Spot Shadow IT
in 3 Clicks.

Not a UI problem. A retention problem.

Scope: End-to-end design of a new product line. Discovery research, IA, interaction design, visual design, design system integration, and design QA. Shared PM across squads. I co-owned discovery and problem framing.

"The infrastructure existed. What didn't exist was a way to act on it. We didn't have a UI problem. We had a product that created anxiety without resolution."

Discovery phase · core reframe
Under 3
Clicks to first insight
−10%
Security risk reduction
50+
Enterprise clients
Context

A strategic shift,
not just a new feature.

Sastrify was built for procurement. As we moved upmarket, IT leads became the real decision-makers, and they cared about risk, not spend. We could already see unsanctioned tool usage in our IDP data.

Two signals pushed this forward: ISO compliance standards mandate shadow IT awareness in enterprise orgs, and the competitive landscape had no clear winner. The market was waiting for a tool that earned trust instead of demanding it.

How we worked

Cross-functional squad.
Shared PM.
Design-led discovery.

Radar was built by a cross-functional squad: one senior product designer (me), a PM shared across multiple squads, and three engineers. The shared PM meant I carried more ownership over discovery and problem framing than a typical designer-PM split.

I co-ran 5+ interviews with IT leads and procurement managers. I wrote the interview scripts, the PM handled scheduling from the account pipeline, and we conducted sessions together. I synthesized findings into the four problem statements that framed the entire project.

The biggest cross-functional moment: the sales team pushed for two separate dashboards, one for IT, one for procurement. I built a prototype showing both workflows served by the same data with different entry points, presented it to the Head of Product, and got alignment in one session. Sales adjusted their demo flow instead of us building two products.

The Problem

The data existed.
Nobody trusted it.

Four problems hiding behind one dashboard.

01

"I don't know if this is real."

Signals were incomplete but the UI pretended otherwise. Confidence was zero.

02

"OK it's flagged, now what?"

No clear next step after seeing a risk. Anxiety without agency.

03

Two personas, one screen.

IT leads and procurement managers shared the same view. Neither felt it was theirs.

04

No reason to come back.

Useful once for a quick scan. Nothing created a weekly habit.

The Insight

We changed the question.

Most teams would have redesigned the dashboard. We built three
activation questions that every design decision was tested against.

01

Does it activate?

Can a new user find their highest-risk tool in under 3 clicks?

02

Does it retain?

Is there a reason to open this every single week?

03

Does it build trust?

Will users act on what they see, even when signals aren't complete?

The Solutions

Three decisions.
Each one counterintuitive.

02

Split by workflow,
not by feature.

IT and procurement don't need different data. They need different paths through the same data. Task-flow structure let one product serve both.

03

Build the unglamorous
feature first.

The Audit Trail was last on the roadmap. But compliance reporting, done manually in spreadsheets, was the weekly job. It became the most-used feature.

How I got there

From interviews to
iteration to handoff.

5+ interviews with IT leads. Trust × agency matrix.
Three fidelity stages. One confidence indicator
that took three versions to get right.

01

Discovery & Research

Week 1–2

5+ interviews, competitive audit of 3–4 tools. Mapped findings into a trust × agency matrix.

02

Explore & Prototype

Week 3–5

Whiteboard sketches → mid-fi Figma prototypes → tested with original participants.

03

Build & Ship

Week 6–8

High-fidelity production screens. Design system integration. Design QA with engineering.

The Results

Proven impact.

Weekly
Audit Trail use

Highest return rate in Mixpanel — the unrequested feature became the habit

−10%
Risk reduction

reported by clients post-launch

<3 clicks
To first insight

activation benchmark confirmed

What I learned

Key takeaways.

Transparency beats polish. Users didn't need perfect data, they needed to know how confident we were.

Design for the workflow, not the screen. The two-persona problem was solved with task-flow thinking, not two products.

'Will they come back?' beats 'is this usable?'. The Audit Trail wasn't requested but it fit a weekly job.

Personal growth: I now co-run discovery research with the PM by default, not because it's required, but because shared synthesis is harder to dismiss in prioritization conversations.