What is a 'website below the research threshold'?
Why are user numbers not shown for all websites? Who are the panellists? What does it mean for a publisher if their website is below the research threshold? Find out what methodology specialists from Gemius have to say on the topic.
Those who are familiar with the results of the online study measurement from gemiusAudience do some websites show number of views, but not number of users?', analysts reveal: 'Because that website was below the research threshold'. But what does this really mean?
The important role of panellists
The phrase 'website below the research threshold of the gemiusAudience study' is used for a website which was viewed by fewer than 45 panellists in the space of a month. The panellists are people who filled out a special questionnaire, as well as installing the netPanel program which was used to report their online activity.
Why forty-five and not ten, for example?
This derives from one of the principles of statistics: the larger the study sample, the lower the likelihood of error in estimating results, which are therefore closer to reality. Gemius carried out an analysis to find out how large a group of panellists must visit a particular website so as not to exceed the assumed probability of statistical error. It turned out that the study sample should consist of at least 45 panellists.
Taking care of data reliability
This number was considered the so-called research threshold level, the minimum number of panellists that needed to visit a website within a month in order to provide reliable estimates of the number of its users. Therefore, if in the space of a month the website was not visited by at least 45 panellists, Gemius did not present results for this website's performance based on number of users (this indicator is called 'real users').
Cautionary notes in blue
However, sometimes data on number of users and other features based on this indicator (such as reach or average time per user) are presented in blue. This is for cases where a website was actually visited by at least 45 panellists, but when certain features are taken into account (e.g. gender or age) it turns out that a particular socio-demographic group is not represented by a minimum of 45 people on that website. So, if a publisher wanted to find out how many men aged 15 to 24 visited their website, but the number is shown in blue in the results, it means that the data for that particular group of internet users has a higher estimation error than data for groups represented by at least 45 panellists.
The advantage of audited websites
Can Gemius give reliable data for publishers of e-services visited by too few panellists in a given month to provide user numbers for their website? Yes, as long as an auditing service is used. This involves placing a measurement script in the source code of the website which accurately indicates values for page views, time and cookies. As a result, the publisher receives data on the popularity of the website, i.e. number of cookies, views and amount of time spent on a given website.
