Coupa Spend Guard

Turning an invisible AI feature into an upgrade engine

The Problem:

Spend Guard’s value depended on one thing: customers noticing and trusting its AI-generated fraud insights. The preview lived deep in the suite, surfaced generic data, and failed to answer a simple question for procurement and finance leaders:

“Is this worth paying extra for?”

Who I designed for:

Primary: Procurement and finance leaders evaluating whether to add Spend Guard to their plan.
Secondary: Existing customers who had access to the preview but had never meaningfully interacted with it.

They needed a quick, low-risk way to see suspicious patterns in their own spend data and understand the potential impact if they did nothing.

My role:

As the Senior Product/UX Designer on Spend Guard, I led the UX strategy, IA changes, and interface design for the new preview experience, working with PM, data science, and engineering to keep the solution technically feasible and aligned with the pricing/packaging model.

What I did:

Made the preview discoverable
- Simplified navigation and renamed entry points so the preview sat alongside other “risk & insights” tools instead of being buried in a secondary tab.
- Added contextual links from high-traffic dashboards where customers already reviewed spend anomalies.
Turned raw AI output into a story
- Partnered with data science to select a small set of high-signal indicators (e.g., suspicious suppliers, out-of-policy categories, abnormal invoice patterns).
- Designed a summary strip that answered, at a glance: “Here’s where your money is at risk today.”
- Used progressive disclosure so leaders could drill into a few concrete examples without being overwhelmed by tables or model details.
Aligned the experience with upgrade intent
- Framed the preview as “a limited look at what Spend Guard will monitor continuously,” so the UI set expectations without feeling like a hard sell.
- Designed an upgrade CTA that appeared only after users had explored their own suspicious spend, keeping the focus on insight first, upsell second.

Impact:

The stat cards show the result. Here's what changed behind it:

Customers found the preview without needing a direct link, because it now sat alongside other risk and insight tools.
They engaged more with real examples from their own data.
They treated flagged items as starting points for investigation, not noise.

What this reinforced for me:

AI features only drive value when people can find them, trust them, and see the business risk in plain language.
In enterprise SaaS, UX is often the bridge between “we have powerful models” and “decision-makers are willing to pay for them.”

This project shaped how I now approach AI-driven experiences in my work and in my book: show the real problem, make the insight unmistakable, then get out of the way so the business decision is easy.

+106%

YoY growth in Spend Guard
customer adoption

$135M

Non-compliant spend flagged

Sales demos

Adopted as a go-to demo for expansion revenue

Outcome:

By redesigning the Suspicious Spend Preview, I made Coupa's AI fraud detection easy to find and easy to trust, turning it into a real driver of Spend Guard upgrades.