Conversions Matter / Case studies
Real problems. Better decisions. Measurable results.
Experimentation should be about more than chasing winning tests. These anonymised case studies show how better measurement, research and decision-making helped businesses improve performance, build more reliable programmes and avoid misleading conclusions.
01 / Commercial impact
From experimentation to business value
Commercial outcomes matter, but the evidence behind them matters too. These two retrospective engagements show different approaches to delivering and evaluating substantial improvements.
International B2B ecommerce · Programme leadership
£50m
Additional annual revenue attributed to the programme in a historical external assessment
From an A/B testing tool to a global experimentation programme
An established testing platform but no consistent ownership, prioritisation or experimental safeguards. See how central governance, local market input and a repeatable decision process brought rigour to international optimisation.
Read case studyRetail and consumer finance · Checkout optimisation
+14%
Relative uplift in online application completion; approximately £20m subsequently reported in annual value
Making a lengthy credit application easier to complete
Analytics, heatmaps and click maps highlighted friction in a substantial online application. A clearer, staged experience helped reduce perceived effort while keeping necessary credit and identity requirements in place.
Read case study02 / Experimentation in practice
The decision matters more than the uplift
Not every useful experiment has a simple success story. Sometimes the right metric changes the interpretation. Sometimes the answer is to fix the underlying customer problem rather than the interface around it.
A/B testing · Guardrail metrics
−32.51%
Change in successful application rate for a variant that improved upstream progression
When a conversion uplift hides a worse outcome
One homepage variant increased progression and completed applications, yet produced a lower successful application rate. A practical example of why experiment decisions need more than a single attractive number.
Read case studyB2B ecommerce · On-site search
+14%
Reported relative uplift in order rate in a controlled experiment
When searches returned nothing, we fixed the search
With around 15% of queries returning no results, the team tested smarter query matching and extended-catalogue suggestions. The result illustrates the value of addressing the cause of a dead end instead of simply redesigning the empty page.
Read case study03 / How we work
A repeatable approach. Not a collection of tricks.
Every organisation starts somewhere different. Some need reliable measurement before they can trust a test result. Others have plenty of data but no prioritisation, or an experimentation platform without the processes needed to make good decisions.
Our approach connects the work from evidence to action:
These case studies are anonymised to protect client confidentiality. Historical commercial figures and test results are presented with context on attribution and limitations in their individual reports. Results are examples of past work, not forecasts or guarantees.
