Designed an automated, context-aware data dictionary with AI-generated documentation (Autodoc) to dramatically reduce manual effort and improve data governance for Salesforce admins.
Company: Sonar (B2B SaaS, Salesforce Admin Management)
Year: 2023
Role: Senior UX Designer (Led UX/UI Design, Research Synthesis, Prototyping)
Duration: ~6 months (Initial Launch)
The Challenge
As Salesforce instances grow, changes by one team will often accidentally break another's processes. Customers struggle with documenting metadata — manual efforts are unsustainable, and massive spreadsheets quickly become outdated and incomplete, leading to errors and inefficiency.
The Solution: Autodoc
We developed an automated data dictionary leveraging Sonar’s dynamic metadata tracking as well as our knowledge of external application API calls. A key enhancement — Autodoc — uses AI (ChatGPT) to generate context-aware help text and descriptions with a single click.
Key Decisions & Rationale
- Understanding the Real Pain: Initial user interviews suggested some challenges around data dictionary upkeep, but having customers actually open up their spreadsheets revealed the true depth of the problem — a patchwork of out-of-date field information spread across multiple sheets. This validated the need for both automation (dynamic updates) and low-effort adoption (AI generation), addressing huge initial setup and ongoing maintenance burdens.
- Communicating AI Interaction: To build trust and clarity around the AI suggestions, we focused on micro-interactions. Skeleton Loaders indicate processing, and explicit "Accept Suggestion" actions ensure user control and provide clear feedback on the AI's contribution. High-fidelity prototypes were used to de-risk these interactions both with engineering and customers before implementation.
Outcomes
Increased Average Contract Value (ACV): Add-on with an average 40% increase in deal size.
High Customer Adoption: Achieved 55% adoption among our customer base, significantly exceeding the initial success target of 30%.
Reduced Manual Burden: Directly addressed the primary pain point of tedious, manual documentation.
Positive User Feedback: Received enthusiastic responses highlighting time savings and improved data management.
Reflection
While the AI-generated documentation significantly reduced manual effort, we learned that building trust in AI suggestions requires more than just accuracy — it demands clear user control and transparency. The success of micro-interactions like Skeleton Loaders and explicit acceptance actions reinforced the importance of guiding users through AI-powered workflows. Looking forward, deeper personalization and proactive AI recommendations could further enhance adoption and engagement, ensuring that Autodoc evolves alongside the complex needs of Salesforce admins.
"Our company is actually cutting software right now and we still pushed through the data dictionary add-on."
John at Alteryx
