Navigation & IA

Jimdo, 2021

Jimdo’s core product is a website builder that aims to be the easiest in the market. Product strategy was increasingly shifting to a connected multi-product experience, while we heard from users that finding what lived where in individual products was challenging.

Years of iterative MVP experiments had led to irreconcilable journeys and mixed metaphors in different products.

Challenge: How to approach a fractured cross-product journey when cross-product is becoming company strategy?

Response: Build a discoverable, extensible IA and a navigation to fit current and future needs.

Background

Over the course of dozens of user interviews across various features (from copy-pasting page elements, to creating onboarding guidance for new users) we were beginning to hear a pattern of feedback - that our navigation was a barrier to users creating their websites effectively.

Since the core product launched 5 years ago, several add-ons, sub-products and features had been added, but often as MVPs without thought to broader product journeys. In each case the process of using the product was simple and effective - but finding out it existed at all, or remembering where it was the next time you logged in, was a challenge.

Discovery

To establish if this was accurate we combed through previous usability studies and surveys, ran new rounds of moderated user testing, and went through Help Centre usage data - as well as speaking to our Customer Support staff and creating custom FullStory filters to watch user navigate directly.

From all these data points we established that something like 1 in 10 support tickets was a direct result of our navigation and information architecture - people struggling to find what features were available or how to find them.

To try to gain momentum for the project and justify some resource allocation, we challenged Jimdo staff and management to complete the same journeys we were expecting users to complete

We did this as a tree test - that is, away from their familiar visual navigation, so muscle-memory and simple recall can’t do the work for them.

For some core cross-product journeys that we intended to result in upselling to more expensive packages, only 4% of Jimdo staff and management reached the goal.

Testing

Establishing that IA is a problem area is the easy part. Establishing what should replace it and how to improve it in a way that was extensible and scalable is trickier.

We took a content inventory of each in the product journey, and went through a workshop with internal stakeholders from content design, product management, and marketing, in order to find better terminology for complex features and establish a clearer hierarchy for information in each product.

With that complete we launched a series of tree tests to check against our Jimdo employee baseline - in almost every case we saw significant improvements.

To accomodate an improved information architecture and to ensure cross-product navigation was possible, we also worked on developing new navigation concepts and interactions.

We did two rounds of unmoderated testing to establish whether one model was preferable to others, and a round of moderated interviews to discover where we’d made errors in our reasoning.

We also made sure to document both the decision processes that led to these navigations and the ways by which they can be extended, amended and improved in future in our internal documentation.

Production-ready proof

Parallel to moderated and unmoderated testing was an experiment to determine the technical feasibility of sharing a navigation component between teams - one more complex than the functional components that comprised the other elements of the design system.