Ryan Martin
Product Design
London
United Kingdom
Kitty Hawk
North Carolina
2022
Forage
Hyper-personalised cannabis recommendations
Background
The Cannabist Company (formerly Columbia Care) is cultivator and retailor of cannabis products in the US.
The were looking to launch a Cannabis recommendation tool that matched consumer contexts, personal experience and purchase history to streamline decision making through to purchase for their retail sites.
My Role
I lead the team that defined the strategy, approach and experience design of the platform. Defining the recommendation engine logic, the location and inventory matching experience - and building out a design language that used colour and iconography in the absence of consistent product photography.
Overview
With an increasing demand and legalisation initivaites in the US. The Cannabist Company (Columbia Care) were increasing its investment in the recreational cannabis market.
In a saturated market that often offers little in the way of visible differentiation between products, espcecially for new-comers, Forage was a direct to consumer platform that connected live inventory to customers through a recommendation system that mixed context, behaviour, customer experience and history — an outcome first purchase tool.
01/03
Understand the market
We started by looking at the market landscape and interviewing customers of different experience levels, use-cases and purchasing behaviours.
We also spent time with operational staff and people within dispensaries to understand where the pain-points existed with purchase flows and inventory management.
Variable product types and use-cases add extra diversity to the decision making funnel,
Inconsistency in product branding and photography make visual consistency difficult.
02/03
Define the recommendation models
We outlined our architecture and personalization framework early and how it fit into the website blueprint. This enabled us to design and build parts of our system concurrently across front-end, back-end and services layers.
We combined multiple different inputs to create a recommendation output
Mapping and associating product "effects"
We built a library of assets for products, effects and contexts
We mapped different recommendation criteria and paths into a single recommendation generator
03/03
Build the platform
We worked over a series of sprints to build out components and user-flows allowing a customer to go from discovery through to building a cart, setting up a profile and account and ordering.
Outcome
A multi-platform launch, winning industry awards.
Our platform launched online, although no longer live (forage.io) and was also available on in-store kiosks.
I can't give you exact numbers in terms of purchaes, revenue, engagement, because I don't have them, but we did win a Clio award.
On reflection, maybe we'd have iterated our way into some of the more advanced recommendation engine ideas we had; but when quiz's and recommendations are a dime a dozen, finding something one that adapts and adjusts over-time as you buy/use more — is kinda cool.