The shopping list application is a tool to help users plan their shopping seamlessly across different digital channels. The design process had several iterations, the initial concept evolved as we learned more about how users plan their shopping. While the initial business focus was about pushing promotions to users, the different user research activities helped identify other opportunities for innovative features.
This first prototype was only intended to illustrate the UX Vision for the pitch, it hadn't been thought through and had many usability problems. Now the real work had to begin.
Working with Raphael Yharassarry to identify key activities, tasks and flows
We considered different effort-saving interaction design patterns for minimising the effort of entering data, such as predictive text, voice input, OCR of handwritten lists, re-using previous lists, and so on. But the most interesting opportunity was generating automated lists based on people's past purchases, applying Big Data and pattern recognition
By analysing people's purchases over time, it is easy to identify algorithmically how frequently products are purchased. Predicting people's future purchases becomes merely a matter of extrapolating the data from past behaviour and guesing when products will be purchased again. Instead of multiple entries and search operations, users could create their next shopping list in just one click.
Raphael created a few screens to illustrate the key functionalities and use cases. A focus group was conducted to assess how much the different features would appeal to customers.
The One-Click Shopping List was very well received, participants were most excited about it. They also found the meal planning function interesting, mostly those who cook regularly. The focus group had a hard time navigating the different functonalities, which was understandable, given we packed all possible features into single prototype. The purpose of this exercise was to assess the relative importance to customers of the different features.
Carrefour was also trialling a geo-location app using beacons to enable new in-store shopping functionalities, such as precise geolocation of items and mapping an optimised journey around the store based on the customer's shopping list. A high fidelity prototype was tested with users on-site but reception was lukewarm: users didn't see themselves using the functionality, saying they already knew the location of the items they buy usually. There was also frustration when items were out of stock, users resented being shown the item's location on the app only to find it was no longer in stock when they got there. Nonetheless Carrefour launched their prototype on the app store, named "C-OU", with little success. This led me to recommend this type of functionality should be a low priority for future versions.
Collating information and insights about user needs and behaviours during the different focus groups and user testing sessions, it became apparent the ability to manage multiple lists wasn't crucial, people were only concerned about their next shopping session.
If our Automagic shopping list was going to be used at all, the UX strategy was to create a dedicated app focused on a single task, and doing it extremely well. If successful, it would then replace the Drive and OOshop apps.
For the sake of simplicity, I stripped down the list of functionalities in order to focus usage on interaction with the list itself and the items suggested by the Automagic Helper.
A standalone application dedicated to a single task, solving a single problem exceptionally well. In addition the Shopping List could work together with one or more Connected Objects in the customer's living environment, to facilitate data entry. Because the Automagic Helper isn't perfect, we still had to preserve traditional input methods, but they now have a secondary role. The quickest way of adding items to a list is accepting suggestions from the Helper.
During the initial pitch we had imagined a connected device for the kitchen to help people add products to the shopping list as they run out of them. But with a Smart List that predicts automatically which items I need, the purpose of the device became redundant. So this idea was shelved in the context of the Smart List.
Although the Automagic shopping list concept was very well received by customers during focus groups, there was an unsurmountable barrier on the Carrefour side to bring the project into production: due to the way internal business units were siloed, it was not possible to convert a shopping list into an online order for home delivery or the click and collect service, rendering it useless for most customers.
In any case, at Carrefour's request a traditional shopping list functionality was added to the main Carrefour app. In particular, Carrefour was keen to introduce a feature to push certain promotions during the list editing process, they called it the "intelligent list" feature, although I remain skeptical about the real value of the feature. But above all else, user's interest for the shopping list functionality will remain low as long as they can not use it to place orders. (Customers asked repeatedly "What's the point if I can't place an order?" during user testing and focus groups.)
My original Automagic Shopping List concept based on analysing user's shopping habits to predict their future shopping lists proved its interest to customers and retains its potential, and although I am no longer involved in the project I know our research contributed to Carrefour's future strategy.
I'm sure we will be seeing similar predictive apps and intelligent agents hitting the market in the near future, along the lines of Amazon Echo and Google Now.