Improve Performance, Scalability, and UX

Your Users Are Already Telling You Something

You don't need a user research project to know your product has problems. The data is already there. Bounce rates that are higher than they should be. Session durations that drop off at a specific point in the flow. Conversion rates that haven't moved in six months. Support tickets about the same three confusing things. These are your users telling you, through their behaviour, what isn't working.

Most product teams see this data and don't know what to do with it — or they make changes based on gut feeling rather than the actual evidence. The result is improvement work that doesn't improve the right things.

At NICKTUNG, improving performance, scalability, and UX is a discipline with a methodology: measure, hypothesise, change, verify. Not "we think this will help." Evidence.

Performance: What We Measure and What We Fix

Web performance is measured against standards that have both user experience and SEO consequences. Google's Core Web Vitals are the benchmarks:

• Largest Contentful Paint (LCP) — the time until the main content of the page is visible. Target: under 2.5 seconds. Above 4 seconds: poor. Every second over target costs you organic ranking and user attention.
• Cumulative Layout Shift (CLS) — how much the page shifts unexpectedly after initial load. Target: under 0.1. High CLS means users misclick; it's a trust signal that says "this site is unfinished."
• Interaction to Next Paint (INP) — how quickly the page responds to user actions. Target: under 200ms. Above 500ms: poor. Slow interactions feel broken.

The interventions that most often move these metrics: image optimisation and lazy loading (biggest LCP impact), eliminating render-blocking scripts, code splitting for large JavaScript bundles, CDN implementation for static assets, database query optimisation for server-rendered pages, and server response time improvements.

We measure with PageSpeed Insights / CrUX field data — real user measurements, not lab tests on our fast development machines. A fix that looks good in the lab but doesn't move field data isn't a real improvement.

Scalability: Capacity Before the Crisis

Scalability problems have a signature: everything was fine until suddenly it wasn't. A promotion drives traffic and the site goes down. A new enterprise client onboards and the database slows to a crawl. A background job that processed 100 records in 5 seconds now takes 30 minutes processing 100,000.

Our scalability improvement process:

• Load testing — simulating real traffic patterns at 2x, 5x, and 10x current peak loads to find where the bottlenecks are before users find them
• Horizontal scaling preparation — ensuring the application is stateless and can run multiple instances behind a load balancer
• Database scaling — query optimisation, connection pooling, read replica configuration, and sharding analysis where volume warrants it
• Caching strategy — Redis implementation for expensive queries, CDN edge caching for static content, HTTP cache headers for browser caching
• Background job architecture — moving expensive operations out of the request path into job queues so user-facing operations stay fast

UX Research That Produces Actionable Changes

UX improvement without research is decoration. Research without action is waste. NICKTUNG runs a focused 2-week UX research sprint that goes from observation to prioritised change list:

• Analytics review — identifying the flows with highest drop-off, the pages with lowest engagement, and the features with lowest adoption
• Session recording analysis — watching what real users actually do (Hotjar, FullStory, or similar) to see exactly where they get confused or give up
• Heatmap analysis — understanding where users look and click vs. where you expect them to
• User interviews — 5–8 conversations with representative users to understand the "why" behind the behaviour data
• Usability testing — task-based sessions where users attempt specific goals while we observe, identifying friction that analytics don't surface

The output is a prioritised change list with evidence — not "we think this would be better" but "users failed to complete this task 6 out of 8 times because of this specific friction."

What This Engagement Typically Costs

Performance + scalability audits and implementation: S$8,000 to S$35,000. UX research sprint and design changes: S$8,000 to S$20,000. Full improvement programme (all three dimensions): S$20,000 to S$50,000. EDG grant co-funding is available for qualifying projects.

Frequently Asked Questions

How do we know if the improvements actually worked?

We define success metrics before the work starts — specific Core Web Vitals targets, conversion rate benchmarks, task completion rates from usability testing. After implementation, we re-measure against these baselines. We don't declare success before the data confirms it.

We can't run a formal user research programme. Can you work with what we have?

Yes. If you have session recordings, analytics data, and support tickets, we can often identify the highest-impact improvements from existing data without a full research sprint. We'll tell you what additional data would be valuable and how to collect it efficiently.

Our site is on Shopify/WordPress. Can you improve performance on a third-party platform?

Yes, with some constraints. On hosted platforms, we can't change infrastructure. But we can optimise themes, reduce plugin bloat, implement proper image delivery, improve caching configuration, and reduce JavaScript payloads. Platform constraints set the ceiling; good optimisation within those constraints moves the needle significantly.

If your product is underperforming, the evidence is already there. Talk to NICKTUNG — we'll turn that data into a prioritised improvement plan.