Improving mobile discovery with fewer but better product suggestions
Do you remember how the internet used to be? I’m talking pre-broadband. You’d sit patiently as you called up the internet, to the sound of a screechy, static dial up tone.
Since then we’ve progressed a lot. We now have much faster, and more powerful devices in our pockets and around our wrists. But as the technology develops and grows, so do our expectations.
Another thing that has increased with the growth of the internet is the amount of choice we have as consumers.
One of my favourite design books — Neuro Web Design by Susan M. Weinschenk, Ph.D. — talks about how if we have too much choice we can freeze and end up not choosing anything at all.
“One interesting thing about choices is that we think we want a lot of them, but in actuality, a lot of choices hinder our decision-making process.”
The Netflix way
Netflix are famous for reducing the amount of decision making someone needs to do through machine learning. Anyone that has an account with them will be familiar with their carousels of recommended videos.
Netflix could just list out the genres of films and leave me to search through their huge catalogue of romcoms on my own. Instead, by giving me options related to films that I have seen, I get a head start and the decision making process is much easier.
Here I knew that The Notebook had me sobbing for hours so Safe Haven seemed like a pretty good bet (and yes, I ended up loving it). To make things even easier, they learn what I like based on how I interact with their content, so they can provide me with better suggestions over time.
Choices during shopping
I am a big fan of Topshop’s clothes. We’ve got a great store round the corner from the office which always seems to have items I love — it’s a dangerous place for me to visit for that reason! It’s one of the smaller stores and they are quite selective about what products they stock. It seems I’m in sync with the typical West End shopper because I always tend to find something I like fairly quickly!
I like visiting the bigger stores too but always find I end up spending more time trying to make sure I’ve seen all the possible options before making a choice.
The Qubit Aura era
We’ve built Qubit Aura to act like my favourite little Topshop store, sitting on top of a full online catalogue. It creates a curated list of the items that are most ‘me’ from my previous onsite experiences to help me get started, but allows access to all products on the site should I want to explore them.
Aura has always had a personalised recommendations system to generate different feeds for different people, but we wanted to make these recommendations even more relevant and so adapted our algorithms to pull in products based on specific interactions. This information is also now displayed alongside the product to show each visitor why each recommendation is relevant. This turns a list of seemingly randomly compiled items into a feed created for each specific individual.
For this process we’ve started out with 3 groups of specific recommendations:
- Based on your recently viewed items
- Popular within a category you’re browsing
- Popular across the site in general
The idea was that this would particularly improve recommendations when Aura is opened on the home and category pages, making sure you get a highly personalised experience wherever you are in the journey. And when we ran a test pitching this new recommendation system against the previous format and saw a 6% increase in product clicks!
Now we know this system improves the shopping experience, we can look at what other contextual information we could introduce and how this could help to give everyone their own specially curated store.
Sarah Benson. Product Designer at Qubit —still crying from The Notebook a little bit. Give me a bell on twitter @sarahbenson18
