Speed is not a one-off project: it erodes as assortments grow, extensions accumulate and media are uploaded unchecked. This category treats performance as part of maintenance — regular field and lab measurements, watching response times under load, caching for category, search and detail pages, database indexes and slow queries, image sizes and formats, and clearing out extensions nobody uses any more. Added to this are further operational topics that decide reliability in daily business: the deliverability of order and shipping emails, the division of tasks in the team and documented procedures. Every measure described comes with measurement, so regressions are noticed early. We additionally show how to keep measurements comparable over months so that changes do not stem from differing test conditions.
On this page you will find all posts on the topic “Performance & shop operations”. The newest posts are at the top.
Below each title there is a short text about it. Click on a post. Then you can read the whole text.
Above the list you also see the other topics. At the bottom, a link takes you back to all posts.
Zero results, missing synonyms and a stale index cost orders. How shop search becomes a fixed, measurable part of ongoing maintenance instead of a side note.
WebP, AVIF, lazy loading and an automated image pipeline keep page weight and load time stable in a running shop and protect the Core Web Vitals over time.
Find 404 errors and broken links in your shop: detect soft 404s, use 301 over 302, avoid redirect chains and make link hygiene a maintenance building block.
When a payment provider, shipping API or CDN fails, the shop must keep taking orders: inventory, blast radius, degraded mode and alerts for external services.
Shop updates without revenue loss: a low-traffic maintenance window, an SEO-safe HTTP 503 mode, cache warmup and a rollback plan with Git tag and DB dump.
Bulk-sender rules from Google, Yahoo and Microsoft: SPF, DKIM and DMARC decide the delivery of shop mails - and how to raise DMARC safely to p=reject.
Black Friday and sales peaks bring a multiple of normal load. Plan load testing 6 to 8 weeks ahead: realistic scenarios with buffer, breaking point and scaling.