- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
08-18-2019 12:12 PM
Looking for any direction or information with folks if you have been successful with my inquiry below.
How do you read/report against input(s) within a conversation message record? I could not find any ServiceNow documentation to answer my question.
The data stored under (sys_cs_message) is not very read friendly to quickly find what you are looking for.
For Example, we want to know the # of times a specific input was used and break down of the values selected within a given topic.
Example #2 List results of query strings used on a KB search within the topic.
I found a previous topic that was similar, has anyone been successful with this so far? Are you using tools such as a JSON Parser Online?
Solved! Go to Solution.
- 2,468 Views

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
08-21-2019 06:47 AM
So first, let me make sure I understand what you are trying to do: create a report based on navigation patterns and string/other variable inputs from the user during conversations in Virtual Agent? That's actually a really interesting question.
Doing a bit of digging, the Virtual Agent Dashboard comes up, and notes "administrators can monitor user engagement with the virtual agent to understand topic usage." This is great, but the functionality seems aimed at topic choices, unique users, and monthly users per channel. You've probably read over this, and want something a bit more specific based on navigation and user input, if I understand correctly.
Next, I found Virtual Agent Scripts, which allows you to store user input in variables:
Okay, this seems to be getting closer but is still mainly aimed at topic control and analytics related to topic choices.
Next, I read over the Virtual Agent Plugin docs, which once activated, gives you a wider array of choices with which to manipulate data, but still nothing in the realm of what you want to do. Your question seems like one that should be simple to answer...yet I am not seeing it.
I have no experience doing this, but I guess I don't see any reason why you couldn't store the input as variables via script and use maybe a JSON parser to get and analyze variable results and create lists/reports. The path to getting those values can be ugly, however:
This is a very interesting question, and hopefully it has an easy answer. I hope you find it.
Cheers
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
11-16-2020 04:47 PM
Not even with Predictive Intelligence?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
01-19-2021 02:00 AM
Not sure Luiz. Have to check the recent Paris release notes if this is taken care. As per my knowledge during the Peak session i got to no such self learning capabilities 1 year ago.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
05-13-2022 06:47 AM
Hi Adam,
Despite it being 3 years since this was posted it seems like there is still no easy solution, I have exactly the same requirement but cannot find anything other than manually scripting the results as Brandon suggested. In case I've missed something, did you ever find a solution for this?
Thanks
Rich

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
12-01-2022 09:36 AM
Maybe this could help you: Improve NLU Prediction
https://docs.servicenow.com/bundle/sandiego-servicenow-platform/page/administer/virtual-agent/concep...
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
12-02-2022 02:01 AM
Hi Patrick,
Thanks for the suggestion and correct me if I'm wrong but this doesn't provide us with the reporting that we (and the original poster) are looking for. The conversational analytics dashboard provides useful data on how the model is performing but it doesn't provide reporting on specific inputs within each topic. For example; if there is a user input in the topic which asks "Have you tried rebooting" we cannot easily report on the number of people that have answered 'Yes' and the number that have answered 'No'.
Regards
Rich