The time dimension

Over the past few weeks we have looked at the inherent dimensional structure of management information and examined the unit, item and mode dimension in more detail.

The next fundamental dimension of MI is time.  The structure of the time dimension is inherently hierarchical (years are made up of months, which themselves are composed of weeks or days etc.).  Which particular time hierarchy to select for MI needs to be a specific choice related to the nature and requirements of your particular organisation.

The time interval “year” is relevant to MI in almost all organisations, and coincides with the cycle of filing annual accounts.  An organisation’s financial year-end may not necessarily coincide with the end of the calendar year, so the distinction between financial and calendar years needs to be made clearly.  This is particularly the case if a business is collecting annual information about its competitors, which may not share its own financial year end.  Data pertaining to annual time intervals includes strategic plans, annual budgets, current year-end forecasts and historical actuals.

Organisations may require the next level of the time hierarchy to be the half year (particularly UK listed businesses that are required to report interim results on a half yearly basis).  Half years can be made up of quarters, or for organisations that report externally on a quarterly basis the half year period could be eliminated completely, with each year being disaggregated into four quarters.

Most organisations report the main set of their management information on a monthly basis, or sometimes on the basis of a cycle of four week periods.  This latter approach creates reporting periods of the same length, except for period 13 which may contain an additional day for a regular year and two additional days when it is a leap year.

Otherwise, monthly periods might be defined as calendar months (common practice in financial services businesses), or alternatively as months made up of unequal numbers of weeks.  In the latter case, a typical quarterly pattern could be two periods of 4 weeks followed by a 5 week period, but organisations sometimes adopt other patterns.  In these cases, care must be taken to interpret such things as monthly sales trends, since periods containing larger numbers of weeks will naturally ‘spike’.  I will look at methods of working around this in a later post.

Depending on the organisation, weekly and daily information may be important for a subset of performance items, such as sales volumes, revenues, stock etc.  Retailers, e-tailers and broadcasters will also need to produce and analyse MI that pertains to hours, minutes or even seconds.  This is particularly the case for daypart analysis, used to analyse patterns of demand within each day.

Time periods may be absolute (e.g. the specific day March 21, 2014) or relative (e.g. the breakfast trading period between 7:00 and 11:00, which might be analysed over weekly or monthly cycles to identify longer term trends in consumer behaviour).

Finally, data for longer time periods such as months may be consolidated from the daily or weekly data that make them up, but in MI it is normally the case that new information is introduced at each higher level period.  For example, it is usual to prepare a 5 year plan as a set of annual values, but not as weekly or daily numbers.  Overhead cost numbers may be reported on a monthly basis, but not on a weekly or daily basis.  So you will usually need to adopt a selective approach to consolidating time-based data for specific items and modes.

Leave a Reply