How Much Time Should It Take to Review a Data Model?

I recently asked the folks who receive my monthly newsletter how they would answer the question, “How long will the modeling take?” The consensus from this group was that in order to answer this question, we would require answers to several other questions first including:

• “What has already been done in the way of requirements gathering prior to me starting this modeling effort?”
• “How do you define data modeling?”(This is not a silly question by the way because it clarifies the data model deliverables, such as will the data modeler be responsible for producing mapping documents from the source system?)
• “How will my data modeling activities be constrained by tight project deadlines?” (This last question is a tactful way of asking “Do you already have in mind how much time you are giving me to complete the modeling deliverables?”)

So some questions such as “How long will the modeling take?” require answers to other questions first. It is rare that there is a concrete and immediate response to this question such as “The modeling effort will take approximately 23 hours and 35 minutes.”

The irony though is that despite asking all of our important questions, if someone did tell us we only had 23 hours and 35 minutes to complete the model, we would most likely be able to finish the model within this timeframe. We would complete as much as we could, relying heavily on generic constructs such as Party and Event, and generously applying the 80/20 rule wherever we can.

This same timeboxing principle holds true for reviewing data models. “How much time will it take to review a data model?”

I get asked this question quite often. I usually start by scratching my chin, looking up at a light, a well-paused period of silence, and then the two well-rehearsed word response every consultant needs to perfect, “It depends.” Sometimes I add a third word: “Well, it depends.” (Note that my consulting rates usually go up when I add this third word J)

All of these factors come into play in determining how long a data model review takes:

1. What is the size of the data model I will be reviewing?
2. Will I have ready access to key business and IT folks as well as relevant documentation such as functional requirements documents?
3. What stage is the modeling effort in? (This question tells me whether feedback provided will actually be implemented. For example, if the project is one week away from go-live and it is the first time I am seeing the model, I would question the value the review provides and therefore the level of review required.)
4. Is the model in an easily accessible format for me to review?
5. How many of the data model checks can I automate?

Often however, just like building the model, reviewing a data model is often timeboxed. That is, there is one question that trumps all the ones above: “How much time and money did you budget for the review?” I have reviewed data models in two hours and also in two weeks, depending on to a large extent on how much time I am given to review a data model.

Until the next blog…