Strategic Product Optimizations at Yext

Yext Search historically powered structured, enterprise-grade search experiences built around links, featured snippets, and structured results. As large language models entered the mainstream, user expectations shifted toward direct, summarized answers rather than lists of results.

For businesses, this introduced a deeper challenge. Beyond generating answers, they needed to understand how responses were created, maintain control over what was communicated on their behalf, and trust the accuracy and sourcing behind AI-generated content.

At the time, enterprise generative search had few established patterns. While answer generation was technically feasible, the primary design challenge centered on trust, governance, and accountability.

Multiple concepts were explored before aligning on an end-to-end experience focused on:

• Clear source transparency
• Configurability and oversight
• Explicit ownership of AI-generated responses

Early concepts were tested with four enterprise customers across different industries to validate expectations around automation, control, and responsibility. The goal at this stage was learning and direction-setting rather than polish.

Despite being in market since 2016, Yext Reviews lacked support for responding to reviews in bulk. For customers managing hundreds or thousands of locations, responding individually across platforms like Google, Yelp, and Facebook was operationally inefficient and increasingly unscalable.

For large consumer service brands, this gap created a competitive disadvantage and became a renewal risk.

The goal was to unblock high-volume workflows quickly while maintaining clarity, control, and brand safety at enterprise scale.

The work was intentionally split into two milestones:

• Milestone 1: Manual and template-based bulk responses to immediately unblock customers
• Milestone 2: Exploration of AI-powered bulk responses, focused on efficiency without sacrificing oversight

Existing design system patterns were reused to accelerate delivery and reduce risk, as speed was critical to near-term business outcomes.

Enterprise and franchise-based customers repeatedly requested the ability to export review data via SFTP or FTP. Many relied on file-based workflows to distribute data to regional teams or franchise owners who were not onboarded to Yext.

Without a native solution, teams depended on manual spreadsheets that were slow, error-prone, and difficult to scale. While APIs existed, they were not viable for customers with fragmented internal systems. The absence of this capability became a renewal blocker for one of Yext’s largest Reviews customers.

The objective was to unblock revenue quickly while delivering a scalable, enterprise-ready solution.

A flexible export experience was designed to allow customers to:

• Export via SFTP, FTP, or email
• Scope exports to specific franchises or location groups
• Automate exports on a recurring schedule

A management view was also designed to support editing, auditing, and disabling exports over time. Existing UI patterns were reused to reduce engineering overhead and accelerate delivery.

Customers relied on the Knowledge Graph to manage business data at scale, but keeping content complete and consistent across thousands of entities required significant manual effort.

As LLM APIs became available, customers increasingly wanted to automate content generation—such as business descriptions—using data they already maintained. At the time, there was no safe, scalable way to support this within the platform.

Computed Field Values were introduced to allow customers to define logic once at the field level and automatically compute values across entities.

For example, instead of manually authoring descriptions for each location, customers could generate them programmatically using existing data—maintaining consistency while significantly reducing effort.