Context

The QuickBooks support site houses a library of more than 5,000 customer-facing help articles, which are continuously updated as the product and regulations evolve. A team of content designers worked closely with subject matter experts to maintain accuracy and deliver new content for major product updates. Intuit prioritized AI as a tool for content generation, and the AI Digital Delivery team evolved to new ways of working during my time as a content design manager.

Problem

As user behavior shifted toward relying more on AI-generated answers for help and support, traffic to the QuickBooks support site began to trend downward. The company responded by spinning up a GEO task force to ensure that AI answers cite QuickBooks-owned content and provide accurate, up-to-date information.

In order to cover both common and long-tail queries, the team needed a process for gap analysis and content governance at a pace that would not be feasible for a team of content designers

What I did

As a Staff Content Strategist and Content Design Manager at Intuit, I implemented processes for Generative Engine Optimization and responded to AI-generated gap analyses. My role was a people manager, but I also worked closely with the team to evaluate prompts, guide content updates, and report weekly on progress and metrics.

I managed up to 15 content designers concurrently across all global regions (USA, Canada, LATAM, EMEA, APAC). Each region had variable release cycles and feature sets, as well as localized support sites.

I worked closely with our Machine Learning specialist to design a prompt for Generative Engine Optimization. The prompt scored each help article against a scorecard that included best practices for GEO, as well as standard content quality indicators such as readability and adherence to the Intuit style guide. We initially trained the team to use the prompt with Gemini, and the output included a detailed scoring grid and a draft rewrite. Once available, we translated the prompt into Claude Cowork skills.

My team also supported an AI content-generation pipeline developed by our Machine Learning specialist. This pipeline compared incoming call transcripts to our library of help articles and identified opportunities to create new or update existing content to address real customer questions. I provided guidance on prioritizing and evaluating AI-generated recommendations.

Outcomes

Content Designers used output from the GEO prompt to improve and re-publish over 400 articles in the first six months of the effort. For articles re-published, GEO scores improved by an average of 1 point on a 5-point scale, and citation rate in Google AI overviews increased from 80% to 92%.

With my guidance, the AI Digital Delivery team members gained confidence using AI tools. When I joined the team, the content designers were operating as traditional technical writers. By the time I left, they were confidently writing prompts, using Claude connectors, and finding innovative ways to accelerate their work. Despite some trepidation, even the most skeptical team members were sharing new AI use cases.