The scenario
This playbook applies to a non-bank lender offering short-term working capital to small businesses, acquiring most of its applicants through paid search and comparison-site placements. The channel works, but competition for lending keywords is fierce and cost per click climbs steadily. The application form is usually a single long page, unchanged since launch and never tested.
A lender in this position typically engages Axoria for a conversion optimisation programme with the aim of getting more qualified applications from the traffic it already pays for.
The challenge
Financial services CRO carries constraints other sectors do not. The form has to collect information required for regulatory and underwriting purposes, so fields cannot simply be removed. Compliance has to approve every change to copy. And the metric that matters is not form completions but applications meeting the lender’s minimum criteria, since unqualified applications only cost the underwriting team time. The job is to find where qualified applicants abandon, understand why, and change the experience in ways compliance can sign off.
Research and analysis
We instrument the form at field level in GA4 so every focus, error and abandonment is recorded, paired with session recordings and heatmaps in Hotjar or Microsoft Clarity. In this scenario drop-off tends to cluster in two places: early, when applicants are asked for turnover and time trading before any indication of eligibility or cost; and late, at the document-upload step, where mobile users in particular give up.
We then run an exit survey and short interviews with recent applicants to test the pattern. The recurring finding is that people do not object to providing information; they object to providing it without knowing whether it is worth their time. Many are also unsure whether checking eligibility affects their credit score, a concern most application pages fail to address.
Strategy
Every observation becomes a hypothesis scored on expected impact, confidence and effort, producing a ranked backlog rather than a wish list. Three themes usually sit at the top.
Eligibility before effort
Move a soft eligibility check to the start of the journey so applicants learn early whether they qualify, and state clearly that the check does not affect their credit score.
Multi-step structure
Break the long form into short, labelled steps with progress indication and saved state, so that leaving to find a document does not mean starting over.
Trust at the point of doubt
Place security, regulatory and data-handling information next to the fields that trigger hesitation rather than in the footer.
We recommend testing the multi-step redesign as one variant against the existing form, rather than as a series of small tweaks that would each take months to reach significance on the traffic a lender of this size has available.
Execution
Optimization
Once the redesigned form has been tested, field-level analytics show which steps still lose applicants, and those feed the next round of hypotheses. Underwriting acceptance is reported alongside completion rate so that any test which improves completions but lowers quality is rejected. Learnings feed back into lead generation: eligibility framing that works on the form is adapted into ad copy and landing pages.
How results are measured
Every test is measured through GA4 field-level events joined to the underwriting system, against a baseline captured before the first test launches. We report significance, not just direction, and we never call a test on completions alone.
- Form completion rate: the share of form starts that reach submission, split by device. A successful programme shows a statistically significant improvement for the multi-step variant, with the largest gain usually on mobile.
- Qualified applications: applications that meet the lender’s minimum criteria, taken from the underwriting system rather than the analytics platform. Success means qualified applications rising in step with completions, confirming that the eligibility check filters out rather than encourages unqualified applicants.
- Cost per application: media spend divided by qualified applications. When the programme works, this falls with no change to media spend, giving the marketing team room to bid more competitively on the terms that matter.
What we have learned
- Diagnose before you test. Field-level tracking and applicant interviews point to specific problems, which is why the first large test is far more likely to win than a guess.
- Quality is a guardrail, not an afterthought. Reporting underwriting acceptance next to completion rate keeps the programme honest and is standard in our work with finance and lending businesses.
- Work with compliance, not around it. Batching approvals in advance turns the most common blocker in regulated CRO into a scheduled step.
- Measurement makes the case. None of this is visible without properly configured analytics and event tracking, which is where every CRO engagement starts.