Skip to end of metadata
Go to start of metadata


There is a grouping of the Level 1 processes GSBPM users will understand, but which does not form part of the model. GSBPM processes 4 through 7 (Collect, Process, Analyze, Disseminate) form what we might call the “work” process of the statistical office. Processes 1 through 3 and 9, in the order 9, 1, 2, and 3 (Evaluate, Specify Needs, Design, Build) form the “change” process. Process 8 (Archive) possesses sub-processes with characteristics of both work and change.

GSBPM contains a convenient fiction that the Evaluate process happens following each instance of a statistical process. The model paper hints at the reality, where “ some cases, particularly for regular and well established statistical business processes, evaluation may not be formally carried out for each iteration. In such cases, this phase can be seen as providing the decision as to whether the next iteration should start from phase 1 (Specify Needs) or from some later phase (often phase 4 (Collect)).”

We think it more useful to consider the work process and change process as being fundamentally different. The work process produces statistics, directly creating value for society. The change process improves or degrades work processes, indirectly creating or destroying value by changing the performance and cost of the work process (or of the stock of work processes if adding or removing a work process). We realize value from the change process only when work processes improve.

Step one for the industrialization of official statistics is to use the change process to automate the work process. Our view is that it would be useful for the GSBPM model to explicitly support this.

For more context see our METIS paper.










  • No labels


  1. Note: Wim raises a similar point 

    Need for clear separation of design (2, 3) and run (4-8)

  2. Discussion 16/9:

    Gary - To keep ourselves our dynamic, shouldn't we be coming back to the needs each cycle? 

    Eden - Specify needs phase may now be a consideration something much broader than just a survey. We want to load the front with more research 

    Wim - Support Gareth's proposal. Main distinction is the 4 - 8 have a strict rhythm but the other phases can be triggered by all sorts of things. Would like this distinction to be less hidden in the model.

    Steve - Could do more to make the distinction more obvious. Different types of the activities have different rhythms (monthly survey vs 10 year census). Distinction between regular activities vs ad hoc or occasional activities. 

    Gareth - First targets for standardisation are often phases 4 - 7. Maybe we could explain that in the document. Automation is a secondary goal for me. 

    Mats - a fundamental issue. Sees a difference between 123 and others. See the models that represent GSBPM as a continuous cycle is attractive. We should think about how we show this in diagrams or if not in the documentations.

    Wim - Process 8 could fit in either half. Would be tempted to reorganise it.

    Gareth - The Information objects in GSIM that go through 4 - 7 are quite different to other phases. The 4 - 7 objects are rather concrete (data etc) 

    Eden - Process 8, looking forward to the future when you are metadata driven, with automatic archiving rules, it is not clear what is left in process 8.

    Gareth - in response to query on work, people doing the design work are not always the people doing the production work. 

    Gary - Process 8, role in the future. Well,  people chose to implement different phases. Is that what we should continue to do? Or should we think about completely remove a phase if no one is really using it? We should raise those issues as a separate discussion topic.


  3. Note: Comments from Statistics Finland

    Separate on-line (1-3, 8) and off-line process phases (4-7) more clearly.

  4. Istat

    1. Comment on “Issue 1: Distinction between "work" processes and "change" processes

    We are not convinced about this proposal of change. One of GSBPM main characteristic (and reason for its success) is its extreme flexibility, and reading the “Instruction” on how to use GSBPM you are allowed to use the phases and the subprocesses in a different order or to skip some of them or to apply some of them only in some instance of the process. Thus this different “View” is not strictly necessary, and, from the point of view of “quality”, it seems restrictive to have phase 9. Evaluation tied to the “change” side of the process, while it should be recommended to perform “9.Evaluation” at each instance of the process. Obviously level 1 processes 1-3 can be performed only if a change is deemed necessary.

  5. Statistics New Zealand response to GSBPM v4.0

    The processes Collect, Process, Analyse and Disseminate are production processes i.e. these four processes constitute the activities involved in the regular production cycle of an output. The processes Need, Design and Build are transformational rather than part of the regular production cycle. Is it possible to highlight this difference in some way?

  6. Statistics Sweden

    Modify the descriptive texts for the preparatory phases 1–3 to make it clear that they are important for making continuous improvements, also with input from phase 9. They are used in each production round (more or less), for instance by modifying the allocation of resources.

    Modify the texts for the preparatory phases 1–3 to make it clear that methods and tools are re-used, rather than developed, also in new designs and re-designs.

  7. Both the rhythm and the expertise involved is different.

    I do not need a revision, it would be sufficient to highlight the point by applying different colours and by an explanatory text explaining the point.