TTFB & backend profiling
Profiling with the tools that show the truth — Blackfire, xhprof, slow query logs — to find where time actually goes.
Magento Performance
Emyrix makes Magento 2 stores fast where it counts — server response time, cache, search, and Core Web Vitals — so pages load quickly and checkout holds up under real traffic. We fix the infrastructure and code ceiling, not just the Lighthouse score.
Magento & Adobe Commerce engineering from Massachusetts, serving e-commerce businesses across the United States.
Overview
Most Magento performance projects start in the wrong place: someone runs Lighthouse, sees red, and spends weeks minifying CSS while the server still takes over a second to return the first byte of HTML. We start with the layer that sets the ceiling for every page — TTFB, full-page cache, Redis, OpenSearch, PHP, and the database — then move to page-type and frontend work once the foundation is solid.
The goal isn't a one-time score bump. It's a store that stays fast as the catalog grows and traffic spikes, measured against real business metrics rather than a synthetic lab run.
Services
Profiling with the tools that show the truth — Blackfire, xhprof, slow query logs — to find where time actually goes.
Correct FPC, Varnish, and Fastly configuration, cache warming, and hole-punching so hits are fast and misses are rare.
Redis for cache and sessions, configured and separated properly instead of the defaults that stall under load.
Search and catalog indexing tuned so category and layered-navigation pages stay quick with large catalogs.
Query optimization, index management, and flat-catalog decisions based on your data, not folklore.
LCP, CLS, and INP work on category, product, and checkout pages, measured on real traffic.
Modules that run a query per product on a category page, an ERP or shipping API called synchronously in checkout, and the front-end scripts that each add a few hundred milliseconds.
Why Emyrix
We profile before we touch anything, so effort goes where the time is actually being spent.
Priority on the templates that carry revenue — category, product, and checkout.
Fixes hold at peak-season traffic, not only on an empty staging box.
FAQ
Usually the backend, not the frontend: slow TTFB from missing or misconfigured full-page cache, an under-tuned Redis or database, or heavy custom code. We profile to find the real cause before recommending work.
Usually because checkout is the one page the full-page cache can't help with, so every problem the cache was hiding shows up there. The common causes: shipping and tax rates fetched from a third-party API on every step, a payment module doing more work than it needs to, customer-section requests firing more often than they should, and a database under load from indexing or a report running at the wrong time. We profile a real checkout to see which of those it is.
Load a category page twice and compare the response times. A cached page returns in well under a second, often under 200 milliseconds. Magento adds an X-Magento-Cache-Debug header (HIT or MISS) in developer mode, and Varnish and Fastly set an Age header on a hit. If every request is a miss, something on the page is marking it uncacheable — a block with a customer-specific value, a session started too early, or a third-party module setting cacheable="false" in its layout XML. That last one is behind a surprising number of slow stores.
Sometimes the hosting, and it's worth ruling out first because it's cheap to check. Time to first byte on a cached page tells you most of it: if a page you know is cached still takes a second to start arriving, the problem is in front of Magento — the server, PHP-FPM, or the network. If cached pages are fast and uncached ones are slow, it's Magento, an extension, or the database. A store can have both at once.
Yes. We work on LCP, CLS, and INP across category, product, and checkout pages, using field data from real users.
Speed is one factor among many, but slow category and checkout pages measurably cost sales. We prioritize the templates that carry revenue so the work pays for itself.
Both. We tune existing Luma stores and, where it's the better long-term move, migrate to Hyvä for a lighter frontend.
Related
Work with Emyrix
Send us the URL and we will look at it — version and patch status, speed, caching, indexing, checkout — and tell you what we found.