Skip to content
All work05/05 · Case study
Spicom · Onboarding & Activation

Seven became one.

7 minutes of database hosts, FTP keys, and DNS A-records compressed into one click and 20 seconds.

Spicom's legacy install flow asked merchants to configure their own database, server, theme, domain, plugins, content, and activation — seven separate steps with 21 form fields and a 60% drop-off rate. We rebuilt the entire flow as one button. The work didn't disappear; the system just runs it. The simulator below replays the compression.

Role
UI/UX Designer · Research → prototype → ship
Scope
Onboarding · Activation · Workflow simplification
Timeline
Apr 2024 – Jun 2024 (7 weeks)
Team
1 PM · 2 eng · 1 designer (me)

Auto-runs on first scroll · click Replay to watch the compression again

Setup wizard · auto-compresses on scroll

OLD · 7 minutes · 60% drop-off

7 manual steps · 21 form fields

1m 30s

Database

  • MySQL host:
  • Port: 3306
  • User: root
1m 00s

Server

  • SSH host:
  • Public key:
  • Path: /var
1m 15s

Theme

  • Pick skin:
  • Manifest:
  • Variant: dark
1m 00s

Domain

  • DNS A-rec:
  • SSL: yes
  • TLS: 1.3
1m 30s

Plugins

  • Auth: y
  • Pay: y
  • CMS: y
1m 00s

Content

  • Bulk seed:
  • Pages: 5
  • Products: 3
0m 45s

Activate

  • Cache warm:
  • Live: yes
  • Confirm: yes

✓ 7 things auto-handled · 0 jargon fields

Time to first live site
7m → 7m 00s100% TASK SUCCESS
Brief

Six in ten merchants quit before they finished their first site.

The legacy installation flow was a 7-step technical configuration wizard. Database setup, server config, theme picker, domain DNS, plugin chooser, content seeding, final activation — each its own modal, each its own jargon. 60% of trial users abandoned before reaching their first live site.

Trial-user friction

60%
What's an A-record? Why do I need an FTP key? I just wanted to launch a store.
Trial user · post-abandonment surveydrop-off · pre-redesign

Setup time

7 min
The first time I tried it took me about 7 minutes — half of which was reading documentation.
Successful trial · pre-launch interviewmedian time-to-first-site
Why seven steps lost 60% of merchants

The form was asking the user to do the system's job.

Three diagnoses surfaced from the research, each pointing to the same root cause — the system had been designed around what was easy to expose, not what merchants wanted to do.

  1. 01

    Each step felt like a separate product.

    Database, server, domain, plugins — each step had its own UI patterns, terminology, and validation logic. Merchants had to context-switch seven times to finish one task.

  2. 02

    The form fields asked for the wrong things.

    Merchants don't have an opinion on MySQL ports or DNS A-records. They have an opinion on whether their site is live. The form was asking the user to do work the system should do.

  3. 03

    Drop-off compounded across steps.

    Each step lost ~12% of users. Compounded across seven steps, that's 60% drop-off — even when each individual step seemed 'simple enough' in isolation.

Three rules behind the one click

Move the work, don't hide it. Default the choice. Show the outcome.

One screen. One button. Three product rules, each with a specific stakeholder concern behind it.

01

for merchants

Move the work — don't hide it.

The system provisions the database, sets DNS, deploys the theme. Merchants don't see jargon disappear behind a black box; they see a progress bar and a working site at the end.

02

for engineering

Default the choice.

Every step had a 'most-merchants-want-this' answer. We picked it. Merchants who wanted custom settings could opt in via a 'Customize' link, but 92% never opened it.

03

for growth

Show the outcome.

The end-state — a live site at yoursite.com — is visible from the first screen. Merchants see what they're getting before they click. Trial-to-paid lifted +20% from this alone.

What the system does in those 20 seconds

The work didn't disappear. The system runs it.

One click on the user side. Seven backend tasks on the system side, running in sequence — Postgres provisioning, nginx + SSL, theme cloning, DNS, plugin install, content seeding, activation. Click below to watch both layers happen at once.

What runs in the 20 seconds the user waits

User view · what they see

Install your site

yoursite.com · ~20 seconds

~/install.log · backendSTREAMING
[0.0s]Provisioning Postgres on us-east-1
[0.4s]queued · Configuring nginx + auto-SSL (LetsEncrypt)
[0.7s]queued · Cloning theme template (default-storefront-v3)
[1.1s]queued · Setting DNS A-records (yoursite.com → 34.x.x.x)
[1.5s]queued · Installing core plugins (auth · payments · cms)
[2.0s]queued · Seeding sample content (5 products · 3 pages)
[2.4s]queued · Final activation + cache warmup
User actions vs system tasks
1 click · 0 / 7 tasks doneALL AUTOMATED
A/B test · Variant B beat the legacy flow

Same merchants. Different flow. Variant B won by 4.2σ.

We split-tested the old wizard against the one-click install for 60 days, sampling ~3,800 trial merchants. Variant B (the redesign) won across every single tracked metric.

Completion rate

60% drop-off → none

40%100%

Time to first live site

median across cohort

7m 12s20s

SUS score

System Usability Scale

5887

Trial-to-paid

downstream conversion lift

control+20%
What changed after launch

Completion went from 40 to 100.

Measured over 60 days of A/B testing on ~3,800 trial merchants. Tracked through completion analytics, time-to-first-site logs, SUS questionnaires, and revenue dashboards.

+0%

Hero metric

Installation completion rate

Same trial pool, same product, different flow. The 60% drop-off vanished — every merchant who started the install finished it. The downstream lift to trial-to-paid followed naturally.

40% → 100% completionA/B winner · 4.2σ confidence
  • +20%

    Trial-to-paid

    downstream conversion

  • −30%

    Support tickets

    install-related

  • 20s

    Median time to live

    was 7 minutes

  • 100%

    Task success rate

    post-launch sample

  • SUS 87

    System Usability Scale

    was 58

Trial user · post-launch interview

“This is the cleanest install flow I've used. It feels like I clicked a button and a website appeared.”
— Trial merchant · 30 days post-conversion
Reflection

Four lessons that traveled with me.

  1. 01

    When the user has to do work the system could do, the system is broken — not the user. Move the work; don't simplify the wording.

  2. 02

    Defaults aren't a fallback — they're the design. Choose the right answer for 90% of users; let the other 10% opt into customisation.

  3. 03

    Compounding drop-off is the silent killer of multi-step flows. Every step is a coin flip; seven coin flips lose 60% of users even if each one is 'easy'.

  4. 04

    Show the outcome before the work. Merchants who saw a live yoursite.com preview converted 20% better — even before they'd configured anything.

All workEnd of case study · 05/05