I joined Grammarly as Head of Design for Grammarly Business — the company’s enterprise product. We built a first-class design team, created a customer-centric development process for a remote and global product/design/engineering team, provided a design strategy focused on solving business and user problems, and created a fun, safe, collaborative, and inspiring environment to drive innovation. 

We grew the team from one to eight product designers, two researchers, and three content designers across three pods to drive innovation, enterprise readiness, and growth. We adopted lean, agile, and customer-centered processes that proved essential when we quickly shifted to focus on adapting to a GenAI world. Grammarly Business revenue and seat size doubled during our two-year stint, all during an economic downturn. The screens below show a slice of the great work produced by the team.

Leveraging GenAI
Like many SAAS companies in 2023, Grammarly needed to reimagine itself in a GenAI world. As market pressures required us to act quickly, we leveraged existing research to ensure our efforts were focused on solving real customer problems.

Effective AI design requires close collaboration. Product designers must ensure user needs are at the forefront of the conversation and work closely with engineers to understand AI's ever-growing capabilities.

Our initial efforts focused on helping customers become more effective communicators in the workplace. The screen below shows how a customer can make the generated text more persuasive, assertive, or confident. They can do the same with the text they generate.

However, it was clear that the greatest value and differentiation could be achieved in the space only if our LLMs could understand the workplace context. Take any random conversation in the workplace — it likely sounds like Martian to those outside the company. Without understanding this context, LLMs will only be marginally helpful within the workplace.

We began a series of hypothesis-driven experiments to understand how Grammarly LLMs can provide value above and beyond publicly available LLMs. This required a fast-paced Lean UX approach. Results from these experiments were promising and knowledge workers’ ways of working will dramatically change as products like Grammarly continue to learn to leverage context to provide assistance.

Actionable Insights
Building useful analytics requires ongoing collaboration with customers to continually hone in on the most useful insights and actions we can facilitate. While our analytics dashboard is shown below, our design strategy focused on embedding actionable insights throughout the experience to give administrators the ability to constantly take action to improve the value that Grammarly Business provides for their team members.

Enterprise Knowledge Management
Knowledge Share builds shared knowledge across an organization by helping team members decode terms like acronyms or internal project names. When team members encounter unfamiliar terms, Knowledge Share provides tooltip-like explanations that include a definition of a term, key contacts, and links to relevant documents.

We built an interface to display the tooltip explanations, a back-end workflow to allow for collaborative maintenance of the knowledge, prompts to encourage team members to create new entries as new terms were formed (e.g., new projects), and an AI component that generated an initial set of terms to mitigate the cold-start problem.

Invite Links
This feature made it easier for team members to invite others to join Grammarly Business. We saw an increase in the percentage of institutions joining by the first day (59% vs. 34%), a higher rate of institutions that expanded seats (19% vs. 15%), and a higher number of average seats provisioned (10.1 vs. 9.7).

True-up Billing
True-up billing allowed additional team members to instantly join Grammarly teams, even if doing so would extend the existing contract's parameters, allowing the organization to pay for their additional seats later. This was another project that allowed us to flex our design principles of transparency and flexibility to reduce account growth friction in ethical and useful ways.

Role-based Access Control
As larger accounts adopted Grammarly Business, we collaborated closely with customers to develop role-based authentication that balanced control, flexibility, and security. For example, companies might need different style guides for their legal and social media teams. In cases with multiple admins, only some may need the ability to provision licenses or access security features. Additionally, not all team members may require access to features like analytics. Our user-centered approach produced a solution that ensured tailored configurations that met diverse organizational needs.

Team Member Analytics
We carefully managed ethical considerations and customer feedback when allowing administrators to track individual member usage. We mitigated ethical concerns by giving members access to the same data their managers saw. The feature allowed administrators to get more value out of Grammarly through more intelligent account allocation and unlocked a $456K sales opportunity for Grammarly.

Brand Tones
Brand tones was one of many features our team created to leverage Grammarly within an organization. The feature allowed organizations to easily customize their brand voice and ensure their entire organization uses it internally and externally so employees could present a united front.

Managed Mode
Many considerations evolved around managing a product that allowed corporate and individual accounts. Managed Mode restricts sign-in to an organization's Grammarly Business subscribers. When this setting is activated, company employees can’t use their personal Grammarly accounts at work.

Request to Join
This is one of several projects focused on making it easier for new users to join existing Grammarly Business teams, which results in higher revenue and retention. We were able to funnel employees signing up for an individual account into their company's Grammarly account by recognizing their domain when signing up or visiting their account hub. Request to join increased the number of users joining a GB team within the critical period of their first seven days by 18%.  

Checkout Experiments
We continually ran experiments to improve checkout flow conversion. Some experiments included varying trial timelines, optimizing plan size selector designs, visualizing pricing displays, and changing content, layout, ordering, and product positioning through the funnel. The biggest win of these experiments was the introduction of a trial timeline that resulted in a 20% increase in checkout completion.