What about your web statistics? – 3 things every company should look out for
Recently, Apple’s Tim Cook stated during Apple’s earnings call that the iPad boasted 84% of web traffic share from tablets. Though an impressive figure, the validity of the statistics was questioned by a number of media and analysts, highlighting the difficulty many companies face when trying to determine how information from different sources compares.
Due to the varied nature of methodologies and different ways of collecting, measuring and comparing data, reaching conclusions from multiple datasets can sometimes be a challenge. Though there are no hard and fast rules. Below are my three top tips for ensuring your web statistics are as strong and reliable as possible.
1. Check the sample size
It is vital that any statistic you use comes from a study with an adequate sample size. This is necessary to ensure a good chance of detecting a statistically significant result. A study with a small or inadequate sample size will have a low probability of detecting a statistically significant result and therefore adds nothing to your overall knowledge. The above statistic, for example, came from a sample size looking at traffic from only 100,000 users. But the mobile web is continuing its climb in popularity all over the world and considering its current reach, some commentators criticised the sample as being too small. Size isn’t everything however, and there are other factors which companies should consider when using statistics.
2. Embrace geography
It sounds obvious but unless you want to analyse one particular region, sampling data from a range of regions is crucial, because different cultures use the web in astonishingly different ways.
The People’s Web Report, a recent consumer poll commissioned by Netbiscuits, found numerous cultural differences between emerging and developed markets. For example, Brazil and China are far more likely to conduct their banking via mobile than their Australian or American counterparts, and the Chinese use their mobile phones for longer time periods than the French. In a separate study, Netbiscuits found that Singapore and Australia also appear to be markedly different despite both being developed markets. The Top 10 devices in Australia contained no Android devices on the Netbiscuits platform, whereas in Singapore, any Top 10 device which wasn’t an Apple phone was always an Android.
Variances in culture can be enormous (and interesting!) and sampling web traffic data from a number of different cultures is both practical and eye-opening.
3. Be aware of timing and duration
The Chikita report that Tim Cook quoted cited data collected from one week between June 15 and June 21st of this year. Many other studies of this kind often concentrate on data from just one month or sometimes even one week.
Often it is difficult to analyse web traffic data from a year or longer, usually due to the size and amount of data. But just like there are regional variations, seasonal variations are also common. In the developed world for example, traffic often spikes in January, possibly due to a large number of electronics given as Christmas gifts. Retail sites get more traffic in the winter than the summer, and typically web traffic decreases slightly in the summer months when people spend less time with their mobile devices indoors. If possible, a good approach is to look for quarter-on-quarter and then year-on-year comparisons. This often gives an indication of web trends, even if the time frame in which the same data is taken from is relatively small.
All statistics are open to debate and discussion, but conclusions are very important. Reports that are compiled in good faith can often have their data misused or misconstrued. By making sure they are not caught out on the above points, companies can be more confident in reaching conclusions about industry trends from the statistics they choose to cite.