Case study 2: SMS IA

SMS help center, re-architected — Bridget Paulus
Case 03 / Klaviyo · 2023

SMS help center,
re-architected

Klaviyo’s SMS docs were organized by product team, not by what readers came to do. I rebuilt the IA around six tasks. Pageviews lifted ~10% in two months. The pattern became the template for every product help center after.

Information architecture Content audit Taxonomy Help center
Role
Lead technical writer
Scope
120+ articles
Timeline
6 weeks
Outcome
+10% pageviews
New IA template adopted by 4 products
01 — Discovery

Docs structured like an org chart

The SMS documentation scaled from 12 articles to more than 120, but the growth was invisible to users. The articles were scattered by product area: Forms, Flows, Compliance. Users had trouble finding all of the information they needed.

I reviewed:

  • 4 competitors
  • 300 support tickets
  • 80 pieces of qualitative feedback
  • 120+ article titles

Before · Scattered by product area

Campaigns
SMS
Getting started with SMS
Set up SMS
Add a phone number
Send an SMS campaign
Compliance
Regional compliance
US SMS laws
EU SMS laws
Forms
Getting started with forms
Create your first form
Add SMS to your forms
Audit snapshot120 articles · 14 sub-categories · zero task-based grouping
02 — Synthesis

What users were actually trying to do

I chatted with 4 customers, 3 onboarding specialists, and 2 CSMs. I presented them with a card sorting exercise to figure out how they thought about the topics we presented.

The results were clear: they all thought about channel first, task second, product third.

Figure 02.1 · Card-sort output
SMS
Setup
Grow list
Send messages
Compliance
Settings
Forms
Keywords
Campaigns
Flows
Card-sort output9 participants · channel → task → product mental model
03 — Architecture

Six categories, action-oriented, lifecycle-ordered

The order matched how an SMS program grows: starting out, building an audience, sending messages, and measuring results.

01
Get started
First-time setup. Account, sending number, first test send.
8 articles
02
Grow your list
Sign-up forms, opt-ins, importing subscribers.
14 articles
03
Send key messages
Campaigns, flows, transactional. Drafting through schedule.
31 articles
04
Segments & profiles
Audience targeting, properties, deliverability segments.
22 articles
05
Compliance & deliverability
Registration, quiet hours, carrier filtering, dispute handling.
26 articles
06
Analyze performance
Reports, attribution, deliverability dashboards.
19 articles

After · Task-centric IA

SMS
01 · Get started
Start here · SMS in 15 minutes
Connect a sending number
Send your first test message
Troubleshoot setup
02 · Grow your list
Start here · How SMS opt-in works
Add an SMS opt-in to a sign-up form
Import existing SMS subscribers
Troubleshoot list growth
03 · Send key messages
Start here · Campaigns vs. flows
Draft & schedule a campaign
Build a welcome SMS flow
Troubleshoot deliverability
Shipped IA6 verb-led categories · “Start here · Procedures · Troubleshoot” pattern
04 — Dependencies and decisions

An IA shift isn’t a call you make purely in isolation

You need buy-in from stakeholders as well as to document the decisions being made.

Stakeholder agreement

  • Leadership to approve breaking the current pattern
  • Engineering to create new categories and organize the migration
  • Team members who needed to be trained on the new IA

Decision documentation

Decision · 01
Should SMS be a top-level category?
Yes. To increase visibility and match user mental models. Having a single category where we group all key information made it easier for both customers and internal teams to view all resources.
Decision · 02
Should “Compliance” be its own category, or split across the others?
Create a standalone section. Weave compliance information throughout articles and link to this section for more. Users mentally bucketed regulations separately, and a dedicated page improved their confidence with SMS.
Decision · 03
Task-led or product-led sections?
Task-led. While internally we thought about the product first, this was never true for customers. Sorting by task also made it easier for users to find features they couldn’t find in the app.
Decision · 04
How do we measure success?
Increase pageviews and ticket deflection. The main goal of this IA change was to increase visibility — more people landing on articles and finding them helpful enough to not create a support ticket.
05 — Results

Two months in, the numbers proved success

After launch, I watched the dashboards for sixty days. Pageviews trended up steadily, and by day 45, saw a 4% increase across SMS-related articles.

In addition, we saw a marked rise in qualitative feedback. Internal teams frequently commented that they loved having one link to send to customers, rather than five or six.

+4%
Pageviews
Across SMS help center, 60-day post-launch
+11%
Search → click-through
From searches containing “SMS”
+5%
Ticket deflection for SMS
Raised ticket deflection from 78% to 83% for SMS
6
Products use the same pattern
Reviews, WhatsApp, Customer Agent, Advanced KDP, Customer Hub, and Push notifications
06 — Reflection

What I’d do differently

Delegate more efficiently. I manually exported and organized the article titles myself when I should have brought in teammates to do this with me. This slowed me down and created another group to get buy-in from later.

Follow up more quickly. Push notifications existed at the same time, but we didn’t replicate the SMS IA for push until much later due to a lack of engineering resourcing. In the future, my project plan would include follow-ups so that work is added to the roadmap.

Share louder. While I did launch announcements in a few key internal channels, most people didn’t realize until nearly a month later. I’d work with leadership to highlight this change more broadly going forward.