What fast-growing startups say they value
Your values will change as you grow
Split the sample by headcount and the value sets move in a consistent direction. What rises is operational: customer obsession, one team, grit, resourcefulness. What fades is founding-story material: mission and purpose, innovation and contrarianism, iterate and experiment, agency and autonomy.
The obvious reading is that companies mature into caring about customers. The more interesting one is what happens to purpose. It doesn't get added as a company grows up; it gets displaced. At 50 people, purpose is what you have instead of a customer base. By 250, the customers are real and the list gets rewritten around them.
Values that move as a startup grows. Only categories shifting 6 points or more are drawn; the 250+ bucket is small (n=28), so read it as directional.
Table view
| Category | 50–99 | 100–249 | 250+ | Δ |
|---|---|---|---|---|
| customer obsession | 34% | 50% | 68% | +34 |
| one team | 21% | 32% | 46% | +25 |
| grit resilience | 12% | 15% | 29% | +17 |
| resourcefulness frugality | 7% | 10% | 21% | +14 |
| speed action | 38% | 37% | 46% | +8 |
| ownership | 36% | 50% | 43% | +7 |
| rigour evidence | 14% | 15% | 21% | +7 |
| transparency | 11% | 13% | 18% | +7 |
| craft excellence | 23% | 26% | 29% | +6 |
| iterate experiment | 11% | 18% | 4% | -7 |
| agency autonomy | 12% | 15% | 4% | -8 |
| care respect | 23% | 22% | 14% | -9 |
| innovation contrarian | 16% | 15% | 4% | -12 |
| mission purpose | 16% | 4% | 4% | -12 |
What everyone says
Below the top three the list drops away quickly: one team at 28%, craft and excellence at 25%, curiosity and learning at 24%. Twenty-eight categories were needed to code the corpus, but a handful carry nearly all the weight.
Set that against the median value set of five, and the arithmetic is unkind. A startup writing five values, picking each one independently from what its peers publish, will land on a list that overlaps substantially with the company next door. The distinctiveness a values exercise is supposed to produce is being competed away by everyone doing the exercise the same way.
Share of the 200 companies whose values include each category. Categories under 6% are in the table.
Table view: all 28 categories
| Category | % of companies |
|---|---|
| customer obsession | 44% |
| ownership | 42% |
| speed action | 38% |
| one team | 28% |
| craft excellence | 25% |
| curiosity learning | 24% |
| care respect | 22% |
| ambition bold | 20% |
| candour feedback | 18% |
| grit resilience | 16% |
| impact results | 16% |
| rigour evidence | 16% |
| integrity trust | 14% |
| innovation contrarian | 14% |
| transparency | 12% |
| fun optimism | 12% |
| agency autonomy | 12% |
| iterate experiment | 12% |
| resourcefulness frugality | 10% |
| mission purpose | 10% |
| focus simplicity | 9% |
| humility low ego | 9% |
| wellbeing sustainability | 8% |
| inclusion belonging | 6% |
| ground truth | 4% |
| disagree commit | 4% |
| talent density | 3% |
| reliability safety | 2% |
What nobody says
We searched every value name and every description for a set of themes that appear constantly in writing about workplace culture. Most of them are simply not there. Not rare: absent. Two of them do not occur once across 200 companies and 1,001 values.
Profit, revenue and margin appear at 7 companies out of 200, which is its own small story about what venture-funded startups consider worth declaring. Diversity, at 17, is the most common item on this list and still sits far below craft or curiosity.
The charitable reading is that these are hygiene factors rather than values: you don't put "we pay people" on the wall. The less charitable one is that published values describe what a company wants from its employees, and this list is what employees want from the company. The absences are the finding.
Themes searched across every value name and description. Counted at company level, out of 200.
Table view
| Theme | Companies | % |
|---|---|---|
| Psychological safety | 0 | 0.0% |
| Job security or stability | 0 | 0.0% |
| Rest / recovery / burnout | 1 | 0.5% |
| Work-life balance | 3 | 1.5% |
| Privacy / compliance | 3 | 1.5% |
| Documentation | 3 | 1.5% |
| Profit / revenue / margin | 7 | 3.5% |
| Sustainability / climate | 7 | 3.5% |
| Kindness (standalone) | 12 | 6.0% |
| Diversity / inclusion | 17 | 8.5% |
Most values can't be lived
We coded each value for whether it describes something you could catch a specific person doing on a specific day. "Move fast" is behavioural. "Excellence" is not. Just under a third clear the bar, and 38% of companies publish a full set without a single one.
The grammar predicts it. 40% of values are bare nouns and 36% are imperatives, and the split runs straight down the middle of the behavioural coding: iterate and experiment is 58% behavioural, speed and action 57%, focus and simplicity 56%, while mission and purpose manages 10%. Values built from verbs are recognisable. Values built from nouns aren't.
This is the difference between a value and a mood. A company whose list is "integrity, excellence, passion" has published three adjectives about the kind of people it hopes to have hired. There is no version of a Tuesday in which someone visibly does excellence, which means there is no version of a Tuesday in which anyone can be recognised for it, and a value nobody is ever recognised for is one nobody has any reason to act on.
Share of each category’s values that describe an observable behaviour rather than a trait. Across the whole corpus, 31% are behavioural, and 38% of companies publish no behavioural value at all.
Table view
| Category | % observable | values |
|---|---|---|
| iterate experiment | 58% | 24 |
| speed action | 57% | 83 |
| focus simplicity | 56% | 18 |
| agency autonomy | 48% | 25 |
| rigour evidence | 47% | 32 |
| ownership | 44% | 84 |
| integrity trust | 40% | 30 |
| resourcefulness frugality | 38% | 21 |
| candour feedback | 35% | 37 |
| innovation contrarian | 34% | 29 |
| impact results | 33% | 33 |
| transparency | 32% | 25 |
| grit resilience | 31% | 35 |
| care respect | 30% | 46 |
| fun optimism | 24% | 25 |
| customer obsession | 23% | 90 |
| curiosity learning | 20% | 50 |
| ambition bold | 18% | 40 |
| craft excellence | 17% | 52 |
| one team | 14% | 58 |
| humility low ego | 11% | 18 |
| mission purpose | 10% | 21 |
Whose values are you running?
Ownership, customer obsession, bias for action, raise the bar, disagree and commit and dive deep are all Amazon Leadership Principles. Extreme ownership comes from Jocko Willink's book of that name. First principles is the Musk-popularised strand of a much older idea. Nearly a quarter of the sample uses at least one of them as a value name.
This isn't copying, and reading it that way misses what is actually happening. It's a shared vocabulary: a set of phrases that became the default way to express an idea, in the same way that "OKR" became the default way to talk about goals. Using them is efficient: everyone knows roughly what "bias for action" means before you explain it.
The cost is that the phrase arrives with someone else's definition attached, formed at a company with a different size, market and set of problems. If you publish "customer obsession" without saying what it means at your company, the definition your team will use is Amazon's.
Value names matching phrases that entered startup vocabulary from a handful of famous sources; six of the eight below are Amazon Leadership Principles. 49 of 200 companies (24%) use at least one.
Table view
| Phrase | Companies | % |
|---|---|---|
| Ownership | 17 | 8% |
| Customer obsession | 15 | 8% |
| Bias for action | 10 | 5% |
| Extreme ownership | 6 | 3% |
| First principles | 5 | 2% |
| Raise the bar | 4 | 2% |
| Disagree and commit | 4 | 2% |
| Dive deep | 2 | 1% |
Culture isn't universal
The startup value set reads as a global standard, and mostly it is: customer obsession sits between 38% and 53% in every region we could measure. But the edges differ. Speed and action is markedly a US preoccupation. Integrity and trust is named nearly twice as often outside the US as inside it. Curiosity and learning spikes in Latin America.
Darker means a higher share of that region’s companies name the category. LatAm, Asia and MEA cells rest on small samples, so read the pattern, not the decimal.
Table view
| Category | USA (n=128) | LatAm (n=10) | Asia (n=19) | Europe/other (n=27) | MEA (n=16) |
|---|---|---|---|---|---|
| customer obsession | 43% | 50% | 53% | 44% | 38% |
| ownership | 43% | 40% | 32% | 41% | 44% |
| speed action | 45% | 20% | 37% | 22% | 31% |
| one team | 30% | 40% | 16% | 30% | 25% |
| craft excellence | 23% | 40% | 26% | 26% | 31% |
| curiosity learning | 24% | 60% | 16% | 19% | 19% |
| care respect | 20% | 10% | 26% | 26% | 25% |
| ambition bold | 16% | 20% | 21% | 30% | 25% |
| candour feedback | 16% | 10% | 16% | 22% | 31% |
| grit resilience | 13% | 20% | 32% | 22% | 0% |
| impact results | 17% | 10% | 11% | 19% | 12% |
| rigour evidence | 19% | 10% | 16% | 7% | 6% |
| integrity trust | 14% | 0% | 26% | 7% | 25% |
| innovation contrarian | 14% | 20% | 11% | 11% | 19% |
What the newest startups dropped
Grouping companies by the YC batch they went through shows the newest cohort publishing a narrower set than the oldest. One team has more than halved. Transparency has almost vanished. Fun and optimism, innovation and contrarianism, grit, and curiosity all fall in the same direction.
Almost nothing rises to replace them. Craft and excellence is the only category that clearly gains, from 23% to 32%. The top three (customer obsession, ownership, speed) barely move at all. The collective vocabulary is thinning: the newest companies are converging on the same short operational list and dropping everything around it.
A caution on causation: these are cohorts, not the same companies observed over time, and the 2022+ group is younger and smaller as well as newer. Some of this is age, not vintage.
Younger cohorts on the right. The collective vocabulary (one team, transparency, fun) thins out in the newest batches.
Table view
| Category | 2018–19 (n=65) | 2020–21 (n=91) | 2022+ (n=44) |
|---|---|---|---|
| customer obsession | 48% | 42% | 43% |
| ownership | 35% | 47% | 39% |
| speed action | 37% | 38% | 41% |
| one team | 42% | 25% | 16% |
| craft excellence | 23% | 23% | 32% |
| curiosity learning | 29% | 22% | 20% |
| care respect | 26% | 18% | 23% |
| ambition bold | 22% | 20% | 16% |
| candour feedback | 15% | 21% | 14% |
| grit resilience | 22% | 12% | 14% |
| impact results | 20% | 12% | 18% |
| rigour evidence | 15% | 18% | 11% |
| integrity trust | 18% | 12% | 14% |
| innovation contrarian | 17% | 15% | 7% |
Position one is the honest one
Order is data. It survives in the collection because we recorded each value's published position, and it turns out to carry information the prevalence chart doesn't. Customer obsession isn't just the most common value; it is by far the most likely to be put first, with a mean position of 2.5 in sets that average five.
Read the bottom of the table instead and you get the more useful signal. Craft and excellence appears at a quarter of companies and almost never leads. One team appears at 28% and leads 17% of the time. These are the values that got included because leaving them out would have looked odd. What a company puts first is what it will actually trade the others against. If you want to know what an organisation really optimises for, read position one and stop.
When a category appears in a set, how often does it appear in position one? Position one is the one the founder actually believes.
Table view
| Category | Listed #1 | Appearances | % |
|---|---|---|---|
| customer obsession | 41 | 90 | 46% |
| mission purpose | 7 | 21 | 33% |
| ambition bold | 12 | 40 | 30% |
| care respect | 14 | 46 | 30% |
| speed action | 22 | 83 | 27% |
| transparency | 5 | 25 | 20% |
| ownership | 15 | 84 | 18% |
| integrity trust | 5 | 30 | 17% |
| one team | 10 | 58 | 17% |
| focus simplicity | 3 | 18 | 17% |
| innovation contrarian | 5 | 29 | 17% |
| humility low ego | 3 | 18 | 17% |
Five is the number
This is the most directly usable number in the study. If you are writing a set of values and wondering how many to have, the answer the market has converged on is five, with three and six both entirely normal and four very common.
Above eight you are an outlier: four companies publish eight, and six publish more. In practice, most long lists aren't values at all: they are operating principles collected under a values heading, which is a reasonable thing to publish but a different thing.
200 companies with a published set. Median is 5. 180 of the 200 (90%) sit between 3 and 6. Anything above 8 is an outlier.
Table view
| Values in set | Companies | % |
|---|---|---|
| 3 | 28 | 14% |
| 4 | 49 | 24% |
| 5 | 61 | 30% |
| 6 | 42 | 21% |
| 7 | 10 | 5% |
| 8 | 4 | 2% |
| 9 | 2 | 1% |
| 10 | 2 | 1% |
| 11 | 1 | 0% |
| 13 | 1 | 0% |
Values, principles or tenets
We recorded the verbatim heading above each list. "Values" wins comfortably at 61%, but a third of companies reach for something else: culture, beliefs, "what we stand for", "how we operate", or a bespoke line of their own.
The rename usually signals an attempt to fix exactly the problem in section 4. "How we work" is a promise that what follows will be behavioural. It's a good instinct, though changing the heading is easier than changing the list underneath it, and both have to happen for the rename to mean anything.
The verbatim heading above the list. "Values" dominates, but a third of companies reach for something else, and 4 publish a set with no heading at all.
Table view
| Heading used | Companies |
|---|---|
| values | 124 |
| other heading | 32 |
| "how we work" | 17 |
| principles | 10 |
| culture | 9 |
| beliefs | 5 |
| unlabelled | 4 |
| tenets | 1 |
Every value we found
All 1,001 values, in the companies' own words. Search the names and descriptions, filter by category, size or region, or click a company to see its full set in the order it publishes them. Descriptions are verbatim, including the punctuation, emoji and non-English text that came with them.
Sixty-eight of these rows are marked with a dashed chip: they sit in the three categories (product doctrine, hiring criteria and uncodeable) that are recorded to keep the taxonomy honest and are excluded from every statistic above.
How this edition was made
- Sample
- 400 Y Combinator companies (active, batch 2018 or later, 50 or more employees), taken from YC's own company directory as of 14 August 2026.
- Collection
- An automated crawl of each company's own site: homepage, about, careers and culture pages, plus careers subdomains and hosted applicant-tracking boards. First-party sources only: no aggregators, no LinkedIn.
- Coverage
- We found values for 202 companies. 157 publish none. 41 could not be assessed, being bot-blocked or entirely client-rendered, and are excluded from the denominator rather than counted as silence. That gives a publication rate of 56%: 202 of 359 assessable companies.
- Corpus
- 1,001 values analysed, taken from each company's primary set where several exist. The analysis below is based on the 200 companies whose sets we could code in full.
- Coding
- The taxonomy was induced bottom-up from the corpus itself: three independent passes using different lenses, merged into 28 categories, against which every value was then coded. Three further categories (product doctrine, hiring criteria and uncodeable) are recorded to keep the taxonomy honest and are excluded from every "values" statistic on this page.
- Regions
- Regional buckets are derived from each company's country. The two companies in North Africa sit in a slightly different bucket in the explorer than in the regional chart, which uses the study's own coding, so explorer counts by region can differ from the chart's by one or two.
- Limits
- The 250+ headcount bucket is 28 companies: the scaling curve's direction is reliable, its endpoint is not. Non-US regions are small samples throughout. And a published value is a claim about culture, not evidence of one. This study measures what companies say, not what they do.
About this report
The Startup Culture Insights Report is an annual study by BuddiesHR. It is put together by two YC alumni, Fabien Pinel and JY Delmotte.
Fabien Pinel
JY DelmotteFound a company we missed, or a number you want to argue with? Write to hello@buddieshr.com. Earlier editions are listed on the report hub.
If only a third of values describe something you could catch someone doing, how would you recognise anyone for living them?
That's the question section 4 leaves behind. Clappy is the BuddiesHR app for it: peer recognition in Slack, tied to the values you actually published, so the behavioural ones get noticed and the decorative ones become obvious.
See how peer recognition works →