Citations are references to a publication from other publications in a given database. The database’s coverage thus has an impact on whether a citation counts.
Citations are increasingly being used to shed light on the use and dissemination of research (impact), and a number of indicator measures have been developed based on citations, for example.
Standards for the calculation of these metrics can be found in the Snowball Metrics recipe book. Citations are also an element in the calculation of the h-index and a range of metrics of journal impact.
You can only make citation analyses in databases and search engines that index the publications’ reference lists.
The traditional citation indexes are Web of Science and Scopus, which are competing products with a considerable overlap. You can use them separately, but often it will be appropriate to carry out searches in both in order to assess which database is best in a given situation.
You should be aware that the databases primarily cover journal articles, and that there is limited coverage of humanities and social science research, for example.
Google Scholar also includes citations and has a broader coverage of publication type, topic and language. Google Scholar's interface can make it more difficult to conduct systematic searches in the same manner one can in the traditional citation indexes.
In the normal publishing cycle, citations typically appear in academic publications referencing previous publications.
New articles, that refer to earlier articles, must be published and included in the citation databases before the citations can be counted. In some subject areas this occurs faster than in others. However, it would ordinally be expected to take 3-4 years from an article being published, before it makes sense to count citations of the publication.
In addition, there are varying academic traditions for how frequently citations are made. In some areas, for example, it is impressive to see 10 citations per publication, while in others you might expect 50. Therefore normalised citations are usually calculated, e.g. Field-Weighted Citation impact.
Self-citations (citations of earlier articles within the same journal) are a common part of ordinary research communication. In an evaluation context, self-citations are often criticised because authors themselves can affect the amount of self-citations. Thus, some subjects omit self-citations, while others do not.
The larger the entity being analysed, the less impact the inclusion of self-citations will have.