Client: Copilot (Xaxis / GroupM/ WPP Media)
Role: Content Strategist / Lead Writer
Focus: UX Content Design, Information Architecture, User Help Writing, Accessibility, Style Guide
Overview
Copilot is the AI platform Advertising Traders at Xaxis use to build digital ad placement strategies. Its audience spans three distinct groups — internal Xaxis traders, the managers who approve their access, and external paying clients — each with different technical fluency. The product was sharp and fast-moving; the help center wasn’t. Confusing navigation and jargon left users stuck and support teams overwhelmed.
I led the content strategy for a full rebuild:
- Developed navigation to align with and support user workflows
- Redesigned the information architecture to simplify and reduce technical jargon
- Created new user paths to support new and frequent users, managers, those who wanted to learn about AI, and those who needed to sell it to clients
- Structured new page layouts to enable scanning, skipping and quick retrieval of information
- Developed standards for font, numbering and formatting, to make instructions clear, simple and actionable
- Upgraded screenshots for clarity, contextuality and accessibility
- Developed new resources for FAQs, troubleshooting and onboarding
- Created a style guide to drive brand standards, support authors and maintain clarity and consistency
Result: A useful and intuitive, fully accessible help experience for all Copilot users and potential customers that is now used as the model for other WPP products. We also reduced the word count by 40%.
It’s no accident that you’re seeing this bus ad on your daily commute.
Artificial Intelligence helps advertisers locate and place ads in places and platforms that are relevant, impactful and profitable.
The Challenge
Copilot enhances campaigns using Artificial Intelligence. As an enhancement to digital ad campaigns, there is very little downside. The AI improves results and also helps traders manage, measure and convey performance to clients easily.
However, some users were weary about using AI to help manage their campaigns and budgets. They didn’t understand how the technology worked and configuration on the tool required somewhat advanced data input. User help needed to be simple and clear.
Most of Copilot’s help articles were well-written but not necessarily in a style that aided learning. There were several authors who had contributed during Copilot’s early days and writing styles and quality varied. Further, many articles had been generated for releases, so there was dated verbiage and a lot of “marketing speak” mixed in with user instructions.
The experience felt fragmented. It was difficult to search and find relevant help.
The Goals
We wanted to create a user experience that was intuitive, easy and as good as the product itself:
- New users should be able to find clear, step-by-step help for onboarding and campaign creation.
- Advanced users should be able to find references and resources to help them when they needed a quick answer.
- All users should be aware of the convenient reporting and analytic tools that also made Copilot convenient and a plus with clients.
- Client services professionals should be able to find information to help clients understand the technology and benefits.
- All WPP professionals should be able to understand and communicate how the AI worked to improve campaign performance.
My Process
I used UX-design methods to build a content system.
Phase 1: Discovery
- Gathered insight via surveys, forum discussions, and interviews
- Partnered with UX to map CMS opportunities and constraints
- Ran a full content audit: visitor habits, content condition, word counts/scope, publishing patterns, and opportunities to automate, consolidate, or cut
- Defined a content taxonomy splitting instruction (how to use the product) from information (how strategies and algorithms work) — so every article stayed on task
Phase 2: Future Vision
- Set standards for copy, screenshots, design, and formatting
- Built accessibility standards for screenshots, fonts, headings, and readability, grounded in W3C guidelines — with particular attention to screen readers and visually impaired users
Phase 3: Development
- Mapped user roles and journeys
- Built a full visual site map, letting stakeholders spot duplicate or vague articles, see the whole structure at once, and rearrange navigation before rewriting began
- Refined copy and visual standards; locked in accessibility requirements
Phase 4: Design
- Identified ways to automate authoring tasks and communications via ticketing, easing Copilot’s busy release schedule
- Rewrote content to be scannable, searchable, learnable, and friendly
Headline result: a 40% reduction in Copilot’s overall word count, with no loss of coverage.
Sample before/after edit illustrating the word-count reduction approach
Phase 5: Handoff
- Wrote a full style guide, adapting AP and Yahoo Style Guide conventions to Copilot’s brand voice: title-case rules, plain-language word swaps (“utilize” → “use,” “input” → “enter”), a glossary, and formatting standards for screenshots, lists, and hyperlinks
- Developed a full author guide to help authors understand all aspects of creating a good help article: language, formatting, accessibility, tips on creating good screenshots, how to number user instructions properly and
Style guide excerpt recreation
The Impact
- Fully accessible and intuitive help experience, built to W3C standards
- Visual site map and IA designed to scale with the product
- A style guide and author tools that helps the team maintain quality independently
- Process now adopted as the model for other WPP Media platforms, including Finecast and the Nexus Growth Platform (NGP)