Let’s get personal…

Personalised Content in Newsletter for Skyscanner Growth

By Iain Pritchard

TL;DR By including content in the newsletter that related to the traveller’s most recent searches, we saw our Newsletter activation rate increase by 65%.

As part of an internal push towards personalisation of a user’s experience when interacting with Skyscanner, a hackathon-style competition was held to generate interest and ideas. One of the ideas floated was to combine the behavioural insights we stored in the State Machine with the huge bank of reusable content that populates our content pages (like this one about sunny Barbados), and sprinkle the result all over our weekly newsletters that are sent to millions of people a week. The result? A newsletter containing content specific to the lucky recipient, delighting said traveller and generally making us look great.

Working across Teams at Skyscanner is common and enabled through our Squads and Tribes model. From top left: Andy McCready, Chris McCluskey, Iain Pritchard, Nina Neissl, Emily Stewart, Jonathan Sepulchre, Naiba De Bona

What’s in a newsletter?

When we typically send a newsletter to a country, we retrieve the most recently published articles for that market from our bank of content. This certainly works and has done since we automated the newsletter-sending process a couple of years ago. However, regardless of a traveller’s interests or behaviour on the site, each user in the market received the exact same content.

Example of a Skyscanner newsletter sent to our UK subscribers

Data-driven Content Curation

Our idea was to make use of data we hold about a recipient’s previous interactions with Skyscanner, which are stored within the State Machine. When a user performs a search for flights, say to Munich, a transport search event is added to their timeline containing the search parameters. We decided that this is a great indicator of the user’s intent for travelling to Munich at some point and the user could be interested in content related to it.

Example data stored on my user timeline. This event shows that I made a search for flights from Glasgow to Munich

Our news pages are populated with content that is stored within a queryable internal system. The article content is tagged with place IDs that are relevant to the article. For example, an article listing the best things to do in Munich might be tagged with the ID for Munich, the ID for Germany and the ID for Europe. This is a great source for us to use here, since we can request content relevant to a destination — data we have access to thanks to the State Machine!

In order to improve its visibility, we decided to put the content at the very top of the newsletter. Our automated newsletter system uses the lead piece in a newsletter as the email subject line, which is highly influential of whether a user opens the email in the first place.

With the data sources and content placement sorted, we had one other thing to sort out: who are we going to experiment on? We got in touch with the squad that own growth in the UK and Ireland markets about our plans. They kindly agreed to let us run this experiment in the UK market, our largest subscriber base.

Left: Original newsletter to be sent to subscribers. Right: newsletter personalised based on my search history

You can see the original UK newsletter and the more personalised newsletter above. I had been searching for flights to Munich within the last 7 days, so our systems sourced some relevant content just for me. Unfortunately, we couldn’t find a suitable candidate for Munich in particular. In this case, it choose the most closely related tagged content available, a piece about Germany.

How did it fare? (Spoilers: quite well with a caveat)

Results of the experiment. We ran these through a statistical significance calculator and they came out as being statistically significant

22.2% of users in the control group opened the email, whereas 22.7% of users in the personalised cohort opened their email. As previously mentioned, open rate is relevant to us as the first article in the email dictates the subject line, which is a highly influential factor regarding open rate.

4% of the control group users clicked on the first article in their email. In comparison, 6.6% of users in the personalised cohort clicked on their first article that was tailored to their searched destination. That’s an increase of 65%.

While these are great results, but they come with a caveat. Comparing interest in different pieces of content is difficult, as there are many factors at play such as the quality of the content, quality of the imagery, and even relevancy to users in ways we don’t and cannot know. However, these results are certainly illuminating, as it shows a strong appetite for relevant content in our weekly newsletters.

The future is bright (and hopefully, highly personalised)

Not only are these results really promising for include personalised content in our weekly newsletter, this is a really great result out of a hackathon like this. Often, projects bring about a learning but don’t end up in the product. After speaking with the relevant squads, we’re planning to roll this out further.

Sign up for our Growth newsletter waiting list as we’ll be sending more Growth Hacks, updates and job vacancies right to your inbox soon.

Work with us

Take a leap and look at our current Skyscanner Marketing, Design and Developer roles within our Growth Tribe across our different offices.

Want to grow with us? We have great perks as well as producing a Growth environment. We’re currently recruiting for the following roles*:

Come and work with us at Skyscanner

About the Author

Hi, my name is Iain Pritchard — thanks for reading my post! I’m a Software Engineer at Skyscanner, working in the Central Growth Tribe from our fantastic Glasgow office. Skyscanner gives us the freedom and opportunity to experiment with the products we’re building. My favourite thing about working at Skyscanner is getting to work with world-class engineers to build and refine the communications we send to our millions of users every day.

Iain Pritchard, Skyscanner Growth