MAILCHIMP, CONTACT IMPORT
How might we help
new users send
their first email?
Every Mailchimp user is required to import their contacts if they want to send an email. The existing Contact Import flow suffered from a notably low task completion rate. It was contained in a confusing and inaccessible wizard which resulted in a substantial drop-off rate among users, particularly at the complex field matching step.
The first feature to ship as part of the company-wide app redesign, my team identified that the crucial task of importing contacts was a significant pain point causing activation drop-off.
Retaining the 14% of users who churned due to the flow’s usability issues represented $10 million in ARP.
Notably, 75% of these users never figured out how to actually use the platform for sending emails after creating an account.
Before
Users could not figure out how to move forward or go back in the wizard, they didn’t totally understand what a CSV was or what types of data could actually be imported and recognized. These complexities led to the high user drop-off rate.
CORE INSIGHT
Active accounts that never import exist, but importers have three times the average revenue per user than non-importers.
Strategic Approach
Facilitated internal cross-functional workshops
Partnered with testing vendor Human Interfaces to conduct competitor benchmarking study
Conducted in-person customer visits to interview frequent users
Used RITE method with low-fidelity prototypes for quick insights during customer visits
Released iteratively to GA
Challenges & Opportunities
I initiated a series of internal workshops with my team to identify areas of improvement based on the current flow, which allowed us to establish our initial requirements.
We also reviewed existing analytics, customer feedback and complaints, taking into account various user journeys and edge cases within the flow.
These sessions were instrumental in shaping the initial problem statements, sketches, and prototypes that we later brought to our customer visits.
We had various issues to address, and the outcome of our benchmarking study pinpointed these specific focus areas:
Improving the tagging feature
Streamlining the process of matching contact fields to Mailchimp's system backend
Facilitating the creation of new contact fields
Improving the usability and legibility of the 'wizard' UI
Transitioning from our legacy framework, Dojo, to React, and aligning with current WCAG accessibility guidelines
These areas were chosen because it was identified during the study:
Only 45% of participants were successful in matching fields.
Only 64% of participants could create a new field.
Only 29% of participants managed to tag an import.
Testing & Validation
Following the benchmarking study and workshops, we did 11 in-person customer interviews while visiting their stores, offices, and factories. During these visits, I used Rapid Iterative Testing methods via low-fidelity prototypes (see below) to gather quick insights.
During our initial visit, we had the pleasure of meeting Leslie, the owner of a jewelry shop called Aro. Leslie's method of gathering subscribers involved a notebook placed on the shop's counter, where she collected email addresses. However, this approach often resulted in illegible or misspelled manual entries.
Insights & Feedback
Of our 11 customers, those who used the import tool frequently found the process to be cumbersome and the value of importing more than the bare minimum information to the CMS was unclear to them.
Conversely, those who infrequently imported experienced difficulty with the field matching step.
We also met with customers who imported primarily from physical sources by adding contacts one at a time or creating basic spreadsheets each time they needed to upload.
When we started this effort, only 74% of new users that started an import successfully completed it. After we shipped the new version, we saw 87% of new users that start an import successfully complete it.
Overall, 92% of all users that started an import successfully completed it, which was an increase of 18%! Because of the project’s success, I partnered with the mobile app team to deliver a mobile version of this flow that also showed a huge uptick in mobile list imports.
Outcomes
Design System
My contributions to the Wink design system were the Combobox component and a modular Wizard pattern.
In addition, my team delivered the new React code for the Wizard, the shell of the application and the navigation.