How grüum grew revenue
13.3% year over year.

When Urumi began working with grüum, customers were experiencing instability throughout the shopping journey. Page loads took 5–7 seconds, cold starts reached around 40 seconds, and checkouts timed out when customers were ready to buy.

Urumi worked directly with George Lagonikas, Founder & CTO, and the grüum team to investigate the causes across caching, server configuration, and application code.

1. Week 0

   Discovery and audit

   Review traffic patterns, the plugin stack, and cold starts with George and the team. Deliver a written audit and fix plan.

2. Weeks 1–2

   Caching and server configuration

   Fix Cloudflare configuration and tune the server and PHP workers. Cached responses fell from 4s to 0.3s.

3. Weeks 3–4

   Application and database fixes

   Optimise frequently used code paths, remove N+1 database queries, and provide a WooCommerce Product Bundles patch. Uncached responses fell from 5.7s to 0.4s.

4. Week 5 onward

   Migration and ongoing operations

   Move the store onto Urumi’s infrastructure. The published case study reports P99 below 900ms and zero incidents through the peak weeks that followed.

The work addressed both everyday slow pages and the unpredictable delays that made the store unreliable. It went beyond adding server capacity to correct the configuration and code behind those delays.

• Corrected Cloudflare and caching misconfigurations.
• Optimised frequently used application code and supplied a performance patch to the WooCommerce Product Bundles team.
• Tuned server configuration so workers could use the available infrastructure effectively.
• Removed N+1 database query patterns that added latency.
• Reconfigured PHP workers to eliminate cold starts.

grüum reported 13.3% year-over-year revenue growth on Urumi. The store handled 1.2 million requests per day at 99.99% uptime, with zero incidents through its peak weeks.

Over the same period, checkout timeouts and carts lost during cold starts were fixed. With a more reliable store, the team also ran more campaigns, launches, and promotions.

The team validated performance through stress tests that followed real shopping journeys: browsing, signing in, adding items to a cart, and placing an order. The source reports average response time of 740ms with p95 at 3.03s. A further stress test recorded a 770.09ms average, 314.14ms median, and 1,620ms p90.

Testing also sustained traffic above the store’s 1.2 million daily requests, with p90 under approximately two seconds at 300 virtual users.

For stores with similar symptoms, the first step is to find what is causing the delays. Misconfigured caching, repeated database queries, and untuned PHP workers can interrupt checkout even when the storefront appears healthy at quieter times.

The published audit process examines traffic, plugins, custom code, and the points where customers encounter slowdowns, then supplies a written diagnosis and fix plan.