Ongoing engineering and technical optimization for a live German vintage-fashion eCommerce store serving customers across Europe. BrandBees supported Vintage Vogue through security incidents, catalog protection, multilingual SEO improvements, translation cost control, feed reliability, and custom WooCommerce functionality — while maintaining a production store that continued to serve customers.
4,552 Products
14 Languages
12,726 Watermarked AVIFs
Vintage Vogue
vintage-vogue.shop
Germany
Vintage Fashion / Online Retail
June–October 2026
Ongoing Engineering Retainer
WooCommerce Engineering, WordPress Security, Cloudflare Configuration, Origin Firewall Hardening, Custom Plugin Development, Multilingual SEO, Translation Optimization, Product Image Protection, Feed Automation, Returns Workflow Development, Performance Optimization
WordPress, WooCommerce
Flatsome Child Theme
TranslatePress
WP Rocket
Cloudflare Pro, nginx, Origin Certificate
WordPress
WooCommerce
Flatsome Child Theme
PHP and Custom Plugin Development
TranslatePress and TranslatePress AI
Cloudflare Pro
Cloudflare Web Application Firewall Rules
Cloudflare Super Bot Fight Mode
nginx Origin Firewall Rules
WP Rocket
AVIF Image Delivery
Internet Archive / Wayback Machine
Custom Copyright Archiving System
Google Search Console
Google Merchant Center
Google Shopping and Marketplace Feeds
Facebook and Bing Product Feeds
DHL and GLS Tracking Integration
Staging-to-Production Deployment Workflows
Let’s discuss how we can deliver the same results for your business.
Vintage Vogue operates a live multilingual vintage-fashion store with thousands of products, extensive product photography, international customers, and multiple external integrations.
The store uses WordPress and WooCommerce with TranslatePress supporting 14 published language targets. Product pages, translation requests, search indexing, image delivery, automated feeds, and checkout operations all need to work together.
During the engagement, the store faced a combination of security threats, performance constraints, SEO inefficiencies, and operational issues.
Key challenges included:
The store also needed a better way to document its product catalog for potential copyright disputes, while ensuring that image protection did not interfere with Google Shopping or marketplace feeds.
The objective was to strengthen security, protect product assets, improve multilingual SEO and performance, and maintain reliable eCommerce operations through controlled, incremental engineering.
BrandBees approached the engagement as an ongoing engineering partnership rather than a one-time website redesign. Work was divided into focused improvements, tested on staging where appropriate, and deployed to production in small, verifiable steps.
Vintage product photography is an important business asset. A screenshot folder alone does not provide the same traceable evidence as a dated archive of published product pages.
We developed a production archiving system that submits product and key shop pages to the Internet Archive, preserving time stamped references for future documentation.
Two independent archiving queues
To avoid overwhelming the live store, the system separates historical catalog processing from new and updated products.
Backfill queue
Live-product queue
When a product is unpublished, pending captures are removed from the queue, while completed captures are retained. A monthly process also queues pages older than 30 days for recapture.
Evidence and administration
The custom plugin records the Wayback link, capture time, and language for each completed entry.
The administration interface provides:
At go-live, the archive queue covered approximately 4,700 products in both German and English, alongside key shop pages. A subsequent repair cleared the failed backlog, bringing permanent failures to zero.
This provided a more structured way to document the catalog and maintain historical references to product listings.
Protecting product photography required more than applying a watermark to every image on the website. The implementation needed to protect customer-facing product imagery without contaminating the original assets used by shopping feeds.
We implemented a selective image-protection approach.
Watermarked product presentation
Watermarked image variants are used on individual product pages, in the lightbox, and in the gallery strip.
The image URL substitution runs only on single-product pages, allowing the shop grids, original image files, and external shopping feeds to remain clean.
This distinction is important because Google Shopping and marketplace feeds need the original product image rather than a watermarked presentation image.
Optimized image delivery
The watermarked assets are stored as AVIF files, reducing image weight while retaining the intended visual presentation.
Production holds 12,726 watermarked image files, with no missing AVIF versions reported in the project inventory.
One measured example illustrates the improvement:
Original PNG 694 KB
Watermarked AVIF 145 KB
Approximately 79% smaller for this measured image.
The result is a practical balance between image protection, efficient delivery, and compatibility with the store’s product distribution channels.
The store was operating under repeated automated traffic targeting its APIs, language-detection functionality, and common exploit paths. The investigation also documented periods of gateway errors while PHP and database resources were saturated.
We strengthened the existing edge and origin security layers while preserving legitimate access for customers and trusted crawlers.
Cloudflare configuration
The live environment uses Cloudflare Pro with Full (strict) SSL/TLS and an origin certificate. This ensures certificate verification between the browser and Cloudflare, as well as between Cloudflare and the origin server.
We configured and refined:
An earlier rate limit had locked genuine German and Belgian customers out for an hour, including requests involving a favicon. We refined the approach so the remaining restriction targeted cart abuse and used a two-minute duration.
Origin-level protection
Additional nginx rules deny known abusive addresses targeting the language-detection endpoint.
Generic download tools receive a 403 response, while Meta’s feed crawler is explicitly allowed so catalog synchronization can continue.
The security investigation documented repeated attempts against shop APIs and shell-style paths. It also recorded periods of gateway errors during resource saturation. The available incident evidence did not show a shell actually running on the shop.
Shipping data that disappeared during the incident period was restored, and edge protections remained in place after the storefront returned to HTTP 200 responses.
Background tasks are essential to a store that relies on scheduled product feeds, translation workflows, and catalog archiving.
During the engagement, Cloudflare challenged the server’s own cron requests, causing scheduled work to stall for approximately 49 hours.
We addressed this by:
WP Rocket remained part of the performance stack, but sensitive transactional pages were kept outside the HTML cache to avoid disrupting personalized shopping and account experiences.
These changes helped restore the scheduled workflows needed by the archive, product feeds, and other background operations.
A multilingual WooCommerce catalog creates additional technical complexity. Product filters, attributes, archive variations, and translated pages can generate large numbers of URLs that provide little additional search value.
During the investigation, the indexed URL count was approximately 110,000–115,000. Many of these URLs involved filters, product attributes, and thin language copies.
We refined the site’s search and indexing structure without removing useful shopping functionality.
Crawl and indexation improvements
We implemented the following changes:
These changes helped separate useful product discovery for shoppers from unnecessary URL variations being crawled by search engines.
Search performance during the engagement
The investigation recorded changes in indexed URLs and search performance over the engagement period.
Google indexed URLs
Initial investigation 110k–115k
Low point ~40,000
Later recorded count ~96,200 Interpretation
Indexing volume changed during cleanup and subsequent recovery.
The investigation also recorded two search performance samples:
Earlier sample
458 Search clicks
24,647 impressions
Later sample
1,497 Search clicks
92,959 impressions
These are investigation samples rather than a controlled attribution study. They document the search performance observed during the work; they do not establish that any single technical change caused the full increase.
Vintage Vogue distributes product data through several external channels. Feed failures can affect product availability, advertising, and catalog synchronization even when the storefront itself remains accessible.
During the engagement, the nightly Facebook and Bing feed refresh process stopped working correctly because the first feed in a batch removed callbacks required by the remaining feeds.
We regenerated all four feeds and restored them to a staggered nightly schedule.
We also verified the image handling rules so Google and marketplace feeds continue receiving original product images rather than watermarked variants.
Merchant Center observations
The investigation recorded the following Google Merchant Center figures:
The Merchant Center account remained open during the incident period. These figures provide a snapshot of the catalog and activity recorded during the investigation, rather than a claim that every listed product was approved or actively serving.
With 14 published language targets and thousands of products, translation lookups were placing unnecessary pressure on the database and paid translation workflow.
Investigations found lookups scanning millions of rows and approximately 930,000 junk strings in the paid queue. These included search leftovers, SKUs, and URLs that did not need translation.
We improved the workflow by:
The goal was to reduce unnecessary translation processing, control costs, and keep the translation workflow manageable at catalog scale.
The daily character limit also provides a defined upper bound for the configured translation process.
Security and infrastructure improvements needed to be accompanied by fixes to the customer-facing experience.
We addressed several operational issues identified during the engagement.
Category pages and product information
Cart and checkout
A white overlay was blocking the mobile add-to-cart experience. We removed the overlay and retested the cart.
A Portuguese DHL label containing broken placeholders was also patched, and the cart was retested following the change.
Customer privacy and forms
A cached contact form was displaying another person’s name. We adjusted the form behavior so it renders as a guest and clears stored fields, preventing the previous cached information from being shown to another visitor.
Email delivery
A four-hour email outage was traced to a staging filter that had been copied into the live theme. The filter was removed on the same day, restoring the intended email behavior.
Page weight and front-end performance
We deployed font subsets and deferred review widgets across product pages.
On one measured page, the raw HTML response decreased from 2.18 MB to 1.44 MB.
Approximately 34% less raw HTML on the measured page.
An image-markup experiment that broke the mobile layout was rolled back the same day. Additional menu-reduction experiments remained on staging rather than being released prematurely.
The store required a returns workflow that could distinguish between German orders, orders from other EU countries, and orders from outside the EU.
We developed a custom returns plugin that handles these different paths and supports carrier tracking.
Returns functionality
The implemented system includes:
The plugin and EU withdrawal form were deployed to production. However, the EU withdrawal form’s activation switch remained off on the live store at the time of reporting.
Alongside the live improvements, we developed and verified additional pre-order functionality on staging.
The staging implementation includes:
Customer-facing sentences remain in German so TranslatePress can translate them through the existing multilingual workflow.
These pre-order features were verified on staging only and had not been released to the live shop at the time of reporting.
Keeping these changes separate from production allowed the team to test functionality without introducing unverified behavior into the active store.
The engagement also addressed storage pressure and the safe management of staging environments.
Disk-space recovery
The shared disk had reached approximately 91% utilization on a volume of roughly 1.5 TB.
We performed several housekeeping operations:
These actions reduced unnecessary storage consumption without indiscriminately removing cached content needed by the live store.
Staging environment safeguards
We also established a server-side workflow for rebuilding staging from production.
Following a copy, the staging environment is configured with customer email, two-factor login, and live payment settings locked down.
This helps reduce the risk of staging activity interfering with production customers, payment processing, or store operations.
Time stamped German and English product-page captures, independent processing queues, capture records, queue management, and CSV exports.
Selective watermarking for individual product pages, lightboxes, and galleries, with original images retained for shopping feeds.
12,726 watermarked image assets stored in AVIF format, with no missing AVIF variants reported in the project inventory.
Cloudflare Pro security rules, bot management, exploit filtering, refined cart-abuse limits, and nginx origin-level access controls.
Cron access restored, scheduled workflows resumed, and sensitive transactional pages excluded from HTML caching.
Filter and attribute URL handling, sitemap refinement, corrected hreflang defaults, and streamlined category redirects.
All four product feeds regenerated and returned to a staggered nightly refresh schedule.
Approximately 930,000 unnecessary strings removed from the paid queue, with a daily translation character limit configured.
Fixes for cached customer information, email delivery, mobile add-to-cart, review badges, category descriptions, VAT presentation, and localized size labels.
Regional returns logic, DHL and GLS tracking checks, calendar-day return windows, and an EU withdrawal form.
Font subsets, deferred review widgets, and a measured reduction in raw HTML size from 2.18 MB to 1.44 MB on one page.
Disk-space recovery, debug-log cleanup, and safer staging rebuild procedures.
The engagement strengthened several critical parts of Vintage Vogue’s eCommerce operation while preserving the ability to serve customers throughout ongoing engineering work.
Cloudflare and origin-level rules now target documented abusive traffic more precisely, while preserving access for ordinary shoppers and legitimate feed crawlers.
The catalog archiving workflow records timestamped references for product pages and separates historical backfill from new and updated products.
More than 12,700 watermarked AVIF files support image protection on product-detail experiences without replacing the original images used by external shopping feeds.
The multilingual SEO work reduced unnecessary URL variations and refined sitemap coverage. Indexed URL counts subsequently moved from approximately 110,000–115,000 toward 40,000 before later recovering to around 96,200.
Four product feeds were regenerated and returned to a staggered nightly schedule, while the cron issue affecting background work was addressed.
Removing approximately 930,000 junk strings and imposing a daily translation cap provided better control over translation processing and associated costs.
Fixes to the cart, customer form caching, email delivery, reviews, product presentation, and front-end weight addressed specific issues affecting real store operations.
Staging safeguards, controlled production changes, and infrastructure cleanup created a safer process for ongoing development on a store that is already taking orders.
Better Security Controls
Improved Copyright Documentation
Protected Product Photography
More Controlled Search Indexing
More Reliable Product Distribution
Reduced Translation Waste
Improved Storefront Experience
Better Operational Resilience
Initial work across the live store, including investigation of abusive traffic, edge and origin protection, and infrastructure reliability.
Refinement of Cloudflare rules, origin access controls, cron reliability, cache exclusions, and scheduled processing.
Development of the copyright archiving system, product image watermarking, AVIF generation, multilingual URL cleanup, and sitemap refinement.
Translation queue cleanup, daily character limits, feed regeneration, and repairs to customer-facing store functionality.
Returns plugin deployment, front-end optimization, storage cleanup, staging safeguards, and continued engineering improvements.
The work was delivered incrementally under an ongoing retainer; the above periods summarize the engagement rather than represent isolated release dates for every feature. The project is still active under ongoing care plan.
Let’s discuss your project, timeline, and goals. No obligations — just a clear conversation about what’s possible.
Tell us about your project and we'll reply within a few hours (11 AM – 1 AM PKT). Chat replies within 1 hour.
"*" indicates required fields