Crunch
Replacing the friction of a 40-minute phone call with a self-serve onboarding journey.
Client
Crunch
Role
Product Designer

Timeline
2017 — 2019
Team
1 PM · 3 frontend developers · 1 backend developer · sales, accountancy, BA and marketing stakeholders
My role
Product designer — research, journey design, prototyping & engine scoping, API integrations, usability testing
What shipped
Reworked lead-gen + a full self-serve e-commerce onboarding journey, went live across Crunch's marketing site
Discovery
Poor conversion since the rebrand
Crunch completed a full rebrand in 2017 just before I joined. After the new site went live, we noticed the lead-generation forms weren't performing — the main call-back request, which booked a 40-minute slot with a sales advisor, was converting at only 11%.
“Our hypothesis: The newer messaging around the 40-minute phone call was too big an ask for an accountancy product signup.”
Mapping the journey, building buy-in
To get internal alignment before designing anything, I mapped the current Crunch digital experience end-to-end and printed it as a wall-sized graphic. It became the centrepiece of stakeholder walk-throughs — sales, marketing and product all standing in front of the same artifact, marking where the friction lived.

Competitor onboarding teardown
I walked through the onboarding of every comparable accountancy product on the market — time to signup, whether software access required a sales call, and what each step asked of the user. The audit gave us a concrete picture of where Crunch was falling behind and which patterns were users expecting in 2018.

Quick Wins
Just button testing
We theorised that the main “Get started” CTA was setting the wrong expectation — users didn't know what would happen after the click. I ran a button-test asking users what they expected next after a range of CTAs, and what would feel appropriate for a callback request.
Releasing the new design
The first release packaged a set of small but pointed changes to the existing callback form:
- Removed the open-ended “when's a good time to call” question.
- Swapped the CTA copy — “Get a callback” tested best.
- Told users how long the call would be, upfront.
- Listed what to have ready before the call.
- Introduced an option for a full online digital sign-up.

Learning from the release
We took the form completion rate from 11% up to 32%. Solid — but we were still losing prospects who didn't want any phone call at all. That gap defined the next phase: a proper self-serve e-commerce journey.

Self-Serve Journey
Internal alignment first
Before designing the new journey I built a presentation walking stakeholders through the pain points, usability testing video, and three next-step options. Sales, accountancy, product and marketing had to align on which option to pursue — the workshop made that decision shared, not handed down.

A recommendation engine for decisions
Accountancy isn't intuitive — picking the right Crunch package during signup was a real source of bounce. Working with the accountancy team and a business analyst, I designed a short set of qualifying questions that recommended the right package based on the user's situation. The questions doubled as relevance signals for the rest of the journey.

Validating a company with HMRC's API
We integrated HMRC's API to verify a user's company directly inside the e-commerce journey. The previous experience was a full-page detour; this one stays inline, so users never leave the flow they came in on.
Accountancy help tips
Earlier in the project I sat down with the sales and accountancy teams to walk every question in the journey. The output was a tooltip pattern attached to every question — short, plain-English explanations that gave users confidence to sign up themselves without a call.

32%
Lead form completion almost tripled
65%
Of new users via self-serve
+25%
Higher completion rate for self-serve vs call-back
Reflections
What I'd do differently
Two months of post-launch data showed the self-serve journey was outperforming the call-back option, but a chunk of users still picked the call. In hindsight, that wasn't a failure — it was a segmentation signal we under-read at the time.




Next project
