Discover XiTi 7.5

Web analytics
News | Contact |
Home > Internet users behaviour > 04/18/2007   

Web sites: frequentation forecasts until September 2007

The slowdown in website activity has been confirmed in 1st quarter 2007. What will the website frequentation level be between now and the start of term in September 2007? Here are our predictions...
  • Study conducted between July 2000 and March 2007
  • Range of 209,938 web sites
The slowdown in growth of website visits has been confirmed in 1st quarter 2007…

The following table shows the changes in the frequentation indicator for websites for each first quarter (January to March) for the years 2001 to 2007.

For a constant range of sites, we thus observe, for 100 visits recorded in 1st quarter 2001, a total of 234 visits in 1st quarter 2007: visits have thus increased by a factor of over 2.3 in 6 years.

However, this long-term progression, as in 2006, continues to set the trend in 1st quarter 2007. In fact, although visits to websites increased by 15.7% between January-March 2005 and January-March 2006 (and even more in the previous two years), they show “only” an increase in 4.8% between January-March 2006 and January-March 2007:


and according to our predictions, continues until the start of term in September 2007.

The following table shows the changes in the website frequentation indicator for visits recorded between July 2000 and March 2007, and our predictions for this indicator from April to September 2007.

According to our prediction model, for 100 visits recorded in September 2000, 268 visits will be recorded in September 2007.


This progress in visits to websites, observed over 7 years, has become less dramatic over the last two years:

  • In fact, a phase of sustained growth can be seen between September 2002 and September 2004: +53.1% of the website frequentation indicator,

  • While during an identical period, between September 2005 and September 2007, the increase would be only 15.9%


This trend can be clearly seen if we compare the changes in frequentation indicator for the month of September from year to year starting in 2001:

  • Change between September 2004 and September 2005: +21.4%

  • Change between September 2005 and September 2006: +7.6%

  • Change between September 2006 and September 2007: +7.7% (prediction)


This change in 2007 will confirm our analysis, published in February 2007, in which we put forward the following hypothesis in an effort to explain this phenomenon. The rate of Internet equipping is rising and numbers of Net surfers are steadily increasing, thus suggesting that the total number of website visits is increasing in parallel. In this case, the explanation for this slowing in the rate of site frequentation growth could be that the increase in the number of pages available on the net is greater than the number of Net surfers. The surfers, faced with a wider choice, are therefore becoming scattered over a larger number of sites. The sites should therefore face up to the steadily increasing competition…

Faced with this slowing in the average frequentation of web sites, which seems to be here to stay, only one expression will do: loyalty cultivation!

Visit XiTi Monitor in the next few months to follow this development.

Methodology :

The indicator used for this study, which shows the number of visits recorded by a constant range of sites, shows the changes in Internet site activity.

In the absence of official data on the number and composition of existing Internet sites, it is in fact impossible to produce an index that reflects the changes in total web consumption by surfers. For this reason, we have worked with the constant range.

We have this chosen to follow the changes in audience in terms of visits to a constant range, to which we successively apply the daily rates of actual changes calculated on the full number of sites audited by XiTi. The rates of changes are those for the XiTi range, which is continually being changed as sites are deleted and added.

The audience recorded for the sites by XiTi very quickly revealed the presence of seasonal effects, which varied in significance according to the year. Tools adapted for processing chronological series allow their various components to be assessed.

A moving average will therefore have the effect of smoothing the monthly average; it corresponds to an estimate of the global trend. This is the annual trend without the seasonal effects that characterise the different months. The peaks and troughs away from this trend correspond to the monthly effects.

The moving average, in time t, requires the series to be measured in time t and measurements around t to be taken. The data series is monthly, a moving average over 13 terms, symmetrical and centred on t, measures the annual tendency of the series. To measure the moving average for a month m, you need the measurements for the month m from months m-1 to m-6 and from months m+1 to m+6. This is the explanation for the absence of a moving average from the start of the series.
The successive stages in the processing of the monthly series X(t), which lead to the predictions made through to September 2007, are as follows:
- Estimate of annual trend per moving average over the series of data.
- Removal of monthly seasonal effects.
- Estimate, for the prediction period, of the annual trend through successive linear regressions, on the hypothesis of quasi-stability of changes in the annual trend in the short term.
- Re-inclusion of monthly seasonal effects for estimating the series over the prediction period.

En savoir plus :
© 1997-2019 AT Internet/XiTi - All rights reserved  | FR  | EN  | ES
deg_gauche deg_droit