
Selected work
Turning customer evidence into better decisions.
As Heidi’s founding designer, I built a research practice without a dedicated researcher. My studies shaped winning experiments. I then created AI skills and a customer panel to help our small team keep learning while delivering designs.
Company
Heidi
My Responsibilities
Creating research system
Leading research
Business case writing
Teaching others to use
Worked with
Design report
Product manager
Developer
Senior leaders
88%
of experiments informed by four research studies were successful.
74%
lower cost per five-person study: £1,135 → £300 through our customer panel.
5
Claude skills supporting the team’s research.
01
Prove the value
In a moderated study, 2 of 5 people missed mobile search. I checked session recordings and redesigned it. A/B tests increased homepage-to-holiday progression by 10.3% and landing-page search starts by 23.9%.

02
Reduce the cost
I built a panel of 700+ opted-in customers, giving us a cheaper way to recruit for studies.


03
Build support for the team
I turned the steps I used by hand into five linked skills.
Brief
Posing questions to the facilitator, to shape the brief
Recruit
Pulling pre-screened participants from the panel
Plan
Setting up the moderated or unmoderated study
Synthesise
Shaping the write-up, with the facilitator
"Research Hub"
Being the librarian who can answer product questions

04
Keep human judgement
Findings keep source links and participant counts. The team checks the evidence and makes the decisions.



The skills in action
Removing a page from the booking journey
Removing the Extras page meant a sizeable development effort. Stakeholders worried it could harm the purchase flow. Our two-person team used ‘Research Hub’ to compare and refine designs quickly across seven moderated mobile sessions. The single-page approach consistently worked best in those sessions. We made **20 design changes**, including clearer room labels and a more visible price summary. The findings reduced risk and gave stakeholders confidence to back the build. Next, we’ll test bookings and booking value.


Earning investment in research
Seven experiments informed by my research won. Those results gave the business confidence to invest further in research. I built the business case for Maze to help us get faster customer insights. We can now run unmoderated studies and refine ideas before committing scarce winter A/B test slots. We’re heading into the season with the tools to support more ambitious goals.
Reflection
What I learned.
01
Start with the decision, then choose the research.
Choose research that helps the team act.
02
Build research judgement into AI tools, and keep checking the evidence.
Give others tools to use the process.
03
Research guides the design. Live tests measure its impact.
Show how customer insight improves.
Research wins
Earlier research wins.
Search
2 of 5 people missed mobile search.
My redesign lifted homepage-to-holiday visits 10.3% (~£370k) and landing-page search starts 23.9% (~£810k).
+23.9%
Rooms
Research shaped clearer room choices.
Two tests lifted Holiday-to-Extras progression 3.5% (~£130k) and 13% (~£460k).
+13%
Purchase journey
Five sessions shaped clearer steps and labels.
Holiday-to-checkout progression rose 19.5% (~£680k).





