Migrating an Intranet from SharePoint to ServiceNow - Conversion of Knowledge Articles

[Angela Cruden]

This post provides a rundown on the process that was used to convert >10k SharePoint articles into ServiceNow knowledge base articles. It was used during a project to move the company’s intranet from multiple SharePoint sites to the ServiceNow Service Portal.

This process was executed for an on-premise version of SharePoint where we had access to the database. The resulting knowledge article in ServiceNow was also primarily an attachment with some accompanying metadata.

Work was done prior to the load to determine the information architecture of the new intranet - this informed any new fields that were required to be added to the knowledge article table. Elements such as division, document type, and location were important due to the intended structure and presentation of the portal and how users would search for articles. Each department such as Finance, Human Resources, Safety had their own knowledge base.

Whilst the process involved some potentially risky manual work in Excel files, it worked well to facilitate the data review with business users and also the population of metadata for the knowledge articles…and we were able to easily reconcile the load. The biggest challenge was supporting the user groups to wrap their minds around moving from a folder based structure to one based on searchable metadata.

1. Perform a cleanup within SharePoint
We requested the departmental owners of each area within SharePoint to perform an initial cleanup to move any documents not required to be converted into an archive area. This would ensure the document was backed up and could be restored at a later date but would not be converted into a KB article in ServiceNow.


2. Extract data and attachments from SharePoint
Our database team wrote scripts to extract an Excel file of metadata from SharePoint and also extracted the file attachments to a file repository.


3. Format SharePoint Extract
The extracted SharePoint Excel file was formatted by the project team to include columns for the new metadata required for the creation of a ServiceNow knowledge article - this included any new fields that were being added as a result of the information architecture work done. Where the data was known - e.g. knowledge base name - it was populated by the project team. For columns where the data already existed in the system such as location, a dropdown list was configured in the spreadsheet.

The data was then split into separate files based on the knowledge base - we used the file path information from the SharePoint extract to determine the department and hence the knowledge base the article would be loaded into.


4. Business Teams Review Data
The business teams were asked to review the data and complete any missing metadata.


5. Project Team Reviews Data
Final review by the project team to ensure all metadata values were created in the system e.g. knowledge categories, user criteria for can read/can contribute etc.


6. Development Team Load Articles into ServiceNow
Our partner had a program that read the Excel file to obtain the metadata and picked up the attachments from the file repository to create the KB articles - the Excel file contained the file name of the attachment and hence provided the reference. The articles were loaded in a published state.


7. Reconciliation and Confirmation of Loaded Data
We reconciled the counts of data loaded and also a sample of the data from each knowledge base by extracting the knowledge articles and comparing back to the load file.


8. Perform a Delta Load
Once the data was loaded into the system and soon after the go-live date, we performed the same process to pick up any SharePoint documents with a created or updated date since the date of the initial extract.

Any documents that had been created in SharePoint before the initial extract but had been recently updated - we manually handled these (there were approx. 50 articles out of a total of 10k).

Once the extract was taken, SharePoint access was removed.