Measure Company Culture With Machine Learning – COTM

Measure company culture with machine learning? Sure…why not 😉

For our January Bay Area Ventures Conversation of the Month, a monthly conversation on Bay Area Ventures on SiriusXM with Donald Landwirth, we took a slightly different approach to measuring company culture.

Bay Area Ventures airs Mondays at 4:00pm Pacific Time / 7:00pm Eastern Time, on SiriusXM’s channel 111, Business Radio Powered by the Wharton School and replays throughout the week.

Almost every one of the guests on Bay Area Ventures talks about the importance of their company’s culture in terms of focusing their teams on the company’s vision or being the compelling force that drives their team to look forward to coming into work every morning.

It’s a recurring theme and hugely important aspect of the most successful companies we study.

As usual, the responses were great and we gained a lot of insight into the commonalities that people share regarding what work means to the individual.

January 2017 – Conversation of the Month

The question and prompt to the audience:

“Share a picture or video that best describes your company’s culture.” We also asked for a brief description of what their picture or video was trying to say.

Gathering the results and analyzing the data

Prompted by the aforementioned question and instructions, we used GlimpzIt to gather and analyze 1,167 total data-points.

The data-points consist of human responses from both:

1. Respondents who took a picture or video and explained their answers with a few lines of text. We call these “Glimpzes.”

2. Evaluators who provided a thumbs up or down, along with a brief text explanation of their own, in response to how well each Respondent’s submission resonated with them. We refer to all the Evaluators as a “GlimpzIt peer group.”

To clarify, a GlimpzIt peer group is independent from the Respondents who supply the actual Glimpzes. This peer group mirrors the demographics of the the Respondents identically.

It’s also important to note that prior to the Evaluators providing their opinions to the Respondents submissions, each of those submissions are validated.

A “validated response” is one that’s determined to include quality data. So instead of including the data from a response such as a picture of a cat along with the text “I love kitties,” we only take into account those responses which provide relevant, valuable insights.

Okay, enough of this though, let’s get to the results. More on our methodology can be found afterward.

Results: what happens when you measure company culture with machine learning?

Remember, the question and prompt was: “Share a picture or video that best describes your company’s culture.”

Most notable Glimpzes

Text: “Teamwork. We all believe in the company values and strive to work together to best serve the community.

Tags: Teamwork, Community Outreach

measure company culture with machine learning

Text: “While work is important, happy healthy moments with those you love are most important to my company.

Tags: Work-Life Balance

measure company culture with machine learning

Text: “Our whole team flew to Hawaii for the weekend to celebrate our first big win. We included spouses and significant others, too. Fun and Food!

Tags: Company Events

measure company culture with machine learning

Text: “We have people from all around the world working for us and helping others all over the world.

Tags: Diverse

measure company culture with machine learning

Visualizing the data

Before looking through the analysis, it’s important to note that GlimpzIt’s machine learning technology doesn’t simply analyze the text. It’s also looking at the pictures and videos submitted the same way a human would, and creating tags based on what it sees.

Most common response by frequency: Teamwork

It’s not shocking that people viewed company culture in terms of their interaction with colleagues. This was true both where people were a part of more horizontal and highly collaborative organizations, as well as for more bureaucratic organizations with hard boundaries between different departments.

Other common responses by frequency included: Diverse, Community Outreach, Training, Caring/Giving

Below is our tag analysis of all the Glimpzes and Evaluations gathered from our study.

When looking at the graph, in green you’ll see the most tagged themes found throughout all the Glimpzes (responses), which were recognized by GlimpzIt’s machine learning technology (e.g. Technology, Practical Use, Entertainment, etc.). In blue you’ll see what resonated most with Evaluators (up-votes and down-votes).

measure company culture with machine learning

Most noteworthy insight from our responses: Community Outreach

People valued Community Outreach so highly that our system identified it as one the key “Opportunity” areas, as seen in the graph below.

measure company culture with machine learning

Further regarding Community Outreach, below are the most used descriptor words associated with it…

measure company culture with machine learning

Finally, in this last image, you’ll see why Community Outreach really emerged as the dominant theme. It wasn’t just identified as an Opportunity, it was also the most highly correlated theme among the other responses and evaluations on our Mind Map.

measure company culture with machine learning

GlimpzIt methodology

You might be thinking, “Why not just send out a survey and ask some multiple choice questions and be done with it?”

Well…simply put, the old cliche, “A picture is worth a thousand words,” is becoming quite relevant in the age of machine learning.

What we’re asking here is not a “yes or no” question. Furthermore, we want to get to the “why” hiding behind these responses; and to get to the bottom of why people feel a certain way, without adding limitations to the potential answers, we need to allow people to express themselves.

Until recently, the most common methodologies for this would have been to field a focus group or use another more in-depth methodology to uncover something that’s more complex than filling in a bubble or putting a check mark into a box.

However, focus groups can cost a whole lot of money and take up a ton of time. Most importantly, these methods of collecting qualitative data simply aren’t scalable.

We want real human insights, and we want to get to those insights as quickly and cost effectively as possible.

This is where artificial intelligence (or machine learning) and a tool like GlimpzIt comes into play, because it makes it possible to analyze qualitative data at quantitative scale.

For more on how GlimpzIt works, see the GlimpzIt Feature Guide

Summing it all up

A good number of respondents think of their company culture in terms of their interaction with colleagues. For example some people emphasized the teamwork aspect of their job. One person mentioned how his company is extremely bureaucratic with clear distinctions between the different departments.

Some respondents think of their company culture in regards to how the company interacts with people outside of the office. This includes things such as how the company treats customers, how the company treats employees, and the role of the company in the community.

According to data from respondents, employees are more happy when company management trusts them, or empowers them to do the work they’ve been assigned without micromanagement. This includes giving them the freedom to organize how they want to approach their tasks and encouraging innovation.


Want to participate in the next Bay Area Ventures Conversation of the Month?

Tune in to Bay Area Ventures on SiriusXM Channel 111 Business Radio Powered by the Wharton School and check out Bay Area Ventures Conversation of the Month. The show airs every Monday at 4:00pm Pacific Time/7:00pm Eastern Time and replays throughout the week.  Its also available anytime on the SiriusXM mobile app.

Want to Use GlimpzIt Machine Learning?

AUTOMATED INSIGHTS, CROWD RESONANCE, RECOMMENDED ACTIONS, DATA MODELING AND ANALYSIS

Contact Us [Free Pilot & Demo]

Leave a Reply

Your email address will not be published. Required fields are marked *