top of page
Reliable access, reduced load
Improving Data Export Reliability
TL;DR
In one line: We turned a fragile, scattered CSV flow into a predictable morning batch with clear status, so responders get data without overloading the system.
Impact: Avg completion time −30 percent. Overload incidents −67 percent.
My role: Product and UX, end to end (research to spec to microcopy to delivery)
Background & Context
Angler is a phishing response platform for Singapore government agencies. When people report suspicious emails, each report becomes a case in Angler. Responders regularly export CSVs for operational reviews, investigations, and cross-agency audits. The challenge was not adding more features but making exports predictable, transparent, and scalable for daily work.
Why This Mattered
​CSV exports needed to be predictable, transparent, and scalable. Frequent failures slowed response and strained systems, so the goal was to make a core workflow work every time and explain what is happening when it cannot.
Discovery & Diagnostics
I interviewed responders, reviewed feedback, audited export logs, and analyzed overload incidents to find where requests failed or were retried. Methods included log and queue analysis, incident review, and UI walkthroughs.
What Was Broken
-
Capacity and timing: heavy exports during peak hours created load and failures
-
Visibility: no clear status led people to retry and create duplicates
-
Fragmentation and input quality: options were scattered, the date picker was slow, and key filters were missing
Solution
How It Works Now
​
-
Request with presets 7, 30, and Quarter and guardrails such as required date range and size limits
-
Process in a predictable 8 am daily batch with visible states: Queued, Processing, Ready
-
Download from a single hub; links expire; repeat requests attach to the existing job to avoid duplicates
Smart request handling
​
-
Required ranges and size caps reduce failures
-
Large exports are scheduled for the next day at 8 am to avoid peak load
-
We chose 8 am based on hourly request patterns to shift heavy jobs to the lowest traffic window
Clear UI and feedback
-
Show date range, status, file size, and download link
-
Preset shortcuts for faster and more accurate input
Unified and secure access
-
One place to request, track, and re-download
-
Role-based permissions for cross-agency access
Results
-
Operational efficiency up 30%: fewer retries and predictable delivery that frees responders
-
System overload incidents down 67% per quarter: caps and the 8 am batch reduced peak strain
-
Clearer outcomes: fewer failed requests, visible status, and explicit expiry
Reflection
When systems fail silently, people blame themselves and try again, which creates more load. Designing for failure states with clear status, limits, and predictable delivery rebuilt trust in a tool responders use every day.
bottom of page