Case Study — Content Standards and AI
Standards That Outlast the Team
Northbeam · Reconstruction · Fictional Brand
I authored the company's first-ever content design guidelines on Floor, led global standards with Storefront—content design at the same tab level as product design—and built the 94-criteria evaluator after the function was cut.
The Challenge
Associate-facing teams and consumer-facing teams were writing without a shared global content design standard. Alerts, modals, text fields, and toasts behaved differently by team. Chat bots had no single standard, which created legal risk. Engineering paid for the drift in rework.
The gap wasn't only punctuation. It was how content behaves inside the components used to build our products and where content design guidance lived in the information architecture of the design system doc site.
Two systems had to come together: Storefront (consumer) Design System and Floor Design System (associate). I owned Floor Design System content, and worked closely across the organization with consumer digital content designers through review and collaboration.
For several years before that merge I served on the Floor Design System team. Floor had product-design guidance. It did not have content design guidelines.
Then the UX organization and content design function were eliminated. A Figma page or design system with no owner doesn't scale. Teams still shipping UI needed a way to align to content standards without a content designer on the squad.
The Approach
Floor First
I authored the company's first-ever content design guidelines on the associate-facing Floor Design System: how language should behave in the components and patterns Floor already shipped.
The Merge
I led content design efforts to bring Storefront Design System and Floor Design System under one unified design system and built the company's first global content design standards. Working collaboratively across the enterprise, I aligned style, patterns, and component language across the organization. Two separate standards were killed — that was the original drift.
Parity
I elevated UX content to full parity with product design by locating content design for components and patterns at the same information-architecture tab level as product design on the design system doc site.
Enforcement After the Cut
Leave-a-PDF-and-hope was killed by the org change. After content design ceased to be a function at the company, I built the first AI-driven Evaluative Content Design Checklist to help enforce the standard. I authored 94 criteria, prompted the model, and built the tool.
Content standards reflected in the tool covered word usage and how content behaves in our components: button labels name the action (verb + noun); one action per button; title case on buttons and labels; sentence case on body, helper text, errors, and badges; alerts inform and recover; they don't alarm.
From Mechanics to the System The first work was mechanical: labels, alerts, how buttons behave. The later work was the system: one standard across Storefront and Floor, content design at the same information-architecture tab as product design, and a 94-criteria evaluator I authored and built so a model could score UI I would never see. I used the model for volume—Pass, Warning, Fail, rewrite. I kept the judgment the model cannot make: Indeterminate means the screenshot can't prove the claim: That gate is the job. The checklist scores real UI against the global standard: error messages and alerts; clear language; capitalization and punctuation; component-specific content behavior; voice and tone; accessibility and inclusion; research and testing; content business strategy.
Each item returns Pass, Warning, Fail, or Indeterminate. Indeterminate means the screenshot cannot prove the claim (alt text, research, KPIs); again, by design, the tool doesn't fake a pass. It also returns a rewrite.
"Multiple model numbers are associated to this item number. Scan item barcode to add item to cart."
→ "This item has multiple model numbers. Scan the barcode to add it to your cart."
Item Entry
→ Add Item. Same pattern: Customer Lookup → Look Up Customer.
A run on raw system-exception copy scored 3 pass / 18 warning / 57 fail / 25 indeterminate.
The Result
One design system and global content standard, with content design at the same tab level as product design, and one source of evaluative content truth in Figma for Storefront and Floor designers and product managers.
This created a workflow product teams could run before launch after content design was no longer staffed across the organization. The evaluator isn't the case: The case is that content design moved from writing strings on Floor to governing language across the system—and still ran after the function was gone because the gates were written down, and the model was not allowed to invent a pass.