Copilot in Quick Search and Advanced Search

A ZoomInfo project to get more out of the Copilot subscription by surfacing its AI signals inside Quick Search and Advanced Search, where salespeople already spent most of their time. I was the solo designer on the project, working with a PM.

ZoomInfo Copilot is the premium AI tier of ZoomInfo Sales. It surfaces signals that would take a salesperson hours to find manually: buying intent on a target account, a funding round, a decision-maker you worked with before joining one of your accounts. Copilot tells you who to reach out to right now instead of leaving you to search manually.

Some time after launch, adoption of Copilot signals was lower than expected. Analytics showed why: users were still working the way they did before Copilot. They opened Quick Search (searching by contact or company name) or Advanced Search (searching with advanced filters) out of habit, prospecting manually and skipping the Copilot homepage where most signals lived.

I ran qualitative research to dig deeper into this pattern. I specifically recruited users who had access to Copilot but primarily used Quick Search and Advanced Search without engaging with signals. The goal was to understand why these power users weren't utilizing Copilot and to explore how signals could improve their search process.

What I found matched what the analytics hinted at. Users hadn't carved out time to learn a new surface. They'd tried the Copilot homepage briefly and hadn't seen immediate value. The companies it surfaced weren't always the ones they were actively working with right now, so they'd learned to ignore it. The reasons weren't dramatic, just the mundane friction of forming new habits.

The PM and I decided to stop trying to change the habit and use it instead. If users weren't going to the Copilot homepage, we'd put signals inside Quick Search and Advanced Search, on the path they were already taking.

We set three goals for the work: raise awareness of Copilot features inside the main search flows, get users to actually engage with signals (click in, act on them) rather than just notice them, and make the homepage's value obvious from search so more users would eventually try it on its own.

The user stories were: as a salesperson, I want to know whether the companies I'm searching for have signals so I can act on them quickly, and I want to be pulled toward signals for my target and whitespace accounts so I don't miss opportunities.

I started by looking at how other products surface AI features and real-time signals inside existing workflows. I don't have my original write-up anymore, so the notes below are reconstructed from the competitor screenshots I collected at the time.

Across Apollo, Lusha, LinkedIn Sales Navigator, Google, Yahoo Finance, Nasdaq, Marketwatch, Figma, and Loom, one pattern kept showing up: AI features land better when they appear inside the flow a user already has, not behind a separate tab or destination that needs to be remembered.

Apollo and Lusha show AI insights directly inside the company and contact views. Apollo also has an AI assistant inside search and list flows, so the assistant meets the user where they're already working. LinkedIn Sales Navigator opens its notifications side panel automatically when the page loads, so updates sit next to whatever you're looking at. Google and Yahoo Finance inject AI-generated summaries and commentary directly into search results and stock pages. Figma introduces new AI features with contextual popovers on the canvas; Loom uses first-run prompts right where users work.

That matched what our own analytics were showing. Users weren't leaving their habits to visit a new surface, so we stopped planning as if they would.

I pulled a few other designers into a shortened Google Design Sprint so I wouldn't over-anchor on my own first ideas.

I presented the problem, the user stories, and the competitive patterns I'd gathered. We ran Crazy 8 on Quick Search (eight sketches in eight minutes, each person on their own), then again on Advanced Search. We scored the strongest sketches together, and I carried the best ones forward into my initial design concepts.

The sprint wasn't about converging on a single answer in a day. It was about widening the pool of ideas before I committed to a direction, so the concepts I took into reviews weren't just the first thing I sketched alone.

Quick Search is the fastest way into the product, and most people open it without thinking. That makes it a good place to put signals, because they ride along with the habit instead of fighting it.

Surfacing top signals before and during search

When a user opens Quick Search, a "Top signals for you" panel sits alongside the dropdown. It shows the user's target accounts that currently have active signals, with a count next to each. The panel stays visible while the user types, so signals stay one glance away during an active search, not only before it.

If a result in the dropdown has signals, the company shows a signal count inline (e.g., "5 signals"). If the result isn't in the user's target accounts yet, a "+ Follow" appears next to it, which ties into the follow flow below.

Drawing attention to signals after selection

When a user clicks into a result, the company profile opens with its Signals section visually highlighted. Signals were already on the profile page before this project, but they often got lost among everything else and users just didn't notice them. Highlighting the section on arrival is a small change, but in testing it reliably pulled attention to the right place.

Turning interested browsing into a follow

Copilot only generates signals for companies a user follows (target accounts) or is assigned (whitespace). If a user keeps landing on a company that isn't in either bucket, they get nothing from Copilot for it. So when a user views the same non-target company more than twice, we invite them to follow it and start getting signals for it. The invitation reacts to observed behavior instead of assuming intent.

If the user follows, a toast tells them signals will start appearing the next day. That sets a concrete expectation instead of a vague promise, which matters for a feature whose value isn't instant.

Advanced Search is the heavier tool for building prospect lists. The surface is a results table, not a dropdown, so signals had to work inside a dense tabular layout.

A Signals column on the results table

I added a Signals column to the results table showing the count of active signals per company. Clicking a count opens a side panel with the full signal breakdown, recommended plays, and inline contact actions, so users can drill in without leaving the list.

Inviting users to follow companies to get signals

Companies outside the user's target accounts show an empty signals cell. Hovering that cell reveals a tooltip with a "+ Target Account" button, so users can follow the company and start receiving signals for it, using the same mechanic as Quick Search.

Copilot signal adoption increased after the rollout. When we looked at where the lift came from, almost all of it came from Quick Search; Advanced Search barely moved.

My best guess is that the Signals column got lost among the other columns in a dense table. Users scan results tables for the fields they already know, and a new column doesn't reliably break that pattern, especially not a column that most rows have no value for. Later projects that I wasn't part of picked up the Advanced Search side and tried other directions.

If I were doing this again, I'd treat Advanced Search less like "add a column to the table" and more like a separate problem that might need a different entry point altogether. The Quick Search work succeeded because signals showed up in exactly the moment a user was already acting, and they didn't have to compete with anything else for attention. Advanced Search didn't have that luxury: the column was fighting for attention on a screen that was already full.