AT&T Consumer Self-Service · Information Architecture · Content Strategy AT&T

Redesigning online self-service at AT&T scale.
Turning customer language into a clearer path to support answers.

AT&T's support content existed, but customers struggled to find the answers they needed because the taxonomy reflected internal product structure more than customer language. As research lead and design strategist, I led discovery, taxonomy redesign, and content strategy for AT&T's primary consumer support destination — helping improve CSAT by 20%, reduce support calls by 5–12%, and increase monthly visitors by 5M.

Outcomes

Less calling. More solving.

Measurable improvement across every metric that matters for self-service — satisfaction, call deflection, traffic, and time-to-answer.

Customer Satisfaction
+20%

ForeSee CSAT improvement post-launch.

Call Deflection
5–12%

Fewer inbound support calls as customers resolved issues online.

Monthly Visitors
+5M

Increase in unique monthly visitors following launch.

Time to Answer
−10–15s

Reduced dwell on upper-level pages — customers finding answers faster.

The Opportunity

The content existed. The opportunity was making it easier to find.

AT&T's support content was organized around internal business structure — product lines, teams, legacy categories. Customers arrived with a problem and used their own language to search for answers. The two didn't match.

When customers couldn't self-serve, they called, chatted, or emailed instead — adding unnecessary load to care channels while digital support sat underutilized.

The Core Insight

The core opportunity was a language mismatch: AT&T organized content around internal structure, while customers searched in their own words. Rebuilding the taxonomy around customer language was the highest-leverage change we could make.

What we set out to change

  • Rebuild taxonomy around customer vocabulary
  • Surface the right content at the right moment
  • Reduce steps between question and answer
  • Design pages for scanning, not deep reading

The solution in brief

  • New taxonomy built from customer language
  • Content optimized for search and scanning
  • Redesigned page structure with clear visual hierarchy
  • Knowledge management system (KMS) to maintain quality over time
Research & Discovery

Four methods. Every one pointed to the same root cause.

We ran parallel research streams to triangulate the problem. The finding was consistent across all of them: the language gap was the issue — not missing content, not broken functionality.

What we did

  • Customer Interviews — heard customers describe problems in their own words, surfacing vocabulary AT&T's labels didn't match
  • Treejack Study — tested findability of key content in the existing IA without visual design as a cue
  • Content Audit — mapped duplicates, gaps, and mislabeled categories across thousands of articles
  • Navigational Path Analysis — showed where customers actually went versus where the IA assumed they would

What we found

  • Customer vocabulary didn't match AT&T's category labels
  • Technical jargon and internal terminology created friction at the first step
  • Significant duplicate and overlapping content across categories
  • Unrelated topics grouped together, related topics split apart
  • The content existed — customers just couldn't navigate to it
Click analysis showing navigation patterns vs. expected paths

Click analysis — the gap between where customers navigated and where AT&T expected them to go. The misalignment was structural, not random.

Design

Every decision traced back to one principle: make the answer findable.

The redesign wasn't about aesthetics. It was about removing friction between question and answer — at every point in the customer journey, from search to landing page to article.

Design Decision

We rebuilt the taxonomy starting from customer vocabulary, then mapped back to AT&T's content structure — not the other way around. This meant some internal category names disappeared entirely, replaced by the words customers actually used.

Strategic Decision

We could have continued improving individual articles and page templates. Instead, we addressed the deeper structural issue by rebuilding the taxonomy around customer vocabulary — a higher-effort path that required alignment across content, SEO, and product, but created more durable improvements than patching content one page at a time.

Content strategy

  • Optimized for customer search terms, not internal terminology
  • Reformatted articles for scanning — short paragraphs, headers, bullets
  • Collapsed duplicate content, clarified overlapping categories
  • KMS introduced to maintain content quality over time

Page structure

  • Clear visual hierarchy — most common tasks surfaced first
  • Defined page areas to reduce cognitive load
  • Removed competing elements that distracted from task completion
  • Made interactive elements obviously and consistently clickable
Content structure optimized for scanning and SEO

Content structure — reformatted for scanning with clear hierarchy, short paragraphs, and the terms customers actually use when searching.

Supporting design artifacts

Final comps and redlines

Final comps and redlines — handed to engineering with complete specifications for the redesigned experience.

Validation

Task success rates nearly doubled. Twice.

We tested iteratively — not just at launch. Each round confirmed the taxonomy changes were working and revealed where further refinement was needed before the next build cycle.

Testing methods

  • Eye tracking — confirmed customers found primary CTAs without scanning full pages
  • Taxonomy review — validated that new labels matched customer vocabulary
  • Task success rate testing — measured findability before and after in two rounds
  • Scroll map analysis — showed engagement patterns with restructured content

Results

  • Test 1: Task success 36% → 46% (+10 points)
  • Test 2: Task success 29% → 56% (+27 points)
  • Eye tracking confirmed CTA findability without full-page scanning
  • Scroll maps showed meaningful engagement with restructured sections
Why This Mattered

At AT&T's scale, a 10–27 point improvement in task success translates directly into fewer support calls, lower cost-to-serve, and measurably better customer experience across millions of monthly sessions.

Taxonomy validation study showing category match rates

Taxonomy validation — confirming new category labels matched how customers actually searched. A critical step before committing to the full content migration.

Supporting validation artifacts

Before and after — att.com support redesign

Before & After — the redesigned AT&T support experience.

What This Changed

The root issue wasn't missing content or broken technology. It was a structural mismatch between how AT&T organized its support and how customers thought about their problems.

Rebuilding the taxonomy around customer language — and designing pages for scanning instead of reading — delivered: +20% CSAT · 5–12% fewer calls · +5M monthly visitors. At this scale, small findability gains have outsized operational impact.