Culture Compass: a Personality Test for Companies
Understand your company's tendencies, strengths, saboteurs, and lookalikes... in 10 minutes
May 18, 2025
Culture Compass is a tool that asks you a few quick questions to help you discover your company’s personality archetype, explains strengths & saboteurs, and find similar companies… all based on 4 simple dimensions:
- Who should make the final decision (leaders, individuals, or committees)
- How employees are motivated (people-oriented, success-oriented, or purpose/mission-oriented)
- How typical interactions proceed (harmonious, principled, or challenging)
- Tempo (urgent or methodical)
This produces 54 combinations, which can be assessed very quickly and easily with ~10 questions. That unlocks all sorts of insight:
- A tangible sense of the day-to-day at the company
- Strengths of the culture
- Saboteurs (I love this term)
- What Role Models look like in this culture
- Other companies that feel similar
You can click around on these to quickly explore archetypes. And just for fun, we built a Sarcastic Mode:
To make the serious version engaging too, we analyzed 100+ companies to suggest examples of each archetype that you can anchor to, as you explore.
But obviously, since there are many more: we also plugged in an LLM to let you paste in any company's values to suss out what their values are (or at least, what they'd like their values to be). Here's what we get when we plug in Stripe's values. IMHO, this is quite reasonable based on what I know from people who have worked there.
Backstory: from psychology nerd to trying to embody my own values
I studied it in college paired with applied math, and I've found that learning incredibly helpful in navigating tricky situations – especially for someone like me who's not naturally high-EQ.
So over the years, I've been very into every framework you can imagine: Big 5, MBTI, DISC, PrinciplesYou, Saboteurs, Key Values… even my first ever angel investment was in a company that provides a personality API!
Around this time I founded AbstractOps and drafted our values. The most fun + challenging part is avoiding universally agreeable statements: I laugh (and die a little inside) when I see a company say one of their values is "excellence."
Really? As opposed to all the other companies out there that love mediocrity?
Values need to represent real tradeoffs. Excellence comes at a cost (long hours, time, money).
I'm glad I spent as much time thinking about this stuff... but I'll admit that over the succeeding years, I/we did not manage to build an excellent culture, despite trying very hard. I did learn a lot about which things worked & didn’t work in trying to build culture though:
- Bad culture will absolutely sink a company
- Good culture requires hard choices, not agreeableness
- Picking the right ingredients is a tightrope act across industry, problem / product, founders
- Trying to fix this without embodying the values is worse than having no values at all — because everyone will rightfully make fun of you
- A company should probably not fit its culture to its employees (we tried); it should be set by founders and executives, and serve as the basis for candidates & employees to opt in or out
- In particular, a distinctive careers page is a great way to attract strong talent
- Culture is enacted in policy, not prose: how decisions are made, how compensation & benefits are set, how transparent or political people are with each other... this isn't some touchy-feely stuff. This is the very soul of how work is done.
How to Draft your Culture
Everything we do at Autograph is tied to the core principles that the right framework can turn fuzzy, qualitative things into structured data. This allows customers to build better mental models, understand the world around them better, and make better decisions. Besides headcount management (our core product), we’ve also built tools for compensation (Fair Offer, v3 launched last week), now culture... and more coming soon.
That’s why I've been working for the last 18 months to make sense of the culture conundrum: how can you boil something so human down into dimensions without making it useless?
The first a-ha moment I had was when I encountered Jonathan Haidt's Moral Foundations Theory. It's a way of categorizing societies:
There are several innate psychological systems at the core of our “intuitive ethics”: Care/Harm, Fairness/Cheating, Loyalty/Betrayal, Authority/Subversion, Sanctity/Degradation, and Liberty/Oppression.
But… companies are just societies. So v1 of Culture Compass was just a Custom GPT, which asked literally one question:
"Stack rank the following values from highest to lowest priority: care, fairness, loyalty, respect for authority, mission-orientation, and autonomy."
The only adjustments were for sanctity (which roughly translates to mission-orientation at a company) and liberty (which translates to autonomy). It was fun, but very obviously a prototype, too plain vanilla. But there was the thread of something interesting there, so we've been tugging on it since.
The second realization was that Moral Foundations is like the Big 5: it's single-pole. It's high or low on a single dimension. Myers-Briggs is less scientific, but it has one useful thing going for it: it's sort of double-pole. Intuition and sensing, thinking and feeling, perceiving and judging aren't traditionally opposites... they're just different.
We completely rebuilt Culture Compass incorporating this flexibility, and it was definitely richer. We were about to ship it:
- Autonomy vs. Authority (1-pole)
- Data-driven vs. Intuitive (1-pole)
- Challenging vs. Harmonious (1-pole)
- Bias to action vs. Bias to process (~2-pole)
- People-oriented vs. Mission-oriented (2-pole)
… But something was off. It didn't capture bureaucratic tendencies, or taste / quality-inclinations. It didn’t allow for “mercenary” cultures that very much exist and succeed. So the final piece clicked into place when it became clear that some of the dimensions actually call for three poles:
- Decision Norms: Autonomy vs. Authority vs. Consensus (3-pole)
- Motivation: People vs. Purpose vs. Winning (3-pole)
- Interaction Style: Harmonious vs. Principled vs. Challenging (2-pole)
- Tempo: Urgent vs. Methodical (~2-pole)
The additional precision also unlocked the "recommended policies” that you’ll see on the app: ranging from expense policy to IT / security norms to recruiting philosophy.
This increased the number of archetypes from 32 to 54 – slightly more unwieldy, but reducing from 5 axes to 4 axes more than offset this. Decider | Motivation | Interaction | Tempo is quite easy to grok.
Speaking of Grok — this project was entirely made possible thanks to LLMs: Vercel v0, OpenAI, Claude, Gemini, and Grok.
- Brainstorming combinations and exploring semantic spaces
- Generating the thousands of possible combinations and whittling them down to a pool that it was feasible / digestible for me to look at
- Creative starters for funny additions like Sarcastic Mode that I was able to pick up and run with
- Cleaning and formatting data to make it 10-100x faster for me to work
- This post was written with lots of feedback from o4
- Last but definitely not least: I vibe-coded the actual app interface. No dedicated engineer — just v0 and Gemini.
Two years ago, it would have taken me 300+ hours of analysis and an engineer 50+ hours to build this. Instead, it took probably 30-50 hours of actual analysis and 10-20 hours of building.
How should you use Culture Compass?
- Take the 10-question assessment first, to avoid bias
- Click around the related archetypes and generally explore the axes, the resulting strengths / saboteurs, and example companies
- Dig into potential policies choices you could make at your company
- Try synthesizing AI-generated archetypes for companies you look up to
- Share & discuss internally… and of course, please share feedback!
Closing Notes
Given the utility enabled by LLMs, we think that GTM, content marketing, and mindshare for the next generation of companies won't be unlocked by cold email or AI / SEO "slop"... but instead genuinely useful tools, inspired by a deep understanding of the subject matter. It truly is the most amazing time to explore one's curiosity for niche passions to the fullest.
So, assuming people like them – and early signs indicate that they do – we're going to keep building & shipping useful apps. Most of them free, like Fair Offer and Culture Compass.
Of course... none of these things are perfect. Even 54 archetypes can only capture so much nuance. But one of my favorite sayings is "all models are wrong, but some are useful."
We hope you find our models useful :).
H/t to Amar Sood and Noah Itovitch for feedback on various iterations of Culture Compass.
Share article
Thanks for your curiosity about our blog! We send curated emails, once a month.