Introduction

Start Utazás (www.startutazas.hu) is the leading travel portal in Hungary. Altering the more common way, Start Utazás is offering package tours from the own database. The products are organized by program (trips), the different starting dates and prices are attributes of a program. The portal normally handles 3-5.000 programs and 100-120.000 dates and prices.

Every program record – above the price information – contains around 100 separate data.

The relative big array of information makes it difficult the user to make decision, even in the case when he/she has a strong orientation on the requirements (destination, time, budget, services). Most cases he/she has no.

The good in-site search engine helps to decrease the stress of inefficient informing, but does not help in the refining of the search results and making the last mile in decision making. A human visitor is able to handle 5-9 instances in the short term memory, while in most cases there is not possibility to constrict the elements in the result list to this level. (There is no possibility/sence to offer all the available options on a search form. A normal and handleable search expression, like seashore holiday in Hurghada, Egypt, in 5 star hotel with all inclusive board on a certain date, in an applicable price during the summer, will result 30-150 results).

A recommendation engine running in the back can help through the visitor/customer in the last mile of decision making, and a good recommender system can take over the full job of the search engine. As a by-product the recommendation will increase the user experience, and will make the user recommending the services her/himself.

The introduction of the recommendation system increased the value of the visitors by 50%
Influence of the recommendation system on revenue on visitor


History      Description
  • 50% groth of the revenue on visitor >>
  • high improvement on marketing >>
  • predictable trends >>
  • avoiding to advertise non-atrctive items. >>
  • fully automatic, fully mathematical method for visitor trending >>
  • unveiling non-visible factors of visitor behaviour >>
  • immediate impact on trend change or changing of external factors >>
  • usable for most cases >>