Editor’s Note: Jascha McDonald is Software Developer and Application Architect for BankVogue, North America’s trusted partner in buying, selling and shipping used goods around the world. One of their companies includes Beyond Retro, a vintage clothing retailer in Europe.
Founded in London in 2002, Beyond Retro’s stores offer an exhilarating pick ‘n’ mix of vibrant 20th century fashion. Led by an inspiring team of true originals, the staff are the beating heart of a creative community that spans from England to Sweden with verve and style. Beyond Retro’s retail network recycles upwards of 500,000 items each year. Purchased directly or indirectly from charities, they’re part of a global movement that brings new purpose to excess clothing and accessories.
Beyond Retro’s Magento-powered global site offers more than 35,000 objects with a twist: they’re all unique pieces! We asked them how they managed to deal with such a difficult constraint.
Algolia: What technical challenges have you faced building Beyond Retro?
Jascha: Since almost all Beyond Retro product is one-of-a-kind vintage there are many difficulties we faced when breaking into the e-commerce world. All items need to be photographed and classified individually. When a garment sells it immediately needs to be removed from the site.
Products being constantly added and removed from the catalogue caused us incredible headaches in terms of balancing website performance and an up to date catalogue. Our products are so varied that most search architectures were not sufficient or returned results too slowly.
To that extent, Magento’s built-in search was not effective and Magento Catalogue had performance issues with our limited computing budget. Filtering through products was slow, hindering our customers ability to find the right product for them. There had to be a better solution than the built-in Magento tools or the Manadev navigation we were using.
A: What did you do about it?
J: Our technical agenda was very clear: we needed to find a replacement for our search to better help customers find product faster. We also needed to speed up the site as the queries took too much time to process.
As were originally looking only for a search replacement, we quickly found out that many were very expensive. We discovered Algolia be a recommendation from an SEO consultant. This was originally suggested to us as just a search replacement but after experimenting with the product we decided that it would be an effective catalogue replacement as well.
Algolia’s price point is great and its speed beat all competitors we benchmarked. The free trial was also a great incentive as it gave us the ability to test the integration before deciding. Integration was straightforward and well documented and developers on Github helped us resolve small technical glitches.
A: What changes have you noticed ?
J: Speed was already a decisive factor for us to implement Algolia. It directly impacted the customer experience on our site. Algolia is insanely fast, it’s almost like it reaches into the future to get your search results. Before, we would have periods in the day when some pages would take up to 10 seconds to generate the HTML.
Now, we are averaging about 60 milliseconds with a more stable performance across the board. We also lowered the average server response time from 0.9 to 0.55 seconds. Server performance has been drastically increased and the load spikes we used to see have disappeared. Better yet, the whole implementation eventually saved us about 30% on our infrastructure costs.
There was also an unexpected benefit to implement Algolia as our search. It allowed us to turn off Magento Full Page Cache invalidation for category pages, a major improvement for Beyond Retro’s site performance.
A: How did you proceed ?
J: It took about 1 month to roll out the full Algolia catalog replacement. First, we installed Algolia’s Magento plugin and iterated with the help of the documentation and the Github community. We then applied our customized theme, tested it and rolled it out. It was a very straightforward process. We are now working on increasing the usability of our search with Synonyms and the other search features available to us.