← Back to work

Enterprise B2B · AI + Accessibility

D&B Hoovers: leading a legacy platform back to focus, accessibility, and AI that fits the work

Reframing a legacy sales-intelligence platform around focus and accessibility, and the cross-team leadership it took to hold that line.

Role
Senior UX Designer · Accessibility program owner · Governance Board · AI co-lead
Timeline
2024-2026
Tools
Figma, UserTesting, D&B design system
Focus
IA, AI search, accessibility program, cross-team alignment
AI

The context

D&B Hoovers is a premier business-intelligence platform used by sales and marketing teams worldwide. Years of growth had layered on complexity: dense screens, deep navigation, and accessibility gaps that no one owned. The tool meant to speed reps up was slowing them down.

The people using it spend most of their day prospecting: building lists, scoring accounts, chasing intent signals under quota pressure. They did not need more features. They needed the features they had to get out of the way.

What I owned

I led design on the redesign, but the larger job was getting an organization to agree on what better meant, then holding the line on it.

Accessibility program, from nothing

No owner, no standard, no measurement when I started. I set the WCAG standard across products, stood up the working group, and produced VPATs for top-tier offerings. 8 to 10 product teams adopted it.

Full design-system adoption

Every product used pieces of the system; none had fully adopted it. I drove the prospecting platform to ingest it completely, which made consistency and accessibility enforceable instead of aspirational.

Design Governance Board

Redesign decisions had to survive review against the shared design system and the needs of other product lines. I sat on the board making and defending those calls.

AI integration, co-lead

Co-led AI in the prospecting workflow with the Sr. Director of Product, which meant deciding where AI belonged and, more often, where it did not.

"Mike took charge and led the effort to overhaul our accessibility compliance process, and was always a great partner in brainstorming."
Edward Garaña · Manager, Dun & Bradstreet

The constraints

Enterprise work with real limits. An established design system and a governance board to align with. A legacy architecture that could not be rebuilt overnight. Accessibility that, until I made it a priority, had no owner driving it.

Under NDA I can't show the final product screens. Everything visual here is an original illustration of the decisions and patterns, not the shipped UI.

The decisions

The redesign came down to a handful of calls I had to make and defend, not a process I ran start to finish.

Removal over addition. Research pointed at overwhelm, not missing capability. The hard part was holding that position when product wanted more data per screen. I cut the default view from hundreds of fields down to the 8 most users actually relied on, and let anyone who needed a different set build their own. Density that stakeholders were attached to had to go, and I made the case with rep footage, not opinion.

Before · hundreds of fields After · the 8 that matter List Builder + Customize Company name Industry Revenue Employees Location Intent signal Last activity Contact
An illustration of the same decision: the default List Builder view cut from hundreds of fields to the eight most reps actually used, with any other field a click away.

AI only where it fit the flow. The pressure was to add AI broadly because it was a company priority. I scoped it to the places it measurably helped, search and time-to-insight, and argued against putting agents on small tasks where they only added noise to a workflow that was already too noisy. AI earning its place was a decision I had to win more than once.

Accessibility designed in, not retrofitted. I pushed to make conformance a property of the components and the governance review, so it could never be a cleanup pass at the end. That meant slowing some work down early to avoid rework later, which I had to justify to product management, who always pushed for more features.

The solution

The redesign pulled years of accumulated density out of the core prospecting workflow. Fewer choices per screen, a clearer hierarchy, and the most common actions always within reach. Four clicks to a buried action became one. AI search was woven into that flow rather than bolted beside it, so reaching an answer became part of prospecting instead of a separate detour.

Before · deep, nested Home Section Subsection Buried action After · shallow, direct Prospecting workspace Search Lists Signals
An illustration of the structural shift: four clicks to a buried action became one, and the most common work moved to the surface.

The outcome

The redesign and the accessibility program landed together. The task-speed numbers come from moderated user tests on a Figma prototype I built, around 10 participants per test. I set the baseline through my own timed runs and a competitive analysis, then re-ran the tests after each feature was designed and launched.

28%
Faster daily tasks for sales reps, by how quickly a participant could reach their data
50-65%
Faster time-to-insight on AI search in prototype testing
40%
Fewer accessibility issues across products, from real scans run across the line

What I'd carry forward

The biggest wins came from removal. Reps wanted a calmer tool, and AI earned its place only where it fit the existing flow. If I ran it again I'd put the accessibility program and the redesign on one timeline from day one, so conformance is designed in rather than retrofitted.

The other lesson was organizational. The design was the easy part. The harder work was alignment: sign-off from the Sr. Director of Product who owned it, five product managers, a VP of Engineering, and a Sr. Director of Engineering, each with their own priorities.

Next case studyRealogy Leads Engine