The IT Industry Forum Benchmark

By Peritus

Today is publishing our IT Industry Forum Benchmark in the first of a series of blogs. 

Our company mission is to use machine learning to both create more IT experts and help them be more effective with our AI assistants. To achieve this mission we have built a curated data set representing all known IT industry knowledge in a cloud based knowledge graph that is continuously updated. This data set includes technical conversations in vendor and open source forums, StackOverflow, and GitHub. We also ingest and index all related published content and profile forum contributors across all industry discussions.

This data set provides an industry wide view spanning people, products, and problems  to develop analytic insights on how machine learning can transform forums from a “best-effort” technical support channel to a best-in-class support experience. 

Our current data set includes over 12 million unique questions and responses from forums for building blocks for multi-cloud IT infrastructure including enterprise networking, software defined data center technologies, edge compute, cloud-first open source, and public cloud services. We are continuing to expand the scope of our data set, but we now have critical mass to see consistent trends.

Here are three top-level findings:

  1. 20% of forum contributors resolve >80% of forum posts
    • This is expertise is highly concentrated in the top 5% of contributors 
  2. 81% of posts are resolved after 24 hours or never resolved
    • These are posts that have highest potential of becoming support cases
  3. Answers from the top rated forum contributors are accepted 10-15% of the time
    • Even the best and most motivated experts could use some help

In our upcoming blog posts, we will publish more detailed metrics and trends across people, product, and problem dimensions including comparisons by industry segment and contrasts between open source and vendor technology.