A way for Shopify merchants to find the creators who already buy from them, see what to do next with each one, and act on it in one click.
Brief
OuterSignal take-home. Not shipped; the live product is theirs.
Product Designer
Sep 2026
PROTOTYPE
Live and coded in React
OuterSignal could already spot the influencers in a store’s customer list, but merchants had no idea what to do with them. I designed a ranking that gives every influencer one next step, explained in a sentence, and built it as a working React prototype. A merchant goes from 6,400 customers to a short to-do list, and can send a gift in four steps.
Live prototype, demo on request.
02 — What the product told me
No merchants to talk to, so the product was my evidence.
Influencers were flagged, not organised. OuterSignal already marked them with a signal, then sorted them like every other customer, so a 200K-follower buyer sat next to a one-time shopper. That became the Influencers tab.
Nothing said what to do next. Every card described a person and stopped there, so the merchant had to decide alone whether to gift, thank or ignore them. That became next steps.
Checking a customer cost the merchant their place. Clicking a dashboard card opened a separate profile page, and the way back landed on a different tab. That became the quick look drawer.
Rank by what to do next, not by a score.
A score tells a merchant who’s “best” but not what to do with them. A next step answers both. Every influencer lands on one of five steps, in rank order: Send Gift, Thank Them, Reach Out, No Action and Can’t Read, and each one is earned by a rule a merchant can read in a sentence. Six years designing lending products taught me that people don’t act on a number they can’t explain.
How Next Steps Work lists every rule in rank order, and the influencer list is sorted the same way.
Thresholds come from each store’s own data, because a 40-customer store and a 40,000-customer store need different bars. Every number shows its source and can be changed.
The awkward cases are part of the rules.
Private accounts. Their posts can’t be read, so they land on Can’t Read instead of being guessed at.
Small stores. Thresholds come from each store’s own data, so a store with three influencers still gets a ranking that makes sense.
Recently contacted. Anyone contacted in the last 90 days drops to No Action until they post again, so nobody gets chased twice.
Big creators who never posted. 100K+ followers earns Send Gift even without a post, because that’s the brief’s hero case.
Most of the work was making existing screens say more.
I worked inside OuterSignal’s existing components and colours, so the feature could ship without a visual rework. Select a number on any screenshot to see that change: New means I designed it from scratch, Extended means I reworked something OuterSignal already had. Before views are recreated from the live product, and all data is sample data. Screens without numbers are entirely my design.
Recorded in the live prototype. The card expands in place, so the list doesn’t jump and nothing on the condensed card moves.
A tab that didn’t exist yet.
The Signal Feed had no Influencers tab, so there was no overview, no sort order and nothing to filter. I chose the metrics for the band at the top and designed each module around a question a merchant asks: How far do they reach? Where do they post? Are they talking about us? Do they keep buying? I also tested four list layouts and kept one sorted list with a Next step filter, because group headers buried the influencers with nothing to do.
See all posts opens the Advocacy module into a wall of every post that tagged the store, with filters for recency, format and platform.
06 — The dashboard quick look
A drawer instead of a page jump.
Clicking a customer card on the dashboard sent merchants to a stand-alone profile page, and the breadcrumb back returned them to a different tab, so they lost their place every time they checked a customer. I replaced the page jump with a drawer that opens over the dashboard. Close it or press Esc and the merchant is exactly where they were.
Influencers open the influencer version of the quick look, with rows that jump straight to their Audience or Posts about you in the Signal Feed.
A gift the merchant can trust.
The brief asked for one action designed end to end. I chose gifting because it costs the merchant real money, so every step has to earn trust: pick a product, choose how it ships, confirm, then track it on the card.
Starting from the brief’s hero case. A claim link means nothing ships until Dana accepts, the summary spells out the cost first, and afterwards her card stays put and opens on the gift tracker.
08 — What I cut
Three ideas that didn’t make it.
A table on the dashboard. The dashboard is an entry point with no tables anywhere else, so a list there would compete with the Signal Feed instead of leading to it.
Demographic brand fit. Matching an influencer’s age or location to your customers measures resemblance, not whether their audience will buy.
Finding posts by product tag. Product tags are unreliable to detect, and the results suit research better than marketing.
09 — Validation
How I’d know it works.
A take-home leaves no room for merchant research, so the rules were built to be tested rather than trusted. First test: does the ranking match merchant instinct? Show merchants their own top ten and ask who they’d act on first, then whether they’d rather tune the thresholds or leave them automatic.
Second test: claim link or ship now as the gift default? Claim links are safer; shipping now is faster and may feel more generous. Compare acceptance rates and what merchants pick.
After launch, I’d watch four numbers: the share of next steps merchants act on, how often they override a step (the clearest sign a rule is wrong), gift acceptance rate, and the time from opening the tab to a first action.
Designing past the brief.
I built the prototype instead of stopping at static mocks, and it paid off. Working data exposed cases a mock hides, like a store with only three influencers, or a gift for someone who was contacted last week.
With more time, I’d put the rules in front of real merchants before polishing the card.
