CultureX

In our new article in The Economist,  Donald Sull and Charlie Sull argue that AI will transform leaders' ability to understand and improve corporate culture. Our key points: 1) Culture matters: Nearly 80% of CEOs and CFOs list corporate culture among the top five factors driving financial performance. Organizations with a healthy corporate culture can attract, retain, and engage the best employees. A study by Alex Edmans of the best places to work in the U.S. found that their returns to shareholders outperform peers by 24% over a five year period. 2) Culture is not where it needs to be: In the U.S., the average employer rates a 3.6 out of 5 on Glassdoor. Most people wouldn't relish a meal in a restaurant or a ride with an Uber driver with a similar rating. How excited do you think they are to work in an average culture? 3) Current tools to measure culture are ineffective: Organizations cannot improve culture unless they can measure it. Employee engagement surveys excel at assessing whether an employee is engaged, but fall short in measuring all the dimensions of culture. Faced with dozens of multiple choice questions, employees zone out and provide the same or similar answers to very different questions. Their numerical scores provide little context or guidance on how to improve. And of course they cannot discuss topics that are not included in the survey. 4) Free text is a goldmine of cultural insights: When employees answer open-ended questions they discuss what matters most to them, provide context on their concerns, and often suggest concrete recommendations for improvement. The sheer volume of employee feedback available to leaders is staggering. For a large organization, the free text from internal surveys, online reviews, 360 performance reviews, exit interviews, and other sources can equate to dozens of novel-length "books" brimming with granular insights from what employees have to say in their own words. 5) AI revolutionizes our ability to analyze free text: In the past, companies resorted to crude tools like word clouds or key-word search to analyze employee feedback. Now, large language models can reliably make sense of free text at scale. Our CultureX platform, for example, classifies employee feedback into hundreds of granular elements of culture. Armed with these measures, leaders can assess how well their organization lives its core values, identify toxic sub-cultures, smooth cultural integration in M&A, and identify what drives critical outcomes like employee engagement or innovation. #corporateculture #ai #culture https://lnkd.in/gUYPCXEQ

If there is one cultural value every CEO wants to cultivate, it is perhaps ownership. Wouldn’t it be great if employees proactively treated the company like their own, and held themselves accountable for results? Netflix, NVIDIA, SpaceX, Goldman Sachs, Uber, Ritz-Carlton, Tesla, and some divisions of Amazon are all examples of companies that significantly outperform their industries on ownership, according to what tens of thousands of their employees recently said on Glassdoor. It is no wonder leaders want to cultivate this trait. But you can’t have the cream without buying the cow. If you want employees to proactively take ownership for results, you need to trust them. At least this is what the strongest ownership cultures tend to do. This chart shows how about 1500 of the largest employers in America recently spoke about ownership and trust in Glassdoor comments, compared to their industries. By the standards of this kind of analysis, the two topics are strongly correlated. This is a real dynamic, and it is also a tension that I see play out in companies regularly. The CEO wants ownership, like all CEOs would. However, when s/he realizes that to accomplish this you need to trust employees and give them autonomy (empowerment is also quite strongly correlated with trust), they hesitate. That is not the culture they feel comfortable with… In our Culture Champions conversation with Katie Burke, recent CPO of HubSpot-- a company that scores high above its industry average on both ownership and trust-- she shared that a big part of the puzzle is selecting talent that you can trust to prosper autonomously, and that actively enjoys being in such an environment. In HubSpot’s legendary Culture Code, there is an all-capitalized section that goes “AUTONOMY REQUIRES TRUST”. Wise words. I would encourage anyone interested in building a strong culture of ownership to read HubSpot’s culture code and to listen our talk with Katie. Katie Youtube: https://lnkd.in/ezTkKn-c Katie Apple Podcast: https://lnkd.in/ehaTiuDr Katie Spotify Podcast: https://lnkd.in/eGcJB_qJ Katie MIT SMR Writeup:  https://lnkd.in/eW7994cY HubSpot culture code: https://lnkd.in/emHSmhzG (About the chart: R = 0.44 [sounds moderate, but this R2 is more than 2 standard deviations above average for this kind of analysis], P-Value < 0.0001, Color coding by industry)

This isn’t going to last another 10 years, is it? With AI around now? Solving a problem as important as culture? I just don’t see it. Better employee listening is surely right around the corner. (Source of image: Average point-scale “engagement” survey response of more than 900,000 respondents in our MIT study. Half of respondents answer long surveys with less variance than this)

In our quest to elucidate the secrets of successful culture, Donald Sull and I have launched a new series: Culture Champions. In this series, we have discussions with the people responsible for creating the strongest cultures in the world and learn their secrets. In contrast to many culture podcasts, books, and articles, which select guests based largely on how well known the company is (“We are speaking with the CEO of Coca-Cola!” “Does Coca-Cola have a strong culture?” “No idea, but it’s Coca-Cola!”), we only interview Culture Champions: companies that *measurably* live up to their values and excel on both financial value creation and the employee experience. This way you know that our guests know what they are talking about. It is important to do it this way, because most companies do not have strong cultures. The first Culture Champion we spoke to was Katie Burke, in her reflective last interview as Chief People Officer of HubSpot. Katie was one of the primary architects of a truly remarkable culture that measurably lives up to its core values in the eyes of its employees (like transparency, below). It was fascinating to learn how HubSpot did it (and we wish her best of luck in her new role at Harvey!) These conversations have been some of the most interesting and insightful that we have had on this ten-year journey. We hope you enjoy. Katie Youtube: https://lnkd.in/ezTkKn-c Katie Spotify Podcast: https://lnkd.in/eGcJB_qJ Katie Apple Podcast: https://lnkd.in/ehaTiuDr Katie MIT SMR writeup: https://lnkd.in/eW7994cY

Many people think the future of employee listening does not involve a survey at all. You can just listen to employees passively through Slack, Teams, emails, etc. At the risk of sounding old fashioned, let me caution against that. Five reasons not to passively listen to employees: 1. Employees do not like being spied on. Did they “agree” to this in the fine print of the legal paperwork they signed during onboarding? According to some lawyer, maybe. Who cares. People do not like having their emails, Slack (and where does it end, meetings recorded, office space videorecorded?) analyzed by big brother. If you do this, you are making the culture more toxic, which should be the exact opposite of anyone’s employee listening objectives. (“But we are just spying on employees to help them!” you protest. Give me a break. Every dictator in history has said the same thing, and don’t pretend your interests always align with theirs) 2. If employees are being surveilled, they will change their behavior to evade surveillance. Think you are capturing candid information? It won’t be candid for long when employees realize they cannot speak freely. Now they don’t write emails saying anything meaningful any more. Great. Psychological safety has just been massively inhibited across the company and you are also getting guarded, low quality data.  3. Most emails, Slack messages, etc. are not very culturally rich. Look at your emails. What is their content? “Let’s set up a meeting for 10am on Monday”. “Sorry, could we do 12pm?” “Sure”. The content of most emails are logistical or about specific details of projects, and do not reveal many insightful details about the culture. A survey may sound old fashioned, but it is also a rare opportunity for employees to directly tackle the big questions: “What are our strengths as a company?” “How can we improve?” With surveys (properly structured), employees freely give you the right answer about exactly what you want to hear about. With passive listening, it is hunting for insight breadcrumbs. 4. Email, Slack, etc. is highly unstructured and contextually dependent (“They were being so rude to me!” says a Slack message. Sounds like an important insight, but who is the “they”? It is not specified in the message since these forms of communication often rely on context outside of the messages themselves). This makes it significantly more technically challenging for an AI platform to effectively analyze data than if the data is at least somewhat structured, like a survey. (Have been researching the latter technical problem for 10 years at MIT and, believe me, even that is non-trivial)    5. There is a good chance your lawyer won’t let you do it anyway (especially in Europe) You want to see what happens to a culture that takes passive listening to its logical conclusion? Look at Bridgewater. They have a great culture, right? Ermmmmm… not exactly.

An invisible conversation that goes on every day at a large number of large companies CEO: So what are we going to do about culture? HR: Culture? CEO: Yes. It is incredibly important. One of the biggest drivers of financial value in our whole company. HR: Right! CEO: So what is our culture? HR: I have a few personal ideas, based on my conversations with colleagues. CEO: Right, but what is the rigorous, data-driven answer? Culture is a multi-billion-dollar phenomenon in a company of this size. We need to measure it very rigorously. What actually is it, in this company? HR: Uh, well… CEO: Just basic questions, I mean. What are its strengths? What are its challenges? How did it change this year? Is it supporting our strategy or not? What are the most powerful drivers of employee satisfaction? How do different segments of the company differ culturally and which do we need to worry about? What data do we have on all of this? HR: Well, we did an engagement survey a couple of months ago. CEO: What did that tell you about the culture? HR: Basically nothing. CEO: Okay, so what are we going to do to improve the culture that it sounds like we know almost nothing about? HR: My team has come up with several ideas for learning and development programs and other initiatives that will definitely improve the culture! CEO: Even though we do not know what the culture is or how to improve it? HR: Exactly! CEO: Great! (Shake hands)

Glassdoor is a place where employees go to vent, right? You can’t take it seriously because everyone is just venting there, right? It’s unreliable data, right? Wrong. Employees are more than 3 times as likely to give a positive review on Glassdoor than a negative review. It is a remarkably balanced review platform. Are all of your employees giving negative reviews on Glassdoor? That sounds like a you problem, not a Glassdoor problem.

I tried to squeeze what I have learned about culture from our past ten years of research at MIT onto a cheat sheet for leaders responsible for shaping it.  

Imagine a dozen people talking about something that none of them can see. Each of them has their own point of view on what the invisible thing is, what it should be, how to get it there, and whether it has gotten better or worse over time. Most discussion points are based on personal observation or intuition. Sound like a productive conversation?   Welcome to most boardrooms when the topic of corporate culture comes up. I know how these conversations go. They can quickly devolve into *deep* BS in a way that discussions about other topics do not, because culture is so intangible and everyone has an armchair opinion about it (“empowerment is key to culture!” really? In every circumstance?) As a corollary to evidence-less discussions, look at the official cultural values these discussions eventually produce. At MIT we did a study of more than one million employees: there is no correlation between a company officially espousing a cultural value and actually embodying it in the eyes of employees. Everything is grounded in air! The only way to have a productive conversation about culture (at least in a big company) is to ground it in high-quality data. High-quality cultural data was not available a couple years ago but, with AI, now it is. Right now most companies fly almost completely data-blind when trying to manage the asset that probably creates or destroys more financial value than any other asset they have. It’s wild.

What drives employee satisfaction? Being respected? Colleagues? Ping-pong tables? Yes and maybe but, also, labor scarcity. This chart tracks the average Glassdoor rating in the USA--- a great metric of employee satisfaction—versus the US Labor Force Participation rate in June of the same year. First: some disclaimers. This is an intriguing chart, but there are a lot of question marks. How much of this is natural economic cycle, and how much of it is COVID? What role is compensation playing here? (Interestingly, it appears to have a notable but not decisive impact on satisfaction, consistent with our research into culture largely driving the employee experience) Are there other unknown variables affecting a simple chart trying to tell such an ambitious story? Almost certainly. Could the most dissatisfied parts of the labor force removing themselves definitionally increase satisfaction? Perhaps, although that doesn’t explain the Seahorse’s tall back. Still, take the chart at face value for a second. What it suggests is that, when labor is scarce, employers try really hard to provide a strong employee experience and satisfaction goes up. And, when labor is plentiful, they deprioritize the employee experience and satisfaction goes down. In 2024, that is the part of the chart we are in. It’s a simple narrative that would please an economist. The dynamic in this chart (and its consistency with my personal experience talking to business leaders) is why I have mostly given up on convincing business leaders to prioritize the employee experience right now. They’ve got other priorities, why fight it? You need a strong culture to attract and retain employees! Sure, but almost no one has that problem right now… if anything, they’re trying to get rid of employees. And the toxic cultures that do probably won’t listen to you, anyway. Does this mean leaders should not prioritize culture? Of course not. I view the current period in the labor cycle as an opportunity to sit back and re-assess the meaning of the word culture. Let’s move from a simplistic view of culture (“Happy employees good!”) to a grown-up version of culture as a remarkably powerful, remarkably multi-faceted “operating system” governing most value-creating behaviors in a company. This conception of culture requires advanced analytics, not broken point-scale surveys, as well as a little bit of thoughtfulness and ingenuity from leaders. If organizations embrace this new era of culture, it opens the doors to a richer way of doing business that creates vastly more economic value and also makes employees happier all labor cycle-round. (R2 = 0.62, P-Value < 0.0002)