NetBase
Pro
A skunkworks search box grew into a full analytics product. I found the middle ground between “too simple” and “too much.”
One product. Powerful enough to be a problem.
NetBase Enterprise could process billions of posts and slice sentiment a hundred ways. It was so dense that the trained analysts who ran it couldn’t hand anything off.
Every number a non-analyst wanted had to travel back through the handful of people who understood the system.
“We need something simpler.”
“We need something simpler for the people who aren’t analysts.”
The insight came from our own enterprise users. But “simpler” is not a spec — it’s a direction with no floor and no ceiling.
Instant Search: “simple” taken to the extreme.
We ran with the most literal interpretation possible — model it on Google. One box, instant basic analytics, no learning curve — reusing the data we already collected for Enterprise. It was no throwaway prototype: it shipped as NetBase Instant Search, a real product the company demoed and sold.
One box
Type a phrase, get analytics. No configuration.
Reuse the data
Everything already collected for Enterprise.
Fresher visuals
A lighter UI on top of the Highcharts layer.
Instant Search, rebuilt live — type any topic and hit search · watch the original demo ↗
Not a screenshot — a working rebuild of the product I designed. Every query processes “millions of posts” and snaps back key metrics, trend, top hashtags, and audience in seconds — the whole promise of search-bar simplicity, made real.
The hard part was culture, not design.
The dev team had always built to the platform’s defaults — that had always been enough. Instant Search asked them to reimagine the visuals on new tools. We ran it as a single co-located week: the team flew in from Asia to California, two continents in one room.
We framed the build as theirs. Given ownership of the stretch, they flourished.
You get engineers to build beyond “good enough” not with a longer spec — but by handing them the ownership.
The idea was validated. So was our mistake.
We demoed at the company’s first-ever user conference and it landed — users loved it. Then, in the same breath, they told us the truth.
Too simple
Shipped and loved — but on its own, too thin for a whole team to work in.
Somewhere in the middle
Simple enough to self-serve, deep enough to be worth using.
Too complex
All the power, none of the reach past trained analysts.
The “too simple” verdict was real — but Instant Search wasn’t a dead end. It was the floor we built up from. We grew Instant Search into NetBase Pro.
Two specific, actionable wants.
Share the data.Give the whole team access to Enterprise data — the handoff pain, named directly.
Modern visualizations.So they’d stop exporting data into other tools just to make it presentable.
A mid-tier product that respected non-analysts — without insulting the data.
Easy for a non-analyst
Self-serve for departmental and non-technical users.
Powerful under the hood
Leverage everything already living in the Enterprise platform.
Companion mode
Analysts stop being everyone’s data concierge.
Open a new tier
Package Enterprise + Pro to compete where we couldn’t before.
We believed users needed a step between the bare simplicity of Instant Search and the great complexity of Enterprise.
A product that respected non-analysts without insulting the depth of the data underneath. Now it had to be tested — because I refused to build a whole platform on a hunch.
I wouldn’t build a platform on a hunch.
I recruited likely early adopters on purpose — pulling them in early built ownership and made them more likely to buy at launch.
One team held that salespeople were the onlypeople allowed to talk to clients. I got shut out of recruiting entirely. I worked it patiently — making the case that structured research is a different job than selling — until they agreed to introduce and step back.
Weighted toward the users I most wanted to reach.
13
Participants
5
Companies
8
Managers & execs
9
Existing users
8 men, 5 women — 5 with no prior platform experience mixed in with 9 incumbents, so I heard the fresh reaction alongside the incumbent’s frustration.
Five things the market rewarded — and Enterprise lacked.
A checklist that shaped every Pro decision that followed.
Easy onboarding & self-service
A freemium / premium upgrade model
Simple but meaningful analytics
Ease of export to other tools
Polished visualizations
The people Pro had to serve.
Active Social Listener
Researcher
Casual Explorer
Executive
Analyst
“I ask the analyst to run reports for me because I don’t know how to use the system.”
“It would be great if I had a tool I could give to people to do the small stuff so I can focus on the bigger questions.”
How might we build something a non-analyst can pick up with minimal help — while leveraging the data proven in Enterprise, and flexible enough to grow for years?
“Grow for years” is the clause that drove the biggest structural decisions in the designs that followed.
Four decisions about where to sit on the spectrum.
Dashboard
Tabs + modular cards. The home base, built for extensibility.
Card Concept
One flexible, uniform card that absorbs many data types.
Filter & Search
Adapted from Enterprise — but curated, not copied.
Compare
A table, not cards — to protect the one-size rule.
Tabs and cards, chosen for extensibility.
Familiar enough to need no hand-holding — but the real reason was that a card-and-tab home could absorb new data types for years without a redesign.
One card. Many data types. Built to grow.
I audited every data typein Enterprise, worked through the permutations, and found the shared representations — so one card format could serve them all and absorb new types later.
One card size at launch. Uniform kept it clean — and the build tractable.
Reused from Enterprise — but curated, not copied.
Adapting the working Enterprise components kept us consistent and cut time to market. But my first assumption was wrong in a useful way:
Wrong assumption: the full filter set is a straight benefit.
Some filters relied on data we weren’t piping into Pro at all.
Importing everything would smuggle Enterprise’s complexity back in.
→ A large-but-curated subset. Simplicity was the mandate.
A table, not cards — a constraint forcing a better answer.
Cards would have forced multiple sizes and broken the one-size rule. So I studied how competitors compared data, then adapted a tableto our modular system — and got something genuinely more informative.
It shipped — idea to launch, and still selling.
Validated at the company’s first user conference and adopted — the requesting client got it, and it was folded into the product line.
Designed end-to-end through the full product and development lifecycle. Beta with 3 Fortune 500 companies, then to market — and it still sells.
The dedicated beta forum fed ongoing enhancements — simplified UI patterns, revised visualization colors, and the Compare feature.
It moved the commercial needle immediately.
15
New users tried or purchased on the strength of the new offering.
8
Existing accounts upgraded their licenses to include Pro.
2
Lost sales recovered— pulled back now there was a mid-tier to offer.
Three things with a longer life.
Compare → upsell
Shipped and drove companion-license revenue, sold as a Pro capability alongside Enterprise — a key upsell lever for the business.
A foundational system
The card-and-tab modular patterns carried forward as a template for later NetBase interface work.
A new way of working
Stood up the company’s first cross-functional design sprints and a documented US–Taiwan handoff that cut QA friction.
“Too simple” is not the answer either.
When a user says “make it easier,” the temptation is to strip everything away — and we did, all the way to one search box. The product lives in the middle, and you only find the middle by shipping the extreme, listening honestly, and building on what users actually said.
Research access is a political problem you solve with patience — not just a methods problem.
You get engineers past “good enough” by handing them ownership of the stretch — not a longer spec.
The first product I saw through the entire lifecycle — and where I learned to think in long-term product vision, not one feature at a time.
A skunkworks search box became a shipping, still-selling mid-tier product — because we found the middle instead of sprinting to the extreme.