Overview
TeamHub is LinkedIn's internal tool for managing accounts and rosters. As reps' books grew, two questions got harder to answer: is this account data even right? and which accounts should I work next? I designed the scrubbing and prioritization flow that made both fast — and helped scale TeamHub to 2,000+ reps.
Note: visuals and specific figures on this page are illustrative and NDA-safe. Swap in your own approved screenshots and metrics.
The problem
A rep's book could hold hundreds of accounts, and the underlying data was never perfectly clean — wrong owners, stale fields, duplicates. Reps either distrusted the data or spent hours grooming it by hand, with no clear way to tell which accounts actually deserved their attention.
"I don't even know if half of this is current — so I can't trust it to tell me where to spend my time."
— Recurring theme from rep interviews (paraphrased)- Messy data: incorrect or stale account fields eroded trust in the whole book.
- Manual grooming: scrubbing accounts was slow, repetitive, and easy to skip.
- No prioritization: nothing surfaced which accounts were worth working first.
Process
I split the problem into two linked jobs — trust the data, then act on it — and designed the flow so cleaning the book naturally fed the prioritization that followed.
Research & framing
With PM and data science, I shadowed how reps groomed their books and where they lost trust. That told us which fields mattered most to scrub and which signals reps actually used to decide what to work.
Explorations
I explored scrubbing as everything from a heavy bulk-edit table to a light, guided review of just the accounts that needed attention — and tested how prioritization should appear once the data was trusted.
Key decision
We chose a guided scrub that feeds a prioritized list: reps quickly confirm or fix only the accounts that need it, and that cleaned book powers AI-driven recommendations for what to work next.
Solution
Three ideas carried the design:
Scrub only what needs it
Instead of a wall of editable rows, the flow flags the accounts with questionable data and walks the rep through quick confirm-or-fix decisions.
Clean data, earned trust
As reps scrub, the book visibly improves — so the prioritization built on top of it feels credible rather than arbitrary.
Prioritized, with the "why"
AI-driven recommendations rank accounts to work next and show the signals behind each, so reps can act with confidence.
Impact
Replace these with your real, approved numbers.
Reflection
The lesson here was sequencing: prioritization only lands once reps trust their data, so the scrub had to come first and feel effortless. Tying the two together — clean as you go, then act — is what made the whole experience click.
If I continued, I'd close the loop further: let how reps actually work accounts refine both the data-quality flags and the prioritization over time.