2018
Pipedrive is a Customer Relationship Management (CRM) software that enables salespeople to better organise their time, so they can focus on what really matters.
Problem
Whereas the automation of workflows might be tricky to learn or even scary to practice — you are indeed going to trust a machine to send an email on your behalf — this project was focused on simplifying the creation of workflows, demistifying the bot factor and making people trust, and ultimately use, the feature more.
Increasing trust in a "bot-like" feature
Platform
Desktop
Team
1 PM, 1 EM, 1 Product Researcher, 1 Data Analyst, 7 Engineers
Role
Product Designer
My Role
I was the lead product designer on this project from ideation through to launch. I defined flows, created prototypes, and ran usability tests with both internal and external users. I drove design decisions throughout and ensured the development team was involved early — bringing their feedback into the process as a contribution, not just a handoff.
Problem
Workflow automation allows users to build sequences to perform repetitive tasks which are part of their usual workflows. It is a complex feature that, when correctly used, can help users save time and focus on closing deals. Despite its potential benefits, the fear of doing something wrong when setting up automated workflows pushed users to do one of two things: (1) visit the feature and never go back; and (2) set up an automated workflow without actually running/activating it. Our goal was thus to empower the user in building workflows, removing the friction of setting up a complicated workflow structure, while promoting trust in the automated sequence created, hence contributing to the user’s autonomy in using the feature.
Impact
With the implementation of Workflow Automation templates, feature usage increased by ~30% in the first three months, growing from roughly 5,000 to 7,000 active companies. We also identified key improvements along the way: improving discoverability, making the automation canvas more intuitive, and accounting for power users' more advanced needs.
WA Templates
As a result of the Design Sprint our solution focused on providing a template list that would target common use cases. A specific action could then be the trigger to some of the most perfomed use cases in Pipedrive:
Add an activity
Send an email
Add a deal
Move a deal to a specific stage
To help the user set up the template quickly and with the least friction, we focused on having a wizard mapping the way. By dividing mandatory data input by sections we allowed the user to focus on a section at a time, avoiding overwhelming situations. By doing this we were pushing the user forward, giving back small hints that we was doing the right thing.
We didn't do this:
Another need we were trying to cover was to start getting the user familiar with the canvas view, where the workflow structure would appear. In order to achieve this we added to the wizard a map of the workflow skeleton, as a preview. During testing we found out the additional information was actually adding an extra layer of complexity which was just confusing people.
We tried to provide the user with a different set of categories in the sidebar, specifically targetting common salesperson’s goals, instead of general actions to guide the user in discovering the most adequate template for a specific need. From Pipeline management and Lead nurturing through to Admin processes, that change only made the whole categorization unclear and proved that sticking to the simple actions was a better solution. We opted to focus on the action being taken.
Test & Iterate
Interestingly, even if users did understand that automation would help them focus on their business, save time and ultimately money, templates alone were not helping users understand how they could incorporate them into their own processes and existing workflows. This was a matter out of the scope of this project, but we considered important to provide as much as a quick introduction to it. In order to do that we added a “Get started” category, in which users would be presented with specific use cases that could inspire some of their own.
During testing we also found out:
We were struggling with feature discovery, specially considering a solid product where new features and updates struggle to get to the user’s attention.
People would like to test automations in order to be 100% sure that they wouldn’t be doing something wrong.
More proficient users found the canvas to be limiting, not really tackling their needs, so they took advantage of the filtering system to create more options targeted to their specific needs.
Users would frequently mention the need to share a workflow with team members, or manage permissions for editing and activating.
What's Next?
Even though our success metrics were reached whithin the expected timeframe, we identified some important action points to keep in mind for future reference.
Future Development
Improve discoverability in context, providing users with a prompt to the usage of automation when some specific behaviour is somehow tracked as a repetition that can be automated. In this scenario it would be more clear how automation could help the user save time in a more straightforward way.
Add an Automation course module to Pipedrive’s Academy course list, so users can be even more confident about how the feature might work and impact their personal workflows.
Improve filtering in the Pipeline view to allow users to clearly understand that adding filters to a specific view will enable more actions in the automation workflows.
Potentially add an option to run tests to a specific worfklow.
Users would benefit from a solution focused for teams, where permission management could be an option.
Looking back
The biggest thing I'd do differently is invest more time upfront in understanding power users. We optimised heavily for the trust barrier among less experienced users, which was the right call for adoption, but we underestimated how quickly power users would hit the ceiling of what templates could do. A parallel track for advanced users from the start would have produced a more complete solution.
