Usability audit and testing of a complex onboarding and KYC process in the web account of a fintech service

Initial situation: a B2C fintech service has a web account (desktop and mobile web) for opening an account, completing KYC and making the first deposit. After the redesign, the conversion from registration to successful completion of KYC fell from 42% to 31%, the number of support requests on topics increased - document verification does not pass, it is not clear where the status is, errors in selfies, unclear photo requirements. There are guesses, but we need confirmed reasons and prioritized improvements.

Technical environment and limitations: Figma for layouts, product in React, analytics in GA4 + Amplitude, events are already partially marked, KYC provider logs are available as a CSV download without personal data. Access to production is only through test accounts and staging. The audit is carried out without recording real personal data - use synthetic data and test documents from the provider. You cannot propose changes that require changing the KYC provider or completely rewriting the front end. The development team will be able to implement only spot edits to the UI/UX and texts, as well as minor changes to validations and the sequence of steps.

What needs to be done in stages:
1) Quick audit of the current UX process - registration - phone/mail confirmation - questionnaire - uploading documents - selfie - waiting for status - first replenishment. Find problem areas using heuristics (Nielsen) and fintech-KYC patterns, separately highlight compliance risks and unacceptable wording.
2) Data analysis: check funnels, drop-off by steps, segments (mobile/desktop, browsers, new/returned), main KYC error scenarios from CSV, compare with GA4/Amplitude events. Formulate 5-10 hypotheses for the reasons for the drop in conversion and measurable expectations for improvement.
3) Plan and conduct moderator usability tests remotely: 6-8 respondents from the target audience (Russian-speaking users 22-45, basic financial literacy), staging scenarios, screen and voice recording. The tests must check your understanding of the requirements for documents/selfies, perception of the reasons for refusal and retry, clarity of statuses and time expectations.
4) Additionally - expert cognitive walkthrough for the retry scenario after KYC failure and for the scenario " document accepted - waiting - additional request. data".
5) Final report and artifacts:
- Table of problems (at least 25) with: where it occurs, description, evidence (screen/timecode/metric), severity (0-4), frequency/coverage, suspected cause, recommendation, complexity of implementation.
- Prioritization by Impact/Effort and a list of top-10 changes that can be made in 1-2 sprints.
- Specific microcopy (RU) edits for key screens: photo requirements, errors, statuses, waiting, resubmission.
- Recommendations for event taxonomy for GA4/Amplitude: minimum 15 events/parameters to measure the impact of changes.
- Short presentation (up to 15 slides) for the product team with conclusions and next steps.

Measurable acceptance criteria:
- The report contains at least 25 unique problems, at least 10 confirmed by data (funnel/errors/segments) and at least 10 confirmed by observations from timecode tests.
- For the top 10 problems there are specific, implementable recommendations without changing the provider and without rewriting the application.
- Test scripts, screener and anonymized notes/results are transferred to the customer.
- The proposed event diagram covers all funnel steps and at least 80% of typical KYC errors from the upload.

Work format and results: all materials in Google Docs/Sheets + Figma comments (if necessary), test records in a secure folder, final meeting of 60 minutes to analyze the findings.