I saw a tweet from Gartner Analytics Cindi Howson that his was available.  I only recently looked into this.  The video below says “This new product extension by Narrative Science transforms data into natural language so that users can easily spot hidden trends in their data, and immediately share those insights with others.” – Narrative Science

Right now you can try the add-in for free by going to Narratives for BI.  There is a desktop and BI platform installation.Installation steps: https://lumira-support.narrativescience.com/hc/en-us/articles/212955046-Desktop-Installation

Installation is easy.  By the way, I was prompted to download 1.31.3, so patch 3 is available.

So I tried it myself, on some ASUG BI webcast data I manually collected for webcasts for 2016.  Sample is below:

The more measures you place, the more it will write the story for you:

You can configure it to be verbose (high) and use bullets instead of paragraphs:

This analysis measures Overall (#of webcasts), Registration, and Customer Story by SIG.

  • Reg has the highest total across all entities and Customer Story has the lowest total across all entities.
  • Overall has the highest degree of concentration among the top entities.
  • Customer Story has the most even distribution across entities.
  • Analytics stands out with the highest values for Customer Story, Overall, and Reg.

For Overall (# of webcasts):

  • Total Overall is 84 across all six entities.
  • The minimum value is 1 (BI) and the maximum is 34 (Analytics), a difference of 33.
  • The average Overall per entity is 14 and the median is 11.
  • Overall is somewhat concentrated with two of the six entities (33%) representing 62% of the total.
  • On the other hand, the four smallest entities (67% of the total) represent 38% of the total Overall.
  • Analytics accounts for over a quarter (40%) of overall Overall, and the top two entities combine for over a half (62%).

For Registrations (for webcasts):

  • Total Reg is 12,484 across all six entities.
  • The distribution ranges from 60 (BI) to 5,240 (Analytics), a difference of 5,180.
  • The average Reg per entity is 2,081 and the median is 1,882.
  • Reg is relatively concentrated with 59% of the total represented by two of the six entities(33%).
  • Put another way, the four smallest entities (67% of the total) represent 41% of the total Reg.
  • Analytics represents over a quarter (42%) of overall Reg, and the top two entities account for over a half (59%).

For Customer Story (webcasts):

  • Total Customer Story is 22 across all six entities.
  • The minimum value is 0 (BI Platform) and the maximum is 7 (Analytics).
  • The average Customer Story per entity is 3.67 and the median is 4.
  • Customer Story is somewhat concentrated with three of the six entities (50%) representing 82% of the total.
  • On the other hand, the three smallest entities (50% of the total) represent 18% of the total Customer Story.
  • Analytics accounts for over a quarter (32%) of overall Customer Story, and the top two entities represent over a half (59%)

I also used to to analyze data from ASUG Annual Conference 2016:

The only limitation I found is it does not write a narrative when you have 2 dimensions in the story.  But it is worth looking at and it will give you insights into your data.  Does this mean no more written reports against analytical data in the future?  What type of tools will business analysts and accountants use in the future – natural language processing to write reports?

To report this post you need to login first.


You must be Logged on to comment or reply to a post.

  1. Katie Hughes

    Hi Tammy! I am the product manager for this Narratives extension for SAP Lumira. Thanks for your post. I noticed your comment regarding the 2 dimension limitation. The product is currently designed to tell a multi-dimension story (2 dimensions), as long as there is only one measure. If this isn’t the case for you- let me know so we can get it figured out! You can email me directly at khughes@narrativescience.com


Leave a Reply