LoveHolidays Use Case

loveholidays increases conversion rate by 11% with Fastly

Faster performance drives conversions

For loveholidays, faster site performance directly translates to higher conversion rates. The faster the site loads, the more consumers complete their transactions.

Led by its Head of Engineering, David Annez, the loveholidays team began work on the future of their platform and strategy in 2020. This meant implementing Fastly’s edge cloud, which seamlessly handles device detection, image optimisation, edge caching, TLS and GeoIP detection and enables real-time streaming of logs to BigQuery. This has simplified the company’s infrastructure and driven improvements in performance, reliability, and cost savings. The move has also put loveholidays in a stronger position for the return of international travel by providing the stability and scalability needed to meet rising demand.

“If you care about customer experience and performance, there’s no other place you should look for content delivery.”

A rapid 51% improvement in First Contentful Paint

The project quickly began delivering benefits. In terms of page load speed, the company saw an immediate 51% boost in First Contentful Paint. After making continuous improvements to both the underlying loveholidays platform and the Fastly configuration, this has subsequently improved by a further 10%. These changes contributed towards a conversion rate improvement from their landing pages of 11% in approximately six weeks.

Source CrUX data

Reducing time-to-first-byte by 80%, from 100 milliseconds to 20 milliseconds

With Fastly’s edge cloud network, David and his team can cache even more on the edge. The team saw big improvements on time-to-first-byte (TTFB) – bringing it from 80-100 milliseconds to 10-20 milliseconds.

The flexibility provided by Fastly's edge cloud network, coupled with its integration with Terraform, supports a culture of continuous innovation and development.

“Fastly has given us an edge on a lot of our competitors, primarily, because simplified infrastructure, stability and scale mean we can always give the customer the fastest experience whatever the device.”

Enabling real-time observability and learning

As the team looks to improve conversion rates, they need the right capabilities to collect and analyse performance data. Following the migration onto the Fastly network, the team has been streaming logs into BiqQuery to see how real-time changes to the site are impacting performance.

With a high degree of scalability, flexibility, and reliability, this integration will be able to grow effectively as demand for travel returns to pre-pandemic levels over the coming years.

“Having access to the analysis and real-time log streaming network data in your platform without the need for huge hardware or software investments is a data scientist’s dream come true.”

Sparking customer imaginations with image optimisation

Image optimisation is another powerful Fastly tool that helps loveholidays increase engagement and boost conversions. Loveholidays has used Fastly to optimise its hotel images since 2014, removing the overhead associated with having to resize images on the fly.

Hotel images create an initial hook for potential customers, and images that are slow to load may make visitors abandon a page. Slow-loading images also negatively impact a site’s SEO performance. With Fastly, loveholidays can improve its performance and continue to be indexed by Google as a fast site.

Simplifying workflows

Loveholidays’ previous architecture was made up of a number of different legacy systems, which added an unnecessary level of instability and lack of scalability.

Thanks to Fastly’s integration with Terraform, the team has complete code visibility and can monitor changes made to it. Now, the loveholidays team can deploy changes through Terraform to Fastly with confidence in seconds.

Pioneering A/B testing on the Compute serverless compute environment

With performance so closely tied to driving improvements in the site’s conversion rates, Fastly lets loveholidays run tests at scale. By testing and learning in real time, loveholidays makes its customer experience more customised and personalised — letting them provide better experience to travel-hungry consumers.

Conducting A/B testing effectively without impacting SEO rankings is an important part of the performance-first culture at loveholidays. The loveholidays team has rolled out a range of innovations in its experimentation platform and processes using the Compute serverless compute environment in Rust.

This lets the team make decisions about what experiments individual customers are involved in at the edge, allowing the team to A/B test at scale with virtually no impact to performance. And as Fastly continues to update Compute, loveholidays will be able to implement multi-armed bandit testing and move all the company’s experiment configurations to the edge.

“I think Compute is going to help us drive more personalisation such as recommendation engines and more targeted results to customers.”