|
|
|
|
|
|
| In 2008, August was confirmed as the prime month for website vacations |
|
| Website traffic is seasonal, and a summer break is the norm. This study compares the seasonal effect of August 2008 with the average effect of the month of August over the 6 previous years. |
| Perimeter: |
- Study conducted from August 2001 to August 2008
- Perimeter of 252,137 websites audited by XiTi
|
|
| The seasonal effect of August 2008 is equivalent to that of August 2007 and stronger than that of the month of August from 2004 to 2006. |
The website traffic analysis enabled us to observe the presence of website seasonality in terms of audience, with key periods during the year, such as the summer low or the January come-back. Thanks to non-parametric deseasonalization methods, we were able to extract these monthly seasonal effects.
The chart below shows the average monthly seasonal effects recorded on website traffic between 2002 and 2007. The months of January and March saw high seasonal effects, with respective highs of +9% and +11% in terms of website traffic. The month of August shows the most significant low periods: -14% on website traffic, followed by July.
|
|
Between 2002 and 2007, an average month of August is characterized by a low of -14% in terms of website traffic. Let’s look at August 2008.
In order to characterize the seasonality of website traffic in August 2008, we compared the seasonal effects of the month of August from 2002-2007 with that of 2008.
Thus, the seasonal effect of August 2008 is more significant than the average seasonal effect of August 2002 to 2007. It corresponds to a low of -14.8%, compared to -13.5% for the average seasonal effect of August 2002 to 2007. The seasonal effect of August 2008 is thus roughly equivalent to that of August 2007 (-15%), but more significant than the seasonal effect of August 2004 to 2006 (3 years during which the August low eased off, with a low in August 2006 less significant than the low of July 2006).
It is interesting to compare these results with those of last month (see our study:« A less pronounced seasonal downturn in July 2008 »): the low of July 2008 (-7.7%) was in fact not as important as the average seasonal effect in July 2002 to 2007 (-8.7%).
Once again, August therefore remains the reference month for web vacations.
Seasonal effects and their development can, of course, vary depending on the business area studied (e-commerce is very busy in December and January, for example). Do not hesitate to contact us if you are interested in this study being applied to your market.
|
Methodology:
Audience recorded on sites in the XiTi perimeter made it possible to detect, very early on, the existence of seasonal effects extracted by processing methods.
In this study, the series we focused on was website traffic. This concerns the evolution of audience generated by a defined number of sites. The data series successively integrates daily evolution rates, calculated on a like for like basis on all sites audited by XiTi. Evolution rates integrate the arrivals and departures of sites in the XiTi perimeter in the calculation method.
Tools adapted* for processing time series enabled us to break down this monthly series. The components are:
- The trend of the series that represents the long-term evolution of the series.
- The seasonal component representing infra-annual fluctuations, or monthly in our case, which are repeated more or less regularly from year to year. It shows growth and recession phases.
We focused on this seasonal component in this study.
* Non-parametric deseasonalization method.
|
|
|
|
|
|
|
|