5 Steps to Better Data Quality (Especially in CMDB)

Step 1 – Select Domain

1. Identify different data areas / domains related to service management
2. Select the most important domain
3. Define who is the owner for the whole domain (from an information architecture point of view)
4. Select most import data class as the root class for your blueprint

Note: “Most important” may vary between organizations and perspectives. Importance may be impacted by data criticality, volumes, current quality, required effort and people’s availability for example.

The Data Domain diagram shown here is an example taken from the “Practical Data Model for CMDB” or PDMC for short created by the Justin Group Oy. You can read more about the PDMC from here.

Step 2 – Define Providers

1. Define how to find the responsible person or provider for the data
   • Starting from the root record, where to find this information
   • For example, a “Managed by” reference from a Configuration Item to a User record
2. Create a scheduled audit for the Blueprint
3. Check the results and inform “Domain owner”, if mandatory data is missing
   • In this example, only “Managed by” reference is set a mandatory
   • Compliance against this blueprint should be 100%

Note: Data provider can be found “behind another record” also. For example, a data provider for an Application Service, can be defined as the Application Manager of a related Business Application. 

Target: Make sure that every record in the selected domain/class has a data provider defined. Providers can be used later to assign remediation tasks.

Step 3 – Minimum Viable Data

1. Leave the previous blueprint as is for continuously verifying that data providers are in place.
2. Expand the previous blueprint to include other minimum data related to applications.
   • For example, connection to Business Services and/or other Configuration Items.
   • Also remember the Field Setup for minimum required attributes.
3. Create a scheduled audit for the new blueprint
   • This should apply to all root records, but audit can be limited to only records that have a provider (meaning that you need to fix the provider issues first, before looking into other problems)
4. Check the results and assign remediation tasks to Managers of the Applications (= data providers)

For example, make a relationship to Business Service or Service Offering mandatory and/or add another link to Server infrastructure. 

Target: Data providers now have the responsibility to fill in the minimum information and relationships for their CIs.

Step 4 – Target Model

1. Leave the previous blueprints and audits running
2. Collect feedback and quick wins from the previous steps.
   • What has been accomplished with better data quality so far
3. Create another blueprint with more details and requirements
   • Requirements can (and should) relate to particular use cases also
4. Create scheduled audit for the new blueprint
   • Audit for target model can be limited to smaller data set
   • For example, only include Critical Business Applications
   • Further iterations of the audit can expand the scope
5. Use the audit results to evaluate readiness for the next phase

Target: Identify target model for the future and follow progress on how to get there. Set targets for data providers to increase compliance over time. Celebrate achievements and communicate quick wins!

Step 5 – Refine and repeat

1. Create new version(s) of the Target Model
2. As previously defined data and relationships are in place, more requirements can be considered as part of data modelling and audit. This can mean:
   • More related data or more detailed filters
   • More root records to be included in audits
   • Changing optional data to mandatory
   • Creating more specific models based on different use cases
3. This is an iterative step that will continue forever
4. Along with this step, one can and should also expand to other data domains.

Target: Make sure that previously achieved data compliance is maintained while constantly evaluating and redefining the target data model based on current demand and future plans.

All in all, many organizations and especially the people assigned to the task can feel a bit overwhelmed or unarmed in from of the “we need a CMDB and we need it now” project. With this kind of systematic approach, you can communicate current state, target models and progress to the different stakeholders and have a much better chances of winning their trust and support. It’s no joke, that CMDB initiatives like any other project require the management support in order to succeed.

If you want to know how DCM can help you walk these steps, please do not hesitate to contact me or one of my colleagues or simply start a free trial and give it a go. You can also book a demo session to see DCM in action and get your questions answered on the call.