The challenge
Thumbtack’s pro profile pages often contained inconsistent, keyword-heavy, and difficult-to-scan content that made it harder for customers to quickly evaluate professionals and make confident hiring decisions.
The business also needed content systems that could:
improve customer understanding and trust,
strengthen SEO relevance and on-page quality,
support conversion and revenue goals,
and scale consistently across hundreds of thousands of service categories and provider profiles.
As the scope of the problem grew, AI was identified as a potential tool to help scale high-quality content generation more efficiently, and this initiative also served as an early exploration into how Thumbtack could responsibly operationalize AI-generated marketplace content at scale.
Baseline experience
Process
To address these challenges, we explored whether AI could help create clearer, more scalable marketplace content while maintaining trust, accuracy, and brand quality.
The initiative focused on redesigning how provider information appeared across search results and profile experiences. We introduced concise summaries that helped customers quickly understand a provider’s services, experience, and credibility.
My work focused on shaping the language systems behind the experience, including:
refining prompt frameworks,
defining tone and readability standards,
creating evaluation criteria and training examples,
and establishing guardrails for trust, SEO, and legal compliance.
Because the system needed to scale across hundreds of thousands of profiles and service categories, we also developed reusable frameworks for factual grounding, QA workflows, and content consistency.
The project was highly experimentation-driven, with A/B tests used to measure the impact of clearer, more structured content on customer engagement, conversion behavior, and revenue outcomes.
The new experience
Impact
The project demonstrated that clearer, more structured marketplace content could improve both customer experience and business outcomes.
A/B testing on service and provider pages showed a marked increase in revenue and in hires, and follow up analysis showed especially strong performance among high-intent, non-direct traffic segments, where improved provider information helped customers evaluate pros more efficiently and move closer to conversion.
Beyond the experiment results, the project helped establish broader standards around prompting, compliance, factual grounding, and review workflows for future AI-assisted marketplace experiences.
Through this work, I strengthened my experience in:
content systems and UX writing,
prompt engineering and applied AI,
experimentation and metric analysis,
trust and compliance design,
scalable content operations,
and cross-functional collaboration across product, legal, SEO, and content teams.