Performance Optimization for E-Commerce Systems

Your shop loses revenue, rankings, and the trust of returning B2B customers with every second of loading time. HQ GmbH takes over performance optimization as a complete system – server, database, application, interfaces, and delivery – entirely in-house, for self-hosted systems such as Shopware, Magento, OXID, WooCommerce, TYPO3, WordPress, and custom developments. For fully SaaS-hosted platforms like Shopify, the focus shifts accordingly to the theme, app, and checkout level, as Shopify controls the server and database infrastructure itself.

Our Shopware Certified Solutions Architects, backend developers, and Linux experts analyze server infrastructure, database queries, PHP execution, caching levels, API runtimes, and ERP connections as a cohesive system, rather than viewing individual adjustment screws in isolation.

Content

1. How do you recognize performance problems?
2. Why performance determines revenue
3. Core Web Vitals: Target values at a glance
4. Why HQ is taking on this task
5. What we optimize
6. Shopware Performance Optimization in Detail
7. For which systems we optimize performance
8. The Performance Audit
9. HQ Performance Diagnostics: Our 5-Phase Process
10. References
11. Frequently Asked Questions

How do you recognize performance problems?

Before you invest in an analysis, it's worth doing a quick reality check. In practice, these symptoms almost always indicate an underlying performance problem, not isolated individual errors:

• Long loading times, especially on mobile devices.
• Poor Core Web Vitals scores in Google Search Console
• High server load even with average traffic.
• Slow search and category pages with a growing catalog
• High bounce rate immediately after the page loads
• Low conversion rate despite stable or increasing visitor numbers.

If two or more of these points apply to your shop, the cause is usually not a single adjustment screw, but the interaction of several system levels – this is exactly what we check in the free performance audit.

Why performance determines revenue

A slow shop is not a cosmetic problem, but an economic one. Every additional second of response time measurably increases the abandonment rate at checkout, lowers the conversion rate, and affects Google ranking, as Core Web Vitals have been considered a ranking factor for years.

In practice, poor shop performance rarely results from a single bottleneck. Usually, the database, cache, server configuration, frontend, plugins, and third-party connections work together simultaneously – which is why the isolated optimization of individual adjustment screws often only brings short-term improvements.

For online shops, there is an additional effect: customers with individual price lists and recurring orders also evaluate a shop based on its availability. A system failure on the day of a promotion or a blocked ERP connection during checkout not only costs the day's revenue but also the trust for the next order.

This means for you:

• ✅ Every optimization to the server, database, or interfaces directly impacts the conversion rate and ranking.
• ✅ Reliability is a trust factor in e-commerce, not a technical side note.
• ✅ Performance issues worsen with every growth step of your shop – not just on promotion days.

Core Web Vitals: Target values at a glance

Core Web Vitals have been part of Google's ranking algorithm for years. To help you classify your own metrics from Search Console or PageSpeed Insights, here is an overview of Google's target values:

If your shop is in the green zone for all three values, you are technically faster than most of the competition – measurable in the Search Console and noticeable at the checkout.

For a customer with a PageSpeed score below 50, we achieved a value of over 80 after the optimization – with checkout times reduced by 70% at the same time.

You can find a more detailed in-depth analysis with further practical examples in our Core Web Vitals guide. The optimization principles described there are platform-independent – the specific implementation only differs in technical detail between the systems.

TTFB as an early indicator: The Time to First Byte (TTFB) is not an official Core Web Vital, but it is the most reliable early indicator for a good LCP value – no element can load faster than the server responds. Values below 200–400 ms are considered healthy; from 800 ms, the response time directly affects the LCP. Since TTFB is generated on the server side, it can hardly be improved by purely frontend measures – this is precisely where our work at the server and database level comes in. We show you how to reduce your TTFB step by step in our TTFB guide.

PageSpeed Score is not the same as Core Web Vitals: The score from Google PageSpeed Insights (0–100) is based on a single lab measurement. The actual Core Web Vitals that Google uses for ranking come from real user visits (field data, CrUX report). A shop can achieve a high PageSpeed score and still have poor real Core Web Vitals values, for example, with fluctuating server load under real traffic. We therefore always check both data sources, not just the score.

What poor values actually cost: According to the joint study "Milliseconds Make Millions" by Deloitte and Google, an improvement in loading time of just 0.1 seconds increases the conversion rate in retail by an average of 8.4% and the average order value by 9.2% (travel industry: +10.1% conversions). Portent data also shows that e-commerce sites with a one-second loading time achieve up to three times higher conversion rates than sites with five seconds – and that the effect flattens most after the first second, meaning early optimization pays off the most. With 50,000 visitors per month, a 2% conversion rate, and an average order value of €80, a noticeable improvement in loading time can quickly make a six-figure difference in annual revenue. We will determine the exact figure for your shop in a free performance audit using your real visitor and conversion data.

Practical tip: Check your Core Web Vitals scores once a month and after every major update (new plugin, theme change, ERP connection). This way, a deterioration will be noticed before it shows up in sales.

Why HQ is taking on this task

Most agencies purchase hosting services and limit themselves to plugin configuration. HQ GmbH has been operating its own hosting since 1998 and has been developing custom software for high-load systems since 2001 – including its own advertising trading platform with double-digit billion deliveries during peak times, realized years before comparable systems became standard on the market. This infrastructure experience flows directly into every performance project.

Our beginnings with the performance optimization of web projects started back in 2003: A client, Cashcrawler GmbH, complained at the time about one of our supposedly "too slow" rental servers. A quick look at the server's load values actually revealed a consistently extremely high load. However, the initial brief analysis showed that the hardware was not the bottleneck, but rather inefficient software – even though the client's own developer had declared their system to be "error-free and 100% performance-optimized." We then made a risk-free offer to the client: We would optimize the code, and if that didn't bring any improvement, we would cover the costs. After three weeks of refactoring, the same server ran so smoothly that the client was able to run four projects simultaneously on identical hardware instead of just one, without any overload. He was very happy to pay the invoice back then and has remained loyal to us – now with his successor company – to this day. This project was our entry into software optimization and development – and the uncompromising focus on software efficiency before purchasing new hardware remains the foundation of our work to this day.

• 100% in-house. Server administration, database optimization, software development, and API integration are handled by one team – not distributed among external subcontractors.
• Experience instead of trial and error. We have been carrying out performance optimizations for customers for over 23 years – we know where to look first, what and how to measure to be efficient, and now have an eye for which plugins in shops can become real performance killers.
• A solid process instead of a gut feeling. Every project goes through HQ Performance Diagnostics – our structured 5-phase process from initial analysis to ongoing monitoring.
• Certified Shopware Solution Architects. Our team holds certifications as Shopware Certified Solutions Architect, Solutions Specialist, Certified Developer Advanced, and Backend Developer.
• Cross-platform expertise. We optimize complete e-commerce system landscapes – as a Shopware agency, Magento agency, OXID agency, Shopify agency, WooCommerce agency, TYPO3 agency, and WordPress agency – as well as individual PHP systems and headless commerce architectures.
• Fairness instead of profit maximization. Since our founding in 1994, one of the cornerstones of our corporate philosophy has been the principle that only long-term, fairness-based customer relationships are the key to success. This core value shapes every collaboration: We have always prioritized the customer's success over short-term profit maximization – as can be read, for example, in the case study " Rescuing a B2C Shop."
• Market position instead of assertion. Consistently ranked among the top 100 owner-managed agencies in the iBusiness/BVDW special evaluation of the Internet Agency Ranking since 2023, reaching the jury list for the Grand Prize for Medium-Sized Businesses in 2026 – achieved 100% through genuine service revenues.

What we optimize

For us, performance is not a front-end checklist, but a characteristic of the entire system. We work on all levels that jointly determine loading time, stability, and scalability.

Server and infrastructure performance

An undersized or incorrectly configured hosting setup limits any further optimization. We design server architectures from single servers to cluster systems with load balancing for high availability and, for customers with an international focus, globally distributed CDN connections that remain stable even with multiple times the traffic on promotion days – as our own enterprise hosting provider, not as a reseller of purchased capacity.

For a customer with over 300,000 existing customers, we replaced an unstable 8-server setup with a lean 2-server architecture – resulting in noticeably higher performance and server costs that were around €600 lower per month.

Your result.

• ✅ Stable operation even during traffic peaks and campaigns
• ✅ Scalable architecture instead of a server at its capacity limit
• ✅ Reliability as a contractually guaranteed standard

Database, Application, and Caching

Our developers analyze Apache/Nginx, MySQL/MariaDB, Redis, Varnish, database queries, PHP-FPM, reverse proxies, and cache layers as a complete system. Where other agencies configure individual components in isolation, we coordinate database indexing, Redis caching, and reverse proxy behavior so that the measures reinforce each other instead of blocking each other.

In an enterprise project with over 75,000 main articles and up to 750 million variants per article, we coordinated Redis, Varnish, and the database layer in such a way that a mobile PageSpeed score of over 90 was achieved despite this data load.

A current project shows how much this coordination now also depends on bot traffic: For one customer, bots and scrapers together generated over 3.2 million requests per day – SEO tools like Ahrefs and SemRush around 1,000,000, disguised scrapers without recognizable names around 1,600,000, GPTBot 650,000, and ClaudeBot around 15,000. For comparison: The Googlebot, which is crucial for ranking, is content with around 18,500 requests daily. Using invented URLs and changing filter combinations, these bots regularly caused the 16 GB Varnish cache to overflow and displaced real pages from the cache. Each of these actually cacheable requests had to be newly generated from Shopware and the MySQL database with every subsequent visit – with the result that the shop and database were indirectly burdened just as much by the displaced bot requests as by the bots themselves. We configured Varnish for this customer so that only meaningful traffic enters the cache – bot accesses to non-existent or irrelevant URL patterns for humans are rejected at the server before they burden the cache, application, or database. Result: Varnish is caching real pages again instead of hundreds of thousands of 404 responses, and the database and shop application are noticeably relieved.

To briefly explain: Redis stores frequently queried data in RAM, thus avoiding repeated, slower database queries. Varnish operates one layer before that and delivers already rendered pages without the shop application or database being involved at all. When properly coordinated, both layers complement each other instead of overlapping. Details on Redis and Varnish configuration can be found in our guide to cache optimization – here too, the logic is transferable across platforms, even if the article is written with Shopware in mind.

The database configuration itself is more of a hosting issue than a caching one: With standard systems like Shopware, Magento, WordPress, or TYPO3, the actual code base is usually unremarkable – which, in our experience, unfortunately often doesn't apply to all plugins from external providers. The actual bottleneck, based on experience (in about 80% of the optimizations we've taken over), is a default MySQL configuration on run-of-the-mill hosting that was never adjusted to the actual query load of the system in use. You can find more on this in our guide to hosting optimization.

Your result.

• ✅ Noticeably lower server load with the same traffic
• ✅ Shorter response times across all cache levels
• ✅ Fewer timeouts during checkout under load

Search, catalogs, and data volume

For growing item inventories and complex B2B faceting, we rely on Elasticsearch or OpenSearch architectures that are operated and indexed by our own team – not as a pre-configured third-party module.

An often underestimated side effect: Googlebot crawls fast systems more efficiently. For catalogs with several thousand items, the crawl budget determines how many product pages Google indexes at all – slow server responses measurably reduce the number of URLs crawled daily.

Your result.

• ✅ Fast search and category pages, even with a large catalog
• ✅ Stable faceting without additional server load
• ✅ Scalability for growing data volumes
• ✅ Better indexing rate for large catalogs through more efficient crawling

Interfaces: ERP, API, and third-party systems

Blocking live queries to SAP, PIM, or CRM systems are one of the most common causes of slow checkouts. Our integration architects develop API and ERP connections that decouple third-party systems instead of making the checkout dependent on their response time.

Your result.

• ✅ ERP and API connections no longer block the checkout
• ✅ Resilient interfaces, even in the event of a third-party system failure
• ✅ Performance and security for complex B2B processes

Frontend and Core Web Vitals

We improve LCP, INP, and CLS at the application level, not just through image compression. Our frontend measures complement an already high-performing server and database layer instead of replacing it.

Your result.

• ✅ Better Core Web Vitals scores as a ranking factor
• ✅ Perceived speed increases in addition to the actual response time
• ✅ No frontend fine-tuning on a structurally slow system

Monitoring and continuous control.

Performance changes with every data growth and every update. We set up monitoring via Prometheus, Grafana, and, if necessary, New Relic, so that deteriorations are noticed before customers notice them.

Your result.

• ✅ Early warning instead of reactive action after a failure
• ✅ Traceable key figures for management and IT management
• ✅ Basis for ongoing maintenance and further development

Shopware Performance Optimization in Detail

Most of the performance inquiries we receive concern Shopware stores – accordingly, we work specifically on this platform. These are the causes that are most frequently responsible for a slow Shopware store in practice:

• Outdated PHP version. Shopware 6 loses measurable execution speed with every PHP minor version that is not kept up to date. We check and maintain the PHP version as an integral part of maintenance.
• Faulty or missing cron jobs. Shopware 6 does not clear its cache automatically. Scheduled tasks must run at least every two hours, including correctly set timeout limits – otherwise, processes will continue to run uncontrollably and block the system.
• Improperly configured Elasticsearch/OpenSearch indexing. A correctly configured search usually responds in under 200 ms. Faulty indexes are one of the most common causes of slow category and search pages in growing catalogs.
• Too many or unoptimized third-party plugins. Each additional plugin loads its own assets and database queries. We check which plugins actually justify the loading time and which can be replaced by leaner alternatives. In one project, we replaced several performance-critical third-party plugins with tailor-made in-house developments and thus noticeably reduced the server load.
• Missing or incorrectly configured caching. Redis and Varnish must be coordinated with each other – configured individually, both deliver significantly less than when working together.

For which systems we optimize performance

We optimize entire e-commerce system landscapes, not just a single platform. However, the depth of technical optimization depends on the hosting model of the respective platform:

Special case Shopify: Shopify and Shopify Plus are SaaS systems fully hosted by Shopify itself. The server, database, and caching infrastructure are outside the access of agencies and merchants. Performance optimization here is limited to theme code (Liquid), app overhead, script loading behavior, and – with Shopify Plus – checkout extensibility. For full control over the server and database level, a self-hosted system like Shopware or Magento is the more technically appropriate choice.

First-hand example: On our own website, hqgmbh.de, simply switching from Elementor to native Gutenberg blocks increased the mobile PageSpeed score from 79 to 98 points – without any changes to Varnish, Redis, or the server.

Despite numerous optimizations, critical CSS, preloading, opcache, Redis, and Varnish caching, a page score of more than 79 was not possible with manageable effort.

After switching to native Gutenberg blocks without any further changes.

The Performance Audit

Before we propose a single measure, we will examine your system. The audit includes:

• Analysis of the server and hosting architecture
• Analysis of database, application, and cache levels
• Testing of API and ERP interfaces under load
• Load test during actual visitor traffic
• Prioritized action plan with cost-benefit assessment

The audit is free of charge and non-binding – with no upfront costs, no commitment to a major project, and without you having to provide your own IT resources.

HQ Performance Diagnostics: Our 5-Phase Process

We internally call our process HQ Performance Diagnostics: a fixed process in five phases, which ranges from the initial request to ongoing monitoring and gives you clarity about effort and priority from the outset.

1. Phase 1 – Contact & Initial Consultation. You describe the initial situation – system, traffic, objectives. We clarify the most important questions directly, without form-based back-and-forth.
2. Phase 2 – Free performance audit. Analysis of server, database, caching, and interfaces during real visitor traffic.
3. Phase 3 – Prioritized action plan. You will receive a comprehensible prioritization of measures according to effort and impact – not a lump sum, but an assessment that takes into account that individual optimizations often have a stronger impact in combination than in the sum of their individual effects.
4. Phase 4 – Implementation parallel to ongoing operations. The go-live of individual measures takes place in a controlled manner outside of main business hours, without downtime for your system.
5. Phase 5 – Monitoring & continuous control. We set up monitoring so that deteriorations are noticed before they show up in sales – including permanent server administration if desired.

References

Two projects from ongoing operations, with the actual figures:

cafiro.de – Rebranding & Record Migration: Migration from an outdated xt:Commerce system to Shopware, over 75,000 main articles, which before our migration sometimes generated up to 750 million variants per article and theoretically several billion combinations in total. Despite this data load, we achieved a mobile PageSpeed score of over 90 – with SEO visibility fully maintained during the simultaneous rebranding.

Rescue of a B2C shop with 300,000+ customers: The initial situation was an unstable 8-server system with a PageSpeed score below 50 and a conversion rate below 1%. After the restructuring: PageSpeed score over 80, checkout times reduced by 70%, infrastructure consolidated to 2 servers (approx. €600 lower server costs per month) and a conversion rate of 3–4% instead of the previous 0.4–0.7%.

"Due to massive performance issues and increasing dissatisfaction with our previous agency, we decided to switch to HQ GmbH. This decision has absolutely paid off for us. During the completely downtime-free migration, the performance of our online shop was improved by around 50%. At the same time, the HQ team identified and fixed various technical problems and errors that had previously gone undetected for a long time. We particularly appreciate the personal contact. For questions or problems, you can directly reach competent contact persons – even by phone. […] It is precisely this combination of technical competence, personal support, and reliable implementation that makes HQ GmbH a strong partner for us in our daily e-commerce business." — Jörg Schiermeier, Owner, schiermeier.biz

You can find more projects with customers such as ski-shop, wamiso, DMT a member of TÜV NORD GROUP and SCHIESSER in our Success Stories.

Frequently Asked Questions about Performance Optimization

Request a free performance audit now

With each day that passes without optimization, your system loses ground to faster competitors. A Solution Architect will contact you personally after receiving your request – no upfront costs, no sales pressure.

Prefer to speak directly? 📞 +49 3662 631760 · ✉️ info@hqgmbh.de

Further information: Shopware Agency · Enterprise Hosting · Maintenance & Emergency Support · Success Stories

Mathias Goldhan

Founder, Managing Director & IT Consultant

SaaS Pioneer, Enterprise Architecture & Cluster Expert

Mathias Goldhan laid the foundation for today’s digital agency in 1994 and has been leading HQ GmbH strategically and operationally since 2005. With more than three decades of experience in software development and systems architecture, he advises medium-sized businesses and brands on the digitalization of business-critical processes. He is the strategic mind behind HQ GmbH’s positioning among the Top 100 owner-managed digital agencies (iBusiness/BVDW).

As a recognized SaaS pioneer and specialist in high-performance hosting, he designs highly available cluster infrastructures and leads the technical rescue of stalled large-scale e-commerce projects. He contributes his expertise in resilient systems and API-based B2B automation internationally between the company’s headquarters in Thuringia and its branch in Spain.

View full author profile & e-commerce expertise