All work

Fintech | Workflow | Customer Journey

Short-Term Loan Journey — 10 min to 2 min

Redesigned a short-term loan application journey and its integrations, taking application time from roughly 10 minutes to around 2 — treating it as a journey problem rather than a technology one.

Journey RedesignAPI IntegrationsAutomationLOSFunnel Analysis

10 → ~2 min

Application completion time

01

Context

The short-term loan journey worked, but completing an application took around ten minutes — long enough that the effort itself became a reason to drop off.

02

Problem

It was easy to read this as a performance problem. It wasn't. Time was spread across data the customer had to type manually, integrations called at the wrong moment, and steps that no longer earned their place in the flow.

Before

Manual coordination, limited visibility, effort scaling with volume.

After

Guided digital journey, automated deterministic checks, shared status model.

03

Customer / User

Customers applying for short-term credit, plus the operations team handling the applications that need intervention.

04

Discovery

  • Walked the journey step by step and timed where the ten minutes actually went
  • Separated time spent typing from time spent waiting on integrations
  • Reviewed which fields were genuinely needed for a decision
  • Analysed stage-to-stage drop-off across the funnel

05

Problem Definition

Which parts of this ten-minute journey are actually adding value, and which are just cost the customer is paying on our behalf?

06

Goals & Success Metrics

Application completion time

Materially lower than the 10-minute baseline

Manually entered fields

Replaced by retrieval where a source exists

Blocking integration time

Removed from the critical path where possible

Funnel completion

Improved stage-to-stage conversion

07

Constraints

  • Live journey with customers in it — changes had to be phased
  • Third-party latency and rate limits outside our control
  • Compliance-mandated steps that could not be removed, only re-sequenced

08

Solution

Fetch instead of ask wherever a trusted source existed, re-sequence integrations so the customer isn't waiting on them, and delete steps that were no longer earning their place.

09

User Journey

StartAuto-filled detailsVerificationOfferConfirm

10

Product Flow

Retrieve

Third-party APIs populate what the customer would have typed

Re-sequence

Slow calls moved off the blocking path of the journey

Reduce

Low-value fields and steps removed rather than optimised

Measure

Stage-level timing tracked after each change

11

Technical Architecture

Customer

Journey layer

Integrations

LOS

12

My Role

  • Framed the ten-minute problem as a journey problem, not a performance one
  • Ran the stage-level timing analysis and prioritised what to change first
  • Defined the target flow, integration sequencing and acceptance criteria
  • Partnered with engineering on phased rollout and validated results post-release

13

Key Decisions & Trade-offs

Fetch instead of ask

Replace manual entry with retrieval wherever a trusted source existed.

Trade-off: More integration dependencies to manage in exchange for a much shorter form.

Delete before optimising

Remove steps that no longer earned their place instead of making them faster.

Trade-off: Required agreement from stakeholders who owned those steps.

Measure before automating

Quantify stage-level time first, then change the largest contributor.

Trade-off: Delayed the first fix, but avoided optimising the wrong step.

14

Execution

  1. Baseline

    Instrument the journey and quantify time per stage

  2. Prioritise

    Rank changes by time saved against effort and risk

  3. Roll out

    Phase changes with monitoring and rollback paths

  4. Validate

    Re-measure application time and funnel completion post-release

15

Results

10 → ~2 min

Application completion time

Metrics represent selected project outcomes.

16

Reflection

  • Instrument stage-level timing from the start rather than as part of the fix
  • Test the reduced flow with real customers before scaling it to all traffic
  • A slow journey is usually a design problem wearing a performance costume
  • The cheapest speed improvement is a step you can justify deleting