Predicting Booking Behavior for Camping Holidays
Vacanceselect offers the most beautiful campsites and holiday parks throughout Europe. 2016 was a year in which we succeeded in greatly enhancing the user experience from the first search query up to the actual booking. Reaching the right person at the right time with the right message. Booking your dream camping holiday with a simple click. That’s what we want to offer to users. We believe the secret lies in serving relevant ads in combination with personal landing pages. And due to automation and machine learning this was also possible in a cost effective way. Together with Vacanceselect we started advertising for about 890 campsites and vacation parks in Europe. Not only did we include Google and Bing, it also spread over 12 European countries for 3 brands (Vacanceselect, Campingselection and Selectcamp).
31% increase in sales with only 2% more media budget
Booking your dream camping holiday with a simple click. That’s what we want to offer our users. We believe the secret lies in serving relevant ads in combination with personal landing pages, based on machine learning. These give the users the exact information they’re looking for, if it’s a campsite in France with a child-friendly pool, or an adventurous camping trip in Sweden. With the help of this campaign, Vacanceselect wants to grow 10% per year in revenue with a steady media budget.
The primary target group are parents with young kids and holiday plans in Europe. But that
sounds easier than it is, as:
- Multilingual: The account contains ads for 12 countries and 9 languages, all with different holiday seasons
- Dynamic offers: The availability of the campsites varies from week to week and the availability decreases when the season comes closer. We call this stock-based marketing.
- Scope: the accounts are so complex and big that we even reached the limitations of AdWords (> 5 million entities)
- Influence of booking seasons: For every season, different ad statements apply which adjust to the sentiment of the moment and to eventual running season specials.
This complexity makes it a big challenge to reach the right person at the right time with the right message.
So, how did we reach our goals?
Strategy and approach
Automation: It is impossible to manage millions of ads by hand.
50% of the campaigns include automatic created search terms and ads for all 890 campsites for all languages and countries. Campaigns in Google and Bing were synchronised simultaneously. Unavailable campsites are being paused automatically and campsites with high availability will get higher bids & positions in the SERPs through advanced business rules.
Relevance: We want to address new visitors, returning visitors and existing customers in different ways with extremely relevant ad copy and (dynamic) landing pages.
Last year all campaigns got split up for new visitors, returning visitors and existing customers. No matter if it’s about the last minute season or the early booking season, due to dynamic landing pages, the user gets exactly the information he needs. On top of that all pages transform to the users individual query, not just by changing the headings on the page, but by displaying photos on the landing page that match the query. Even the order of the photos in the slider is changed to represent the likelihood the photo matches the users query based on machine learning (Google Cloud Vision API). We realised an accuracy of 93% of matching the photo to the users query.
Flexibility: We adjust millions of ads as soon as a new season begins, per country.
All ads (millions!) are equipped with a label which includes information about the season and the features. This way we are able to shift all statements in the ads – in all accounts – for the new season in only a few days. Whereby we remain capable of optimising towards relevance by using A/B tests.
Voice search: After research on how people use voice search within the travel industry, we found that most people use longer and more questions in their queries. We anticipated on this by integrating these search terms as actual keywords within our campaigns.
A total increase in sales of 30.96% with only 1.90% increase in media budget, 3 times more than the set objective. (October 2015 till September 2016). In addition to that more than 360 working hours were saved by automation on a yearly basis. Logically Vacanceselect was extremely happy with the results. Their revenue directly attributed to AdWords has risen from €16.6mln (2015) to €21.8mln (2016). Our dynamic landing pages significantly outperformed the old static pages, up to 60%.