Leading content modeling for Shop's Home feed
Responding to an urgent need to add principles and guardrails to how we shipped content on Shop’s Home feed, I developed a series of content models that continue to inform product strategy for one of the highest-visibility surfaces in the app.
The problem
When I first joined Shop, it became clear that there was no firmly-held strategy that explained why and where some content appeared on the Home feed. It was sought-after real estate that nearly every team wanted to leverage, but only one team was truly responsible for optimizing — the Explore team, where I was allocated.
As new people joined the team, we spent weeks trying to clarify the job of the Home feed, especially as it pertained to “cold start users” — a large subset of users who were engaging with our app for the first time without having any previous order tracking experience. Complicating matters was that the Shop organization was making a concerted philosophical pivot away from a sole tracking app to a shopping destination. The Home feed was deemed a major ingredient to the product’s success, and our team’s mandate was to make it useful. Trouble was, no one knew quite how to do that and, as my team worked to devise a solution, teams kept shipping new content on the feed all the time that harmed user trust.
The proposal
Everyone saw the problem of making the feed matter to the cold start user, but no one knew how to solve it. The team’s product manager spearheaded a series of successful experiments that helped us learn how users behaved on the Home feed and what content they were interacting with versus what they ignored. I joined a series of jam sessions and brainstorms intended to help us design the perfect feed using the small amounts of data we had at our disposal. Admirable as they were, those efforts didn’t yield much fruit, so I decided to tackle the issue from a different perspective.
I launched a self-initiated content hierarchy project that gradually evolved into an IA model/artifact. I theorized that the Home feed’s primary problem was a content-relevance issue and would require a content-driven solution. So, I worked to create principles for each “parking spot” on the feed and the corresponding content that would go there. My reasoning: Each piece of content needed to work hard and earn its place on the feed.
The process
Before starting my work, I studied the outcomes of a few of those aforementioned jam sessions. One of those rituals helped develop a series of Jobs To Be Done (JTBDs) that our UX leaders began to socialize across the org. I created a Figma file, took those JTBDs, and plopped them there for later use.
Because Shop was transitioning from a tracking utility to a shopping app, I set up a series of meetings with our head of merchandising to pick her brains on how she thought about merchandized content on the Home feed, which would fuel the bulk of user interaction on that surface. Initially, we paired and started creating a merchandising story arc for the feed. I quickly realized the work we were doing, while valuable, was too limited in scope for the recommendations I needed to broadly socialized. So, I pivoted, taking the content rules the merchandising manager and I co-created and transforming them into stronger content principles that would inform strategy.
With the JTBDs and merchandising content story arc that never was in my Figma and the cold start user on my mind, I began sketching a new Home feed with low-fidelity wireframes. I mapped each JTBD to an area of the feed, using UX best practices and our own data and research to determine where spatially those jobs should be performed on the feed (top, middle, bottom). I developed content types based on the content that already lived in the app (editorial, collections, categories, etc.), and I mapped them to a JTBD. I then synthesized the JTBD with a recommended content type to name a placeholder content module (“Latest from your favorite shops,” “Brands to get you started,” etc.). I then replicated the model a few more times to represent the cold start user’s journey, showing how the content can evolve, shift, and move hierarchies as time goes on.
The outcome + impact
I shared Version 1 of this model to much fanfare and interest from peers in and outside the Explore team. I presented it as an artifact that I planned to keep fresh as we learn more about the feed and hoped that teams would use it to reinforce their own thinking about what kind of content they want to push to Home.
Since then, we’ve transitioned from our laser-sharp focus on the cold start user. My work has still helped inform strategy for the Home feed, giving rise to a new set of principles and JTBDs focused on relevance, urgency, and freshness — themes I first identified in this model. The naming conventions I introduced have spurred new experiments and modules for the feed. And this work has inspired my fellow content designers and I to adapt a similar model that we can embed in Shop’s design system as a template for doing IA work.
Since then, we’ve transitioned from our laser-sharp focus on the cold start user. My work has still helped inform strategy for the Home feed, giving rise to a new set of principles and JTBDs focused on relevance, urgency, and freshness — themes I first identified in this model. The naming conventions I introduced have spurred new experiments and modules for the feed. And this work has inspired my fellow content designers and I to adapt a similar model that we can embed in Shop’s design system as a template for doing IA work.
