Where users find their starting point.

In Q4 2017, the conversions team formed to tackle some of the most complex problems facing new customer acquisition. I joined this team for the first year and a half, contributing designs to the highest converting page on Squarespace.com — the Template Store.

My Role

In this project, I led the UX, UI, and visual design for all of the about and careers pages, and worked under the guidance of my team lead. We collaborated with design leadership, executive stakeholders, the people team, content strategy, the content studio, and frontsite developers.

Problem

Business Problem: We wanted to increase landing-to-trial conversions, and felt that elevating search would give users a higher chance of success with finding the right template to start with.

User Problem: We wanted to give users more than a category list, and offer them a recommendation that felt curated and personalized.

Process

Working within the constraints of the template store, we tried to create an enhanced search that gave prominence to the search itself but didn’t take away from the templates.

The first part of the process was capturing the top 100 keywords that users searched, and plugging them in the ML model. We mapped the keywords to top recommendations, then found a way to surface the recommendations in the UX.

The resulting pattern was to give the search typeahead examples with a fixed character count, and give prominence to the top recommended templates, with a section of ‘other templates you might like’ below.

Challenges

• Having an open search can be complicated when there are only 90 templates to search through. What ends up happening is a keyword is mapped based on relevance to a template that might not match the persona of the user. The goal is to have the right messaging that lets users know that a template can be customized and is only a starting point.

Results

• We saw a 5.4% increase in subscriptions compared to the previous version of the template store.

Learnings

• From our learnings with Linear Onboarding, we found that users respond better when asked what the purpose of their site is.

• Based on our user inputs, we were able to learn more about our users through their search keywords and train our ML model for a smarter search.