Now Assist Suite | UX research case study
Nancy Nguyen
UX research case studyConcept Testing
Company: ServiceNowProduct: ServiceNow Store

Now Assist Suite: Making enterprise AI updates easier to manage

Trying to update Now Assist applications can be complex due to technical version conflicts, where updating one app can accidentally break another.

This research study tests early designs with real customers to evaluate whether bundling apps into pre-validated packages is intuitive and ready for an official launch.

Desktop monitor showing the Now Assist Suite install window, with a Suite version picker and the related apps that will be updated

Role

UX Research Intern

Timeline

3 months

Skills

  • Concept testing
  • Research planning
  • Synthesis and analysis
  • Workshop facilitation

Tools

Figma
Miro
Great Question
Zoom

Results at a glance ✨

Research Impact: Simpler App Management

Research findings led to clearer documentation, early error awareness, and increased visibility of cross-app version dependencies ahead of the March 2026 launch.

Secured a seamless launch

Achieved 100% user comprehension and high perceived value, delivering validation needed to finalize the launch roadmap.

Direct Impact on Shipped Product

7 recommendations are visible in the launched product, from upfront error warnings to demo availability.

Optimized platform guidance

Collaborated with technical writers to revise documentation, aligning content with validated user insights to drive a friction-free, self-service experience.

Aligned the team on a design plan

Spearheaded a cross-functional prioritization workshop with PM, engineering, and design, aligning the team on key feature improvements that protected engineering bandwidth.

My role

UX Research Intern
Led the study end to end from planning and participant recruitment through analysis and final delivery

Nancy Nguyen introduced as an Emerging Talent UX Research Intern at the ServiceNow intern showcase
  • Created a research plan and interview guide organized around 3 core workflows.
  • Collaborated on adjusting and refining task-based prototypes to prepare for user testing.
  • Translated findings into design recommendations, ranked by user impact and split into quick wins and longer-term opportunities.
  • Maintained cross-functional alignment by delivering immediate, actionable session recaps.
  • Facilitated a prioritization workshop with the designer, PM, and engineering team to agree on what to fix and when.
  • Mocked concept prototypes based on research insights to deliver actionable design recommendations.

🌱 Why I took this on

The ServiceNow Store team planned to finalize designs by October for a December launch. They needed research to validate the concept, but no one owned the research. And that's where I came in!

Background

Thousands of customers face significant friction to keep their Now Assist apps running

Now Assist is ServiceNow's family of generative AI apps (ITSM, HR, and CSM). Most companies run several of them, with deals including three or more Now Assist products growing nearly 70% year-over-year.

The problem? Admins face a critical bottleneck where a single update can trigger platform-wide failures.

Solution

Now Assist Suite solves this issue by bundling compatible versions together into a safe, one-step installation.

Application Manager
Launched Review Installation Details screen: pick a Suite version, see each app's status, app version, and a demo data option
Suite today: pick one bundle version, and every related app updates to versions tested to work together.

Why it mattered

Suite supports three goals for ServiceNow as an AI platform company:

✨ Easier AI adoption

Customers want AI without a complicated setup. Simpler installs make it easier to start using it.

📈 Managing many apps at scale

Bundling can extend beyond Now Assist to other products with shared components.

🧩 Less complexity and cost

Fewer errors, fewer support tickets, and less time spent on updates.

Research approach

What we needed to know before launch

Problem

  • Updates are causing older apps to break
  • Users are often unaware of potential issues until it’s too late
  • There is no clear method to verify compatibility

Goal

  1. Evaluate the usability, comprehension, and perceived value of Now Assist Suite
  2. Identify pain points and opportunities to improve the experience

Core Evaluation Metrics and Results

Evaluation dimensionSuccess metricTarget benchmarkActual resultStatus
ComprehensionConceptual Comprehension RateAccurate definition of a “Suite”80%+ clarity100% unanimous clarity✨ Exceeded
DesirabilityPerceived Value AlignmentBelief that bundling solves pain points70%+ positive91% strong agreement✨ Exceeded
UsabilityTask Completion RateUnassisted click-through success80%+ completion86% task success✓ Passed

Research Phases

Validating the concept with real users

6 admins, 3 real scenarios, 2 weeks: I watched how people actually moved through the new flow, then turned what I saw into recommendations the team could act on.

  1. Phase 1PlanRecruit the right admins
  2. Phase 2Test6 prototype sessions
  3. Phase 3SynthesizeFindings and workshop
  4. FinalOutcomeLaunched in the admin tool

👆 Select a phase to see how it unfolded.

🧩 Phase 1 | Planning and recruiting

Finding the people who manage these updates

Suite was a brand-new idea, so feedback had to come from people who actually live with these updates. I recruited 3 customers who manage their own ServiceNow setup and 3 implementation partners who install apps for their clients, all with hands-on experience with these AI apps.

🔬 Research methods
  • Participant screening
  • Research plan
  • Discussion guide

🧭 The rationale

Why customers and partners

Customers live with the results of an update. Partners run updates across many clients. Together, they covered both everyday and at-scale use.

Why hands-on experience mattered

Most participants had also customized their apps, which let me test the trickiest scenario: updating an app that no longer matches its original version.

Participant overview: 3 Now Assist customers and 3 implementation partners, with job titles and whether they manage customizations
Who I talked to: 3 customers and 3 implementation partners. Company names are blurred for confidentiality.

Key insights

Before vs After

Four selected insights, each tied to a change the team made to address concerns regarding late visibility and need for additional context.

1

Indicate the Suite version associated with errors

Participants were facing version overload and couldn’t understand which version had errors.

✅ Outcome

The UI explicitly maps the selected version in the drop-down rather than listing versions for every plugin.

Before
Install Now Assist for HRSD modal listing a version beside every dependent app and plugin, with license errors
Versions listed for every plugin
After
Review Installation Details modal with a single Version drop-down highlighted above the dependencies
One version picker tied to the selection
2

Early error awareness prevents backtracking

Identifying errors late in the process can result in significant rework and delays, requiring additional time for repeated research and testing.

"I would appreciate seeing [errors] straight away so I can deal with the problems before I go through my rabbit hole."

✅ Outcome

The modal now notifies errors early and blocks users from clicking ‘Continue’ until requirements are met.

Before
Select Now Assist Suite version modal with Continue enabled and no error shown
Installation errors occur silently and users can still proceed
After
Suite version list with Installation blocked highlighted on one app and Continue disabled
Gating and a warning before proceeding
3

Trust is the ability to view impacts before committing

Participants needed firmer guarantees against unexpected changes that could impact their users.

"Understanding that compatibility is important... not having that granular control freaks me out."

✅ Outcome

Including prominent “demo data” options would alleviate a significant amount of user friction by allowing users to preview changes before committing to an update or install.

Before
Select Now Assist Suite version modal with no demo data option
No demo data
After
Review Installation Details modal with a Demo data column highlighted beside every app
A demo data option for every app
4

New workflows require up-to-date and detailed documentation

Participants came into sessions already distrusting ServiceNow documentation. Past run-ins with broken links and outdated knowledge articles left them skeptical that help content would be accurate or current.

"There should be detailed documentation: 'What does this Suite mean?' so that it doesn't confuse the admins."

✅ Outcome

ServiceNow published public setup guides for Suite. And I was invited to review the draft.

What launched
ServiceNow documentation page: Now Assist suite versions in the Application Manager
ServiceNow KB article, view here
Support article: Install Now Assist Using the Suite Model, with step-by-step instructions
Instructions to use Suite, view here

Challenges

⚠️ The challenge

Before sharing my report, I was eager to take as much advice from researchers as possible. And by trying to satisfy every piece of feedback I got, my story completely lost its focus.

Steps I took to overcome it

  • Trust the research. I had to have confidence in my findings and the story they told.
  • Sort the feedback. Not all feedback is created equal. I grouped comments into three buckets: critical, secondary, and stylistic opinion.
  • Strip it back. I wasn't afraid to cut the report down to its outline and make sure the essentials came first.
  • Prioritize the goal. Every slide had to serve what the study set out to answer.

Impact

Now Assist Suite was officially released March 2026

This represented ServiceNow's first-ever unified packaging and dependency framework for generative AI applications. By leading end-to-end concept testing with enterprise admins prior to launch, I directly de-risked the deployment experience by uncovering critical usability friction points in how complex technical dependencies were communicated. Validating these designs pre-launch provided the product team with early error awareness, revealing that unclear documentation and complex app management weakened product trust. This led to UX guardrails that contributed to faster, simpler app deployments and reduced post-launch errors.

Reflection

This project laid my foundation in UX research.

I'm grateful for the trust and autonomy my manager gave me to lead the study end to end, and for a team that always made time to answer my questions. Three lessons stuck with me:

Lesson 1

🔍 Report what you saw, not how people felt

My first draft said participants "liked" Suite. My director pushed me to state the observation and what it means instead. "All 6 could explain Suite after one try" says far more than "they liked it."

Lesson 2

🧩 Bridging research and engineering through prototyping

I translated user data directly into high-fidelity prototypes. Mocking up prototypes that visually advocated for customer needs led to greater engagement and productive discussions by directly connecting data to opportunity.

Lesson 3

📋 A readout isn't the finish line

The final readout didn't mean my job was over. Laying out next steps and scheduling a dispositioning meeting with design, product, and engineering made sure the research was heard.

"I'm really impressed [with] how you've approached this entire project... this is above and beyond."Feedback from my manager (UX Research Director)