Product Hunt is a community-driven platform that helps people build, share, and discover the best tech products.
Our primary taxonomy was a flat hierarchy that encompassed everything from "iOS" to "DJ Khaled" or "calendar apps". I evaluated our existing "topics" system and designed a solution and rollout plan to move products to a fixed, hierarchical, category system that would improve our data structure as well as SEO performance.
We need to build a stronger foundation for richer, more accurate data sets and a web of connections between products
I manually categorized over 5,000 popular products to prevent a cold-start problem.
Over 6,000 makers added categories to their products in the first few weeks of the feature being live. We were able to begin using this more detailed taxonomy to improve other aspects of the site.
Our product ranking algorithm wasn't good enough; this led to confusing default ordering on some category pages.
Makers began to use categories that didn't quite fit in order to increase their visibility. We investigated category assignment using AI, and I prototyped several prompts to automatically categorize products based on our internal guidelines.
Over time, the launch process on Product Hunt grew more complex and confusing. New features were jammed into a flow and architecture that wasn't originally designed to accomodate them. The process suffered from unclear expectations, a manufactured reliance on Hunters, lack of key features, and confusion around scheduling. I leaned heavily on the expertise of our support team and interviews with makers in order to re-design this key experience.
I did an audit of all the fields we wanted and needed for launches and re-organized this information into steps that are more in line with how makers think about product launches.
I worked with our support and content teams to write content that would address common questions. These questions would adapt based on whether the maker was posting for the first time, or returning after a previous launch.
I collaborated with engineers to develop an approach to autosaving and draft content. Previously, there was no way to save an in-progress post – meaning that makers needed to have all their info ready to go at once.
I worked on UX microcopy and introduced a new 'Launch Checklist' step that would allow makers to review both the required and suggested fields.
We were able to validate the success of these changes with reduced support requests related to posting, a decrease in manual moderation actions and a significant increase in the percentage of products including our "suggested" information
I prototyped a new scheduling experience that focused on minimizing errors, providing contextual explanations, and improving maker launches. I spoke to makers who had recently filed support requests, as well as our product operations team, who fields these issues regularly.
Switch the primary action to schedule -- this relieves stress from our product operations team, and ultimately results in improved outcomes for makers.
Provide clear warnings when makers attempt to launch outside of suggested launch windows. Force an acknowledgement of the action and provide an alternative.
As we added new features and content types, the need arose to revisit the design of our search experience. We wanted to improve the presentation of our search results, and also the speed at which visitors could reach product pages.
We tailored our sorting algorithm to prioritize string matches on product names in the modal (for quick access to direct searches), and prioritize topic/category matches in the search results screen (Ex. calendar apps, notion templates, no-code, etc).
We saw improved click-thru rates in both the command-k modal and the search results screen, leading to increased page views and traffic to products.
Product Hubs are the overarching pages that include all launches from a particular product. After the initial launch of these pages, we watched and evaluated usage to determine what needed to change from v1. In particular, we wanted to streamline and focus these pages on mores specific KPIs.
We wanted to focus on driving traffic to maker's websites and App Store links. This allowed us to focus our design decisions on these key goals.
The existing design was overflowing with information – much of it duplicated in two places.
With clear goals, we were able to completely remove extraneous information or make decisions about de-emphasizing various UI elements.
Redesigned Product Hubs resulted in a 32% increase in CTR to maker websites and a 21% increase in pages per session.
The de-empahsization of alternatives decreased pass-through page views. We addressed this through additional updates and were able to recover to previous benchmarks.