Apollo.io – AI Email Writing
Apollo.io is a comprehensive sales engagement platform that provides data-driven solutions to accelerate revenue growth and enhance customer interactions. Apollo.io’s AI in email messaging aims to optimize outreach efforts by providing personalized and contextually relevant content to maximize engagement and response rates. This project was a revamp of the MVP.
Product Designer and UX Researcher
Product Manager, Front-End Developers, Back-End Developers
Research, Wireframes, HiFi Designs, User Testing
In the fourth quarter of 2022, Apollo embarked on a significant venture known as “Generative AI.” Our objective was to develop an exceptional email writing tool specifically tailored for the sales domain, harnessing the power of our prospects’ data assets and leveraging our existing sales engagement features. By undertaking this initiative, we aimed to provide end-to-end coverage for the entire email workflow.
In lieu of relying on third-party partnerships for integration, we made a strategic decision to internally build the Email AI platform. This choice stemmed from our hypothesis and research learnings that the technological challenges involved in this endeavor were not insurmountable. Moreover, opting for an in-house solution allowed us to exercise greater control over the language model and content generation direction, aligning them precisely with our vision.
Overall, AI email writing can simplify the sales email writing process by providing time-saving assistance, personalized content, optimized language and tone, continuous improvement, and maintaining consistency and compliance. These benefits allow sales professionals to be more productive, efficient, and effective in their email communication, ultimately leading to improved sales outcomes.
Review previous research, personas and other findings.
User Flows and Wireframing.
Hi-Fi Designs and Prototypes.
Working with our real users to determine usability issues.
Iterating the design based on testing.
Hand off design to development.
QA with development.
Launch and iterate based on quantitative data.
Users and Audience
Our users included SDR’s and AE’s who use Apollo’s email sequencing tools after having already prospected their potential leads.
SDR’s are responsible for generating quality leads to pass on to AEs and engaging in cold outreach to generate interest.
What do they have in common?
Both of these personas spend copious amount of time writing personalized emails to each prospect.
Scope & constraints
This project was a redesign of the MVP version of AI. The MVP version wasn’t performing well due to a lack of user testing and was generally unscalable, with emerging AI features coming to sequences.
Still, with a short timeline of less than a quarter and changing scope, I was able to convince the company to take the time to research our ideas and fine-tune our AI.
We also lost research support on the squad, meaning I would have to act as a product designer and researcher.
Auditing the MVP
I began the process by performing a comprehensive audit of our existing AI iteration and engaging in thorough discovery work.
The initial step involved meticulously mapping out all the essential features required for seamless sequencing within the AI feature area.
However, upon evaluating the current layout, it became apparent that the slide-in panel, consisting of just three sections—a form, preview, and email body—lacked the scalability needed to accommodate our planned roadmap of features.
To address this challenge in a dynamic and forward-looking manner, I introduced an innovative solution: the incorporation of a versatile sidebar named the “AI toolbox.” Within this toolbox, all AI messaging features could be conveniently housed, empowering users to harness the power of AI or proceed with manual composition. This strategic implementation also provided ample room for future expansion, accommodating the seamless integration of numerous AI tools, fostering limitless growth prospects.
Following implementing the redesigned “AI toolbox,” I initiated a comprehensive user testing phase, conducting 60-minute moderated sessions involving 6 SDRs and AEs. Taking charge of the decision-making process and crafting the test script myself, the outcomes yielded valuable insights:
- Users overwhelmingly expressed a desire to leverage AI for personalizing email content, leading us to prioritize and integrate this feature into our roadmap within the “AI toolbox.”
- A prevailing sentiment among users was that they perceived AI as a tool to assist them in writing rather than automating the entire email composition process. This sparked discussions around the concept of inline editing with AI and the potential development of a freeform AI helper, akin to the chat GPT.
- Feedback indicated that users found the AI form to be excessively lengthy, prompting considerations on optimization. While initially bound by engineering constraints for training the AI model, I suggested future enhancements that could streamline the process by automatically populating certain fields, such as business name and offerings, while enabling users to input their pain points and integrating with our “book a meeting” feature, effectively reducing the form fields by half.
In terms of design improvements, we introduced directional buttons atop the AI panel, allowing users seamless navigation between prompts and outputs, and when the AI toolbar is closed, the email preview now conveniently occupies its space, eliminating the need for constant toggling.
My Key Learnings & Final Designs
The efficacy of AI lies in reducing the user’s workload; thus, the upcoming form iteration must be significantly more concise.
- Our objective became to integrate AI seamlessly into all aspects of email writing across the product.
- Remarkably, our AI feature garnered an astounding 10,000 signups in just one week.
- The redesign yielded exceptional results, propelling a remarkable 40.7% increase in our baseline conversions.