Project: Inclusive Language Guide *shared with permission

Context and Process

The Senior Leadership of the Environmental, Social, and Governance (ESG) vertical at Indeed requested a global reference guide to encourage employees to incorporate more inclusive language in their interactions, both internally and with clients. This language guide (shared with permission) was designed and developed for the use of all Indeed employees worldwide and was localized in French, German, Italian, and Japanese. The guide was designed in Google Slides as a desktop-readable scrolling document similar to a glossary or dictionary. While comprehensive, it functioned more as a passive reference tool that required users to either stop their work to seek out information, or to correct themselves post-communication. The intent of the guide was for "just-in-time" information retrieval, but given the spontaneous nature of communication, the need and the access to the resource were rarely aligned.

Problem

A glossary is only as effective as a user’s forethought. It takes time, effort, and the ability to anticipate in order to consult a static resource in a preventative sense. To be a truly seamless experience for business interactions, the language guide needed to be integrated into the tools employees were already using to communicate. Moreover, after launching the Inclusive Language Guide, there was no way to determine how language use was changing, if at all. We could see how many users opened the document because it was added to our LMS. But we could not determine how the guide was being used and what learning or behavior change was happening.

Therefore, I identified two core failures in the original model:

1. Workflow Disruption: Consulting a slide deck during an active Slack conversation or email thread created too much friction for the average employee.

2. Data Blindness: While LMS metrics showed views, they couldn't measure application. We had no way to verify if the guide was actually changing communication habits or influencing behavior.

Solution

I led the transition from a static document to an AI-powered writing assistant offering in-the-flow of work performance support. Partnering with Grammarly, I curated and integrated a custom style guide featuring approximately 300 flagged terms and phrases. By embedding the guide directly into the tools employees employees were using daily (Slack, Gmail, and Google Docs) we were able to instantly deploy an internal language guide that transformed a glossary searching task into a real-time learning moment. In some instances, definitional information popped up to inform the writer, and in other instances suggestions for replacement words and phrases were given in order to make the language more inclusive. When a non-inclusive term was used, the AI provided:

• Context: Pop-up definitions explaining why a term might be exclusionary.

• Redirection: One-click suggestions for more inclusive alternatives.

Results

By removing the friction for our US-based workforce, we didn't just increase usage, we changed habits. While the localized slide decks provided a necessary global baseline, the US-integrated AI pilot provided the data-driven proof of concept I needed to measure impact. Putting this resource at employees' fingertips created a frictionless learning enablement solution. Over a three-month period, our data analytics showed significant adoption across the US workforce across the following metrics: adoption, friction, and proof of concept.

• In the disability category alone, employees accepted the inclusive language suggestion 54% of the time.

• We successfully moved the needle on corporate culture without requiring any formal "training" time.

• This pilot proved that "in-the-flow" support drastically outperforms static resources, providing a clear roadmap for future global AI deployments.