi itch or shopping online—is more powerful than you may think.
What are product recommendations?
We kind of just went over this, in a slightly weird way, but here’s a more straightforward definition:
Product recommendations are part of an ecommerce personalization strategy wherein products are dynamically populated to a user on a webpage, app, or email based on data such as customer attributes, browsing behavior, or situational context—providing a personalized shopping experience.
They’re the targeted suggestions you get that make you stop and go, “Huh! That does look like something I’d want!”
And ya know what? They work.
Product recommendations really work
Invesp, that oracle of conversion optimization, ran a survey on product recommendations.
Some eye-opening stats:
- 49% of consumers say that, after receiving a personalized recommendation, they have purchased a product that they did not initially intend to buy
- 54% of retailers reported product recommendations as the key driver of the average order value in the customer purchase
- 75% of customers are more likely to buy based on personalized recommendations
Think about those numbers.
Half of all the people Invesp spoke to revealed that product recommendations led them to buy something they weren’t going to. That’s the definition of increased order value.
In fact, product recommendations are so good at enticing people to buy more that almost 6 in 10 ecommerce operators cited them as the main reason for people putting more in their carts.
The last stat shouldn’t come as too much of a surprise, since we’re all consumers ourselves; who among us hasn’t seen that special little recommendation made just for us and been convinced to click and buy?
And one last one before you go: it’s estimated that personalized product recommendations are responsible for fully 35% of purchases on Amazon.
Now, none of us are exactly Amazon, but who couldn’t use a boost like that to their monthly sales?
Product recommendations really work.
Some common product recommendations
There are a few common ways that product recommendations show up on our screens.
Frequently bought together
This one is a classic.
You’re innocently browsing items and one catches your eye, so you click on it. Lo and behold, somewhere below or beside the product’s image and info is a list of other items that, apparently, are frequently bought together with the one you’re considering.
Best sellers and top rated
These types of product recommendations tap into our lizard brain and appeal to our need for acceptance.
If everybody else is buying this product, if everybody else loves it, surely I should get it too.
And hey, who doesn’t want the best?
For the most part, the products that get “featured” are the ones that retailers are keen to offload.
And unless it’s at a reduced price, what does “featured” really mean, anyway? What good is it to me?
These types of product recommendations are all well and good, but they each have a major flaw: there’s nothing personal about them.
So what if Mary from Arkansas also bought that scarf when she bought this jacket? I’m not Mary.
Where these types of product recommendations fall short is this: they are a one-size-fits-all approach.
In some cases they are static value judgements, and in other cases they’re simply too broad (rd: not personalized enough) to be effective.
Delivering effective product recommendations comes down to data.
Using AI to drive product recommendations
We’re going to be honest with you.
If you don’t collect meaningful data about your visitors/customers, your product recommendations will be generic, subjective, and less powerful than they can and should be.
Sure, the types we laid out above are okay—they have their uses—but don’t get them confused with robust, true data-driven product recommendations.
It would be like playing tennis in Birkenstocks. There’s nothing wrong with Birkenstocks, in fact they’re quite comfortable, but you can do better.
You’ve probably heard of the three main types of commonly collected data:
- Aggregated data
- Category/product views, adding to cart and purchase data, internal search queries
- User-specific data
- Which categories and products the user viewed or bought from
- Static product data
- Information statically pulled from product feed
These types of data are important to the running of any ecommerce site, but they’re also superficial. They leave the searching and information submission strictly up to the visitor, who may or may not know your product offerings. What you need to be able to do is talk to your customers and understand their problems and needs, and then provide them with hyper-personalized product recommendations that address those needs.
It comes down to personalization.
Picture yourself on a beach an ecommerce site. You’re browsing, whistling while you shop, and every once in a while some products pop up with little headers saying “Try this one” or “Top rated”.
Hmm, I guess.
Now imagine, instead of generic prompts that may or may not apply to you, you get a nudge with a message that says, “Hey, I noticed you’ve been looking at tennis shoes. I don’t think Birks are the right choice. You mind if I ask you a few questions to help you find the right pair for you?”
Because you’re a kind and curious person, you say yes.
The messenger then begins to ask you things like how often you play, and if you find you need arch support when you do something physical.
At the end of the short conversation, you get a product recommendation tailored exactly to you and your needs.
Boy, that feels nice.
Personalization, as we outlined in a recent post, truly does make a difference. And personalized product recommendations are a huge part of that, and could be a huge part of your ecommerce success.
The ball’s in your court.
What’s in it for me?
As we said in that personalization post, 56% of customers are more likely to return to a site that recommends personalized products.
Repeat customers? That’s a good thing.
But the benefits don’t end there.
Accurate, data-driven product recommendations do a lot of good for your business:
- They improve conversion rates
- They boost loyalty and spur repeat purchases
- They increase average order value and the average number of items per order
Furthermore, by collecting personal information in real time, your site is able to provide customers with product recommendations in the moment. This isn’t something that takes time and then depends on the visitor returning when the numbers have been crunched.
This is immediate conversion and long-lasting loyalty.
When your site can offer suggestions that resonate with a visitor, you build a connection with that person. They feel listened to. And there’s huge power in that.
Take amika for example.
The Brooklyn-based haircare company launched an AI data-capturing program with Automat, and after three months, their customers were raving about the accuracy of the product recommendations. And they showed their appreciation with their wallets.
Still not convinced about product recommendations?
Then maybe a quick chat with our team of AI experts will convince you. Reach out to learn exactly how Automat can implement the systems you need to become a master of product recommendations and a pillar in the eCommerce community.