Meta: Category Naming & Guidance
The Goal
To use instructional prompts to guide advertisers through the campaign creation process.
Objective
My key objective was to help advertisers get better results on Meta Ads.
I could do this by designing a clearer suggestion of what they might try. I could take what we knew about them from the back end, or maybe other businesses like theirs, and translate that into valuable tips and guidance.
And those tips and guidance would hopefully help them spend less and make more.
Team
These are the good folks I collaborated with throughout the process:
🐴 Brad - Product Designer
🐸 Vivek - Engineer
🐹 Varun - Engineer
🐲 Gaurav - Engineer
🐯 Sharon - Lead Engineer
🐻 Suz- Product Manager
🐼🐭 Charles, Mary - Data Scientists
🐰 Fred - UX Researcher
PROCESS
4 months (Jan. 2023–April 2023)
2 rounds of lightweight user testing w/ UX Researcher
Design team collaboration in Figma and content docs
Legal and Privacy reviews
Multiple content quality reviews with feedback
OPPORTUNITIES
A/B test clearer, simpler content
Show that advertisers prefer simple language
Build a story for scaling key results and learnings
Make machine learning in Meta Ads work better for advertisers
Clean things up and make them look better
TEAM
Core Team: Content Designer, Front-end Engineer, Back-end Engineer, Product Design Lead, Product Manager
Stakeholders: Engineering Lead, PD, CD Lead
KEY RESULTS
Positive Monetization leadership reaction
Advertiser reaction (<Help tickets, +Sales feedback, +UXR)
Adoption (lift, no negative trend)
Decrease load time
Process
1 — Questions, Relationship Building, and Alignment 🤝
Get to know engineers
Ask clarifying technical questions
Get to know UX researcher
Ask clarifying UXR questions
2 — Audit ✅
Content audit of Ads Manager table
Deep dive into rabbit hole of legacy features
Sharing crucial learnings with product manager
3 — Strategize🤔
Write hypotheses
Re-write hypothesis based on XFN feedback
Clarify hypothesis with Design and Eng
Estimate content load
Push back on content design timeline
4 — Develop Content and User Experience ✍
Lead name exploration
Use info from audit to inform naming exploration
Partner with product design to explore color treatments and content patterns
Write 60+ strings, CTAs, helper text, and headlines
Focus on accessibility
Review with Legal and Privacy
5 — Localize, Launch, and Test 🚀
Final Legal check
Final accessibility check
Deal with any last minute surprises
Consult with Localization team to ensure appropriate translation into 49 different languages
Dogfood and de-bug
Experiments
There were two. 🧪🧪
First, we tested simplified language. We wanted to see if sounding more human made any impact on conversion.
I got this added to the roadmap.
Second, we tested a new column in Ads Manager Table.
The column was called Recommendations and it was going to have a few of our most common “guidances” for advertisers.
I named this and developed content patterns and guidelines.
Experiment #1 🧪
I pulled together a HUGE A/B Content Test to see whether simplified language saw better results.
I defined “better results” in a lot of ways. Here are a few learnings that would have made my content designer heart happy:
⬆ click through rate (potential revenue)
⬇ advertiser confusion in user experience interviews
⬇ Sales team confusion and complaints
⬇ Help Article click through rate
Cleaner, simpler UI design
Better readability (8th grade, decrease word count by 30%)
Removal of unimportant jargon
Clearer CTA buttons
Addition of alt text and basic accessibility
Support of established content designers in Meta Ads
Here’s a peek at what that content looked like
Communication & Content Organization
Because this test needed 60+ strings with only slight variations, I worked super closely with my engineer to make sure everything was good to go.
Clarity: Stronger headlines
Instead of sharing the error, express the opportunity.
Old: Action + Confusing problem statement
New: Verb + Opportunity + Action to take
1 — More human: Add filler words
Adding in words like your, the, and so makes the rec sound less robotic.
2 — Empathy/Trust building: Acknowledgment
Adding a second sentence acknowledging that advertisers may have set up their audience like this on purpose expresses a vote of confidence and may increase advertiser trust.
3 — Readability
The control is a giant block of text. So I cut it down by ⅓.
sentiment_dissatisfied
Before
rocket_launch
After
And here were the key results:
⬆ click through rate
⬇ advertiser confusion in user experience interviews
⬇ Sales team confusion and complaints
⬇ Help article click through rate
Cleaner, simpler UI design
Better readability (8th grade, decrease word count by 30%)
Removal of unimportant jargon
Clearer CTA buttons
Addition of alt text and basic accessibility
Support of established content designers in Meta Ads
Experiment #2 🧪
In the second experiment 🧪, I did a content audit of Ads Manager table, named a new column, and developed content patterns and guidelines for it.
I defined “better results” in a lot of ways. Here are a few learnings that would have made my content designer heart happy:
⬆ statistically significant click through rate
⬆ advertiser satisfaction in user experience interviews
⬇ Loading times
= No increase in confusion with the added column
= Nothing breaking
Clear, solid name
Design team feeling satisfied
Completing a launch of something advertisers’ had requested for 2 years
Clear guidelines of column
Approval and support of Guidance team, especially the Data Scientists, UXR, and PMs who pushed for it
And here were the key results:
⬇ Loading times
= No increase in confusion with the added column
= Nothing breaking
Coulda been better 🤷
Clear, solid name
Design team feeling satisfied
Completing a launch of something advertisers’ had requested for 2 years
Clear guidelines of column
Support of Guidance team, especially the Data Scientists, UXR, and PMs who pushed for it