Sublium

1,000+ installs in 90 days. Setup time down from 6 hours to 4 minutes. 94% of merchants completing setup without filing a support ticket. The company's previous product took 6 months to hit those install numbers. Here's how it happened.

This project started with a Hotjar recording. Sarah, a tea subscription merchant, spent 6 hours trying to do something that should take a few minutes. She wasn't confused because she couldn't figure things out. She was confused because the tool was built for developers, and she thinks in terms of "customers get tea every month."

---

I watched Sarah's recording on a Saturday afternoon. She started at 2:14 PM. By 8:47 PM she'd closed the tab. Six hours. Zero subscriptions launched.

That quote became the north star. I recruited from the company's support ticket backlog - merchants who'd filed subscription-related tickets in the previous 3 months. 11 agreed to recorded sessions. Every session looked the same: long stretches of confusion followed by giving up. Merchants think in outcomes - "deliver coffee monthly with 10% off." Plugins demand system fluency - "create a variable product with 30-day billing intervals, configure a subscription coupon with renewal conditions."

The company had 50,000 WooCommerce merchants on its checkout tools. Support data kept surfacing the same pattern: merchants trying and failing to add subscriptions. 32% searched for "subscription" in docs within their first week. Most never made it.

---

I coded 847 support tickets from the previous 6 months alongside 12 user interviews. The market leader at $199/year was capable but built for developers. Shopify's native subscription setup takes under 3 minutes - but you need the Shopify ecosystem. WooCommerce merchants had nothing comparable.

I installed every major competitor and ran identical setup tasks:

---

Two rounds of affinity mapping. I expected failures spread across setup steps. Instead, 71% of breakdowns clustered in the first 20 minutes - before anyone even reached pricing or trial configuration. Merchants couldn't establish basic orientation.

I stress-tested three explanations: poor copy, missing documentation, mental model mismatch. Prototyped fixes for each, tested with 5 merchants per variant. Better copy cut setup time by 11%. An in-app help sidebar: 14%. Only the mental model reframe produced a real step change.

A guided five-step wizard: all 8 participants completed setup in 8 minutes. Then I asked them to go back and change one setting. The wizard forced them through all five steps again. "This is great the first time, but now it's in my way." Wizards optimise for first use and penalise everything after.

Relabelling alone - "30-day billing interval" became "Monthly" - dropped setup time to 2h 40m. Better, but still nearly 40x longer than the target. The labels were right. The structure underneath was still wrong.

---

It clicked during a research session. A merchant wasn't setting up one subscription - she was doing two separate things. First, defining her options (monthly, quarterly, annual). Then deciding which products each option applied to. Two distinct activities. Every plugin on the market treated them as one.

I checked this with the next 6 merchants. 5 of 6 described subscription options as a separate business decision from the product catalogue.

---

That observation led straight to the architecture:

Tier 1 - Plan Library: Create reusable plans once. Edit a plan, changes cascade to every product using it.

Tier 2 - Product Application: On any product edit screen, one toggle. Select which plans apply. Done in 30 seconds.

Built a prototype in 3 days, tested with 8 merchants. All finished setup in under 5 minutes. Asked them to add subscriptions to a second product the next day - every one did it in under 30 seconds without any guidance. That two-tier split is why setup dropped from 6 hours to 4 minutes.

PM pushed back: "Merchants with only one product type won't need a plan library." I ran a stress-test on single-product merchants - even they created an average of 2.4 subscription tiers. The library was simpler in every case. That conversation also pushed me to document data more rigorously for stakeholder discussions, a habit I've kept since.

---

Radio buttons over dropdowns for billing frequency. Dropdown users took 4-8 seconds, radio users 1-2. Small change, surprisingly big difference.

Trial period collapsed by default. Only 18% of subscriptions include trials. Showing trial options confused the 82% who didn't need them.

Customer portal. Tested action-first (pause immediately, confirm after) vs. preview-first (show what happens before committing). Action-first had a 23% "I didn't mean to do that" rate. Preview-first: 2%. The extra step wasn't friction - it was confidence.

Live preview panel. Merchants kept tabbing between settings and their live store to see changes. Real-time preview cut configuration errors by 67%.

Accessibility. I audited every competitor for keyboard support before wireframing. All had failures - radio groups not announced by screen readers, modals not trapping focus. I tested with VoiceOver during prototyping.

Engineering navigation. Lead engineer flagged that global plan management would add 12 weeks. I scoped V1 to ship product-level plans first using the same UI components, just pointing to a different data source. When global plans shipped later, merchants didn't need to relearn anything.

---

I underestimated brand diversity. 45% of merchants wanted to change copy, colour, or positioning. A brand personality preset during setup would've solved this. I was so focused on simplifying the flow that I forgot - merchants have their own identity. Simplicity without personality feels generic.

I cut something merchants actually needed. A "subscription notes" field turned out to be the third most-requested feature after launch. I'd run it through my simplicity filter and removed it. But simplification shouldn't mean deleting things people genuinely use for their operations.

I tested the wrong segment. All my participants were merchants who'd failed at setup. I didn't test with merchants who had working subscriptions on competitors. Turns out migration friction was much higher than I'd expected - merchants with 500+ active subscribers weren't going to manually recreate their plans, no matter how much better the product was. That blind spot cost us early adoption from exactly the people we wanted most.