My first Publication Agile-Data-Warehouse-Design-eBook | Page 120

Modeling Business Processes
99
While dimensions are frequently shared across many business processes, facts are
typically specific to a single process or event. However, they can be used to create
conformed measures if they have compatible calculation methods and common
units of measure that allow totaling and comparison across processes; for example,
if sales revenue and support revenue are both pre-tax dollar figures they can be
combined to produce region totals. Conformed
Conforming data is not so much a technical challenge, as a political one, requiring
consensus on data definitions across many departments within an organization as
well as operational systems. By modelstorming with stakeholders you highlight the
value of conformed dimensions to the very people who can make them happen.
Modeling multiple events by example, as BEAM ✲ encourages you to do, quickly
reveals inconsistencies that would otherwise thwart conformance. Stakeholders will
work to address these inconsistencies and conform dimensions when they see the
potential business value they provide. Conforming data
Homonyms are data terms with the same name but different meanings. They are
the opposite of conformed dimensions and attributes. For example, both Pome-
granate’s Sales and Finance departments use the term “Customer Type” but Sales
has five types of customer and Finance only three. If stakeholders cannot agree on
a conformed customer type then you would have to define two uniquely named
details: SALES CUSTOMER TYPE and FINANCE CUSTOMER TYPE. However,
taking this approach for every homonym perpetuates incompatible reporting and
weakens the analytical power of the data warehouse. Perhaps by discovering this
problem through modelstorming examples, Sales and Finance stakeholders could
agree on a new conformed version of Customer Type with four descriptive values. Homonyms are non-
Synonyms are data terms with the same meaning but different names. Organiza-
tions will often use different names across different departments/ business proc-
esses for what could be the same conformed dimension or attribute. For example,
an insurance company might use the terms Customer Enrollee, Subscriber, Policy
Holder and Claimant interchangeably, while a pharmaceutical company may refer
to the same person as a Physician, Doctor, Healthcare Provider or Practitioner. Synonyms are
The value of modeling with examples, to help define conformed dimensions,
cannot be overstated. Stakeholders often think they fully understand the meaning
of their data terms, until ambiguities and differences of opinion are quickly ex-
posed when they provide examples to their peers.
measures rely on
compatible facts
with common units
of measure
is a political
challenge. BEAM✲
tackles this by
modeling with
stakeholders who
can make it happen
conformed data
terms with the same
name but different
meanings
conformable data
terms with the same
meaning but
different names