It's raining data

[britta]

Third in my project management blog series.

The classic song "It's Raining Men" says that for the first time in history it will be raining men. When it comes to data, it's been raining way before technology, which was supposed to fix the data problem. But it hasn't. In fact, it's made the proliferation of data an even bigger challenge.

So, before you start a ServiceNow implementation, it's very important to think about all the data needs you’ll have, as well as the time and resources needed to import or integrate the data, and how to make sure the data is good. In this blog, I'll be interviewing Mary Vanatta, one of ServiceNow's resident ITOM and data experts to provide guidance.

QUESTION: What types of data do customers need to collect and organize prior to implementing ServiceNow?

MARY: Here's a comprehensive list of data types to be considered.

o Core data for organizations that support business operations and capabilities for employees, and/or customers who will interact with ServiceNow. This includes user roles, access permissions, etc.
o Company data such as department, location, business units, etc.
o Product and product models -- Software, hardware, applications and services and their specific versions If you need to explain the items you purchase or create to operate your business.
o Workflow and process data — Gather current process documentation to determine future state processes
o Integration data — Active Directory, third-party monitoring tools, and HR systems are some common integrations
o Security operations — User roles, ACLS, etc. to ensure access is controlled and secured
o SLAs, OLAs —These define and organize work, response times, and escalation procedures
o Historical data — A decision needs to be made on whether to migrate current open incidents and requests or to close them in the current system, and what the plan is to archive the historical data.
o Service catalog — Define and organize (often through categorization) to import only Service catalog items that are current and operational
o Knowledge Management — Review, organize, and import current, relevant KB articles

QUESTION: What are some examples of the data sources that would be in scope?

MARY: There is almost always Active Directory for user data, roles, locations, etc. There are also IT tools for monitoring and access, like LogicMonitor and SailPoint, that provide asset and configuration item and asset management data. Lastly, sometimes an HR platform is also in play like Workday.

QUESTION: What does “organize data” actually mean?
MARY: In the simplest of explanations, it means aligning to CSDM(common service data model). This includes categorizing user, foundational, process, and referential data—and putting it in the right tables. Basically, following the CSDM data model.

QUESTION: Why is data cleanup, categorization, normalization, etc. so important?

MARY: Three major reasons: deduplication, normalization, accuracy. Deduplication is critical because if you leave duplicated data in the system it can lead to issues such as outages or longer mean time to repair (MTTR). Normalization ensures consistent naming for data throughout the platform such as vendor names. For example, Microsoft is not MS, MicroSoft, MS Microsoft, etc..,which means reporting will be accurate. This leads me to the final reason for making sure data is clean: it ensures the accuracy of how an organization is operating.

QUESTION: What are some of the other consequences of not having correct data prior to an implementation?

MARY: You know the old saying, “Garbage in, garbage out.” The obvious ramification is not getting end-to-end visibility. More critically, it can cause an organization to have to re-deploy ServiceNow. The business needs good data to from the operations and services teams to make decisions on key business initiatives.

QUESTION: How does something like this play out in the real world?

MARY: One example is a company where I helped them set up Discovery and got it running, but they stopped it because they were so concerned that scanning was going to kill their operations. The company is depending on people to manually update company data, and they have poor CMDB health. This leads to the problems they have with their incident management, where incidents take more time to resolve because their monitoring tools are shoving incidents directly into incident management without using the filter capabilities of Event Management.

QUESTION: How does an organization recover from having brought in bad data?

MARY: First, they recover by recognizing that they're not getting what they need out of reports.

Typically, they’ll have to put time and resources into remediation because it’s hard to remediate data while it’s in use. Before fixing a data issue, they need to understand where it came from—an integration, bad manually-added data, etc. I say they need to quarantine that bad data, if possible. Then they need to use data validation tools that can help determine which data is good.

However, they may still need to do some data enrichment to fill in what data is missing. For example, when setting up the CMDB, they might not have paid attention to product models and will need to add them now.

The way organizations get to the needed discipline for data management is through governance. Organizations need to document processes to clean up bad data and to implement auditing and monitoring of data going forward. A key factor that is often missed is user training and awareness—good organizational change management. Users need to be educated on the effects of bad data as well as what's in it for them to adhere to processes to keep data clean. Lastly, organizations need to have the ability to verify accuracy and validity of data.

Thank you, Mary, for all that great information! If you want to learn more about how to manage data in ServiceNow, visit our community product hub, Data Foundations and check out Platform Live on ServiceNow events.