Slack - from a Data Science point of view


Conversational data is incredibly rich and valuable, but it’s also unstructured and often unruly. sifting through mountains of disjointed data to glean insights is a frustrating and time-consuming process.

This is where natural language processing (NLP) comes in. NLP is a branch of artificial intelligence that deals with the interpretation and manipulation of human language.


Slack is a popular workplace messaging app that allows employees to communicate with each other in real-time. Many businesses are now using Slack to increase productivity and collaboration. Researchers in the field of NLP are interested in using Slack to improve communication among employees.


Slack is a communication platform that is quickly becoming the de facto standard for workplace collaboration. And, as more and more organizations adopt Slack, there is an increasing demand for tools that can help make sense of all the data being generated.


TheGist has developed a suite of NLP tools that can be used to analyze Slack data. We have used these tools to help analyze millions of Slack messages, and we have found them to be incredibly valuable in extracting insights from conversational data.


The academic community offers a few public datasets for Slack conversations, but it's rare. Hence, the datasets we started with include a mixture of public datasets, Slack communities, tech associations, and partner data sets.


Here are some examples of the kinds of insights that can be gleaned from Slack data using NLP:

  1. What are the most popular Slack channels?

  2. What are the most common questions being asked?

  3. What are the most common emotions expressed on Slack?

  4. Who are the most influential members of a Slack community?

  5. What are the most active times for discussion in a Slack community?

  6. How does the use of different words affect the way people communicate on Slack?

  7. Which topics are being discussed the most?

  8. Are there any warnings or red flags in the data?

These are just a few examples of insights that can be extracted from Slack data using NLP. The possibilities are virtually endless, and we are continuing to explore new ways to extract value from conversational data.



What are the most popular Slack channels?

Some of the most popular ones include:

  • general

  • random

  • announcements

  • support

  • development

  • design

  • marketing

  • product

  • sales

  • success



What are the most common questions being asked?

The most common questions being asked on Slack are related to technical support, such as how to use Slack or how to fix a technical issue. Other common questions include how to find specific information or how to contact a specific person.

Also, besides just asking “why” to get some explanation

or inquire about how to find the PM 😉

The most popular questions focus more on getting feedback, help, and learning from our peers.

Here are some examples…

Feedback

What are your thoughts? Does this make sense? What am I missing?

Help

Any suggestions? Any advice?

Learn

What are the best practices? How would you do it?


What are the most common reactions to messages?

This one caught me by surprise 😅 `eyes` are the most used reaction, which is probably due to the default setting on Slack.


Here are the 10 most popular: 👀 🙌 👍 🔥 ❤️ 💯 👏 🎉 🚀 ➕


And the least used reactions (without counting the costume ones) and without the flags are reactions that we usually label as inappropriate for the workplace: 🤥 😡 😢


Who are the most influential members of a Slack community?

There is no definitive answer to this question as it varies from community to community. However, some factors that could influence a member's level of influence include the frequency and quality of their contributions, their ability to engage other members in meaningful conversations, and their willingness to help others.

Due to the fact influence in the workplace is something that is hard to define, and even more, to measure, let’s check which job titles were most active.


🥇 Bots

🥈 Product

🥉 Developers



What are the most active times for discussion in a Slack community?

As expected, while the most active hours for discussion are during business hours(9am-5pm) there are significantly more active people between 10am-2pm.

An interesting behavior happens during lunch when people are still very responsive.



I imagine it must be a bit difficult to get through lunchtime on a Wednesday in September.


How does the use of different words affect the way people communicate on Slack?

The use of different words on Slack can affect the way people communicate in a few ways.

First, the use of different words can change the tone of the conversation. For example, using informal words can make the conversation seem more casual, while using formal words can make it seem more serious.


Additionally, using different words can change the level of detail in the conversation. For example, using specific words can provide more information than using general words.

Finally, the use of different words can affect the sentiment of a conversation on Slack by making the conversation more positive or negative.


For example, if two people are having a conversation and one person uses positive words, the conversation will be more positive. However, if one person uses negative words, the conversation will be more negative.

From our research, there is a very strong sentiment change when using some phrases of words.


The sentiment of the conversation will change 50% of the time from positive to negative when twitter is mentioned, and sometimes when money is mentioned.


The sentiment of the conversation will change very often from negative to positive when we become truthful or thankful. Comments like super helpful and make sense go a long way toward changing the tone of the conversations.


Which topics are being discussed the most?

There is no definitive answer to this question as it largely depends on the specific team or organization using Slack and the types of conversations they are having. However, some of the most popular topics discussed in Slack conversations may include work-related matters, current events, and general idle chatter.


Engineering - development, code reviews, debugging, troubleshooting.

Product - productivity, development, disagreement, squads, demands.

Growth - monitoring, behaviors, sales, marketing, advertising.

Management - morale, investments, scheduling.


Are there any warnings or red flags in the data?

We should be aware that the dataset may be incomplete or inaccurate, that it may be subject to change or manipulation,

For example, if a user edits a message, it can change the outcome of a simple summary.


Overall, NLP research can help businesses to improve communication among employees.

This research can be used to identify trends and patterns in employee communication.

This information can be used to improve the design of Slack and to make sure that employees are using the app in the most effective way.


To be honest, most of this blog was generated using our internal AI tools. Which some will publish in the future.


So If you're interested in learning more about NLP in the workplace, or if you're looking for a tool to help you make sense of your text data, we invite you to check out theGist.

Cheers,


Sahar Millis

Lead Data Scientist, theGist