Social Post Content Generation
Enabling brands to create unique, on-brand social captions at scale using AI
ROLE:
SR PRODUCT DESIGNER
COMPANY:
YEXT
TIMELINE:
JUN - AUG 2024
SCOPE:
End-to-end feature (research→launch)
OUTCOMES:
↑ Increased posting volume • ↓ Reduced time-to-publish
Overview
Yext Social helps multi-location brands manage social posting across platforms. To better differentiate the product in a competitive market, our team interviewed existing customers to understand how they currently engage with social media management software and where workflows break down.
As the sole designer on Yext Social, I partnered closely with PM and engineering to improve usability and integration with the rest of the Yext ecosystem.
Problem
Most Yext customers have hundreds, if not thousands of global franchise locations, each with their own social media accounts across various platforms. For the first phase for this project, we sought to address the most common complaint among customers:
Creating unique content for social media posts is time-consuming and expensive.
If I post the same caption for multiple locations at the same time, it can be penalized by Instagram because it looks like spam. I usually just ask ChatGPT to rewrite the same caption a bunch of different ways and copy/paste it.
— Customer
We have copywriters and brand-guidelines, but we don’t post that often because getting copy through all the internal channels takes too long.
— Customer
Based on this feedback, our team set out to answer the following:
How might we allow users to create unique social media content at scale?
Success Metrics
↑ Increase the number of customers creating social posts
↓ Reduce average time spent creating post captions
Approach
Evaluating the competitive market
With this customer feedback in mind, I set out to see what features our competitors offered, explored user flows and designs, and facilitated weekly design reviews with PM and engineering to align on scope, technical feasibility and future expansion.
Some of the feature goals top of mind were to
ensure scalability for future features and content types
consider reusability elsewhere in the platform
incorporate user feedback loops to refine content quality
Wireframing and iterating
Early explorations ranged from lightweight rewrite tools to fully automated post generation. Through stakeholder feedback and customer interviews, we learned that most users value control and transparency, and still hold mistrust in agentic automation. This insight directly shaped the final experience.
Solution
An immersive caption generation feature that balances automation and user control.
The interface allows users to guide context and content, while capturing implicit feedback to improve the quality of the output. This solution was intentionally designed to scale into new features as we seek to improve it over time.
Impact
85% acceptance rate of generated content
↑25% increase in posts created MoM
↓7 min less time to create social posts
Following the release of the Content Generation feature at the end of 2024, Yext Social has seen an increase in both generated content and in the number of posts created month over month. It is delightful to use, and cuts the average time taken to think of a post caption in half. These results directly supported our primary goals of increasing posting and reducing time-to-publish.
Reflection
This project was an early favorite because it gave me valuable insight to AI’s capabilities and how it can be applied practically to improve user workflows. It was also a solid UX & UI exercise in how B2C features accommodate consumers, different than our primarily B2B audience.
There are endless feature upgrades I’d like to see for this product, like using AI to publish at peak local times, or add neighborhood-specific lingo to engage with customers. With additional testing, we could refine the flow further, as we noticed that it was not always clear that the preview can continue being edited after its generation.
These learnings have already informed how I approach AI-assisted experience, prioritizing transparency, control and clear system feedback.