The Real Rates of Business Survival | Hacker News Reader

The Real Rates of Business Survival

fleximize.com

174 pointskiriad69 comments

The second slide says "If 8 out 10 (or 80%) businesses fail after the first year, the remaining 2 (or 20%) would probably not survive to see their fifth year" - how does this follow? The 'myth' only talks about the first year of the business, and doesn't say that 80% of businesses fail every year.

Indeed, this is a complete straw-man argument. Obviously, the longer a business has been around, the less likely it is to fail -- which is exactly what the article itself points out on slide 9.

The argument sounds like it's based on a geometric expansion of 80% failure rate for each year.

    (1.0 - 0.8)^5 = (0.2)^5 = 0.00032

That would mean that in 5 years, 99.968% of business would fail.

Clearly that's a fallacy; an 80% failure rate in the first year does not imply an 80% failure rate in subsequent years.

It assumes that "80%/year" failure rate implies something like exponential decay. In this model of business, they behave something like particles decaying; the business environment might be treacherous enough that failures are due to factors that are essentially random and unmanageable.

While others (and you) have pointed to the deficiencies of this model, or at least to its assumptions, it is not fallacious, per se. The presence of other models--for example, that the first year is the most treacherous because the new owner lacks experience--does not make the current model, nor conclusions drawn from it, fallacious. Indeed, the implied size of the population, "all businesses," is a warrant for some pretty strong claims. Additionally, treating a saying as data, rather than respecting the context in and modality with which it is offered is probably fallacious in itself.

So, quoth the Bard, "If you're wondering how he eats and breathes/and other science facts/la la la/repeat to yourself, "It's just a show,"/"you should really just relax!" :D

No, whenever this is said "80% don't survive the first year" it's specific to the first year. Infant mortality basically

Half of all Carbon 14 decays in 5,730 years. So, we're not saying anything about the next 5,730?

Language doesn't work like code. Many interpretations can be sustained by a given text.

Why would you assume that company survival follow an exponential decay in the first place?

This is obviously not the case

Exactly, your first sentence isn't saying anything at all about the next 5730. Without outside knowledge of exponential decay, there's zero reason to imply anything about the next 5730.

If you instead say "the half-life of Carbon 14 is 5730 years" then you would be saying something about the next 5730 as well.

The English of "8 out of 10 businesses fail in their first year." is completely unambiguous.

The statement "Half of all Carbon 14 decays in 5,730 years" does not say anything about the decay in years after that.

Half lives on the other hand are defined as being a re-occurring phenomenon, such that it _does_ say something about future years.

Language does not work like code, yes, but that doesn't mean it can't have a clear and defined meaning. If it couldn't, language would be useless as a communication method.

8 out of 10 Start-ups(!) fail, not businesses. There are loads of businesses that have a very low failure rate, but also a very low quadrupletuple-the-evaluation rate.

Example: You work for company X and build wordpress blogs for companies A,B,C since 5 years. Now you think it's not reasonable to have company X in the middle since your relationship to A,B and C is good enough. So you quit and work with them directly as a freelancer. At that point you started a business and the risk of failure is as low as your estimation of the relationship to A, B, and C is as good. If you are not too unrealistic you only do this when it is very likely to succeed. But your business value may stay about the same the next 5 years.

A startup is just a new business. In the first year of existence, every business is by definition a startup unless created by spinning an existing unit out of a business. Some people might say that the term "startup" is only applicable to new tech companies aiming for absurd growth, but those people are defining too narrowly.

The "8 out of 10 startups fail in a year" claim also rings hollow without a citation, even if restricted to "tech startups". Is there some source that demonstrates that 8/10 high-growth startups fail within a year or is this as unsupported as the more general claim?

In general you are right. Start-up should be the word for any new business. But in recent years the venture capital driven fast growing new-tech business type was mostly associated with that word, especially on HN (since Y-Combinator is a school for exactly that kind of business). That's why I gave an example that is pretty much on the opposite scale. It's also a new business but not what people think about when they hear the word "start-up".

I edited to address this a little more specifically and you likely missed it.

It's still unverified that even high-growth tech startups fail 80% in the first year. How many "Facebooks for {foo} and Uber for {bar}" actually fail in the first year? If you define a high-growth tech startup from the time it takes investment or the time it incorporates, I bet you still don't hit 80% failure rate in the first year. (If you count abandoned ideas that were never meaningfully pursued, then sure, but in that case I've got a dozen failed startups under my belt and a similar number of failed restaurants.)

Was start-up really simply the word for new businesses in (american?) English?

I heard it only for the business you describe, but lately every newly created design/tech-consultancy here calls itself a start-up.

I'd say if a company takes venture capital it's a start up, whereas something based on bank loans and self capitalisation is just a business. But I don't have any justification for that - just my association of the word.

I thought a start-up needs to have a product/service that aims for ultra-high-growth.

That seems to be a relatively accepted definition around here, but not a dictionary definition.

Yeah, it became popular, so no definition is really true anymore. Everybody tries to claim it for themselves to get the coolest tech people and the most VC dollars.

It's often a bit confusing.

I get a job offer from people who say "We are a start-up" and then they are just another dumb agency. If I want agency work, I freelance.

The way people I hear people use it to describe their own businesses, it seems to describe a business that isn't making money yet (large or small).

If I start a McDonalds francise, even if it's my first-ever business, it's DEFINITELY not a startup.

It's a fuzzy definition like any other social category, but I'm not sure a restaurant is a startup unless it aims for significant growth via some novel approach to running and scaling restaurants.

What's "significant growth?" That's one fuzzy area. I'm not sure you have to be aiming to become Fortune 500 in less than 10 years, but the aim should be for at least one or two standard deviations above normal for the sector.

The best definition I can think of is that a startup aims to deviate significantly from the industry norm through the application of technology or higher order conceptual reasoning about process.

> A startup is just a new business

I've only ever heard of startup used to refer to a new business that hopes to become extremely scalable. In my opinion, starting a burger restaurant isn't a startup, unless you're hoping to be on every streetcorner competing with McDonald's in the next 5-10 years. It's just much easier to scale with a tech/software company.

The 8/10 claim is still baseless any way you define it though. The Forbes article this presentation links to only cites "Bloomberg.com" as the source, not linking to any specific article.

> I've only ever heard of startup used to refer to a new business that hopes to become extremely scalable.

I think the recent popularization of the Facebook story, Google rise to dominance, and Apple's romanticized origin story have all contributed to many people thinking of "startups" as isolated to tech or to high growth scalable business. But before this, the term "startup" was used to refer to many things. A chemical company goes through iterations of plant "startups", even while railing many hundreds of millions in capitol, that will only pay off after a decade of plant operation for example.

Personally I don't like to think I'm starting a "startup", but rather founding a business.

It does seem to be the definition bandied about here as espoused by pg's essay: http://www.paulgraham.com/growth.html

As a software engineer working nearly 2 decades, prior to discovering HN some 5 years back I always just thought it was a small, relatively new tech company that was designed to grow beyond a lifestyle business. I never thought about the hypergrowth factor. I still use the term in that fashion.

Interestingly enough, my gf works in commercial finance and they use the term to apply to any new business where there is no like prior for the applicant. In your example above a brand new burger business is certainly a startup until they open up a second location. IOW, they're no longer a startup once they've proved successful enough to grow to a second location with the ability to use the first as collateral reducing their inherent risk.

This has been shown to be a nonsense myth too. Someone's put some hard figures here:

https://www.quora.com/What-is-the-truth-behind-9-out-of-10-s...

sama also posted some stats here a while ago about yc companies that were wildly different from the "8/10 startups fail" myth.

[Edit: Adding specifics link on survival rates]

Shows the ~50% 5 year survival rate and goes out to 14 years: https://www.sba.gov/sites/default/files/FAQ_Sept_2012.pdf

[Original comment]I don't have specific slice of the data to link right now so summarizing how we dug in to look at business life for US small businesses. We focus on US 2-25 employee small businesses and as we dug into numbers at https://www.census.gov/econ/susb/ and https://www.sba.gov/advocacy/firm-size-data we found that each year ~500,000 businesses enter and exit the 2-25. Most of the entrances is from one of the 15-23 million sole proprietors adding their first employee and most of the exits are a business shutting down or returning to a sole proprietorship. In total for the category there are ~3.5M that employee ~40M people in total which would be an average life of 7 years.

Most 'statistics' cited in the media are completely made up, this isn't at all unusual. I spent about 18 months trying to track down the primary source of every interesting stat I could find, and most of them didn't pan out at all.

There is a saying that "reality has a well-known liberal bias" because most left-leaning publications base their arguments on statistics, whereas most right-wing arguments are based on emotion. But in fact most of the statistics that used to support left wing policy are either completely made up (e.g. "137 million American adults don't have dental insurance") or else wildly misleading (e.g. the stats on college tuition increases), to the point where most of the arguments being made are completely divorced from reality.

> There is a saying that "reality has a well-known liberal bias

you do know that was said by a comedian as a punchline in a standup routine, right?

Really? The only thing that comes up via Google is Colbert using it to satirize the right.

It was from Colbert's White House Correspondents dinner speech in 2006. Not quite standup.

No, that's pretty much the definition of standup.

{Shrug} Jokes+microphone doesn't make it standup. The content was closer to a roast than traditional standup. Especially given the satirical style, I wouldn't consider this to really be standup. But I also don't really care. I was just providing a reference to the specific event. Call it standup if you want.

The UK has a rather excellent radio programme devoted to explaining and debunking topical political and media statistics. Available as podcast with all back episodes for download.

http://www.bbc.co.uk/programmes/b006qshd

You might want to track down the primary source of that saying.

https://en.wikiquote.org/wiki/Stephen_Colbert

> because most left-leaning publications base their arguments on statistics, whereas most right-wing arguments are based on emotion. But in fact most of the statistics that used to support left wing policy are either completely made up

The claim you make in the first sentence above seems equally completely made up to me.

> The claim you make in the first sentence above seems equally completely made up to me.

Right, because it's outside the scope of science to put any sort of exact number on this. But how often do you hear arguments for right-wing positions (pro life, anti marriage equality, trickle down economics, abstinence-only education, war on drugs, etc.) that are actually based mostly or entirely on statistics? You certainly get folks who throw out a number or two are part of their talking points, but rarely do you get entire articles for these sorts of positions that revolve wholesale around some purported stat.

Whereas by definition you can't really have an article about rising college tuition without some sort of statistics about college tuition, regardless of whether or not they're accurate. Same thing for income inequality, access to healthcare, the quality of the school system, etc.

> that are actually based mostly or entirely on statistics? You certainly get folks who throw out a number or two are part of their talking points, but rarely do you get entire articles for these sorts of positions that revolve wholesale around some purported stat.

The current right wing UK government is at present busy churning out statistics to "prove" why it's best for the UK to remain in the EU, for example.

E.g most recently they've trotted out claim of worse GDP development outside the EU.

The pro exit crowd in their own party is using exactly the same tactics in favour of their views.

Both sides have focused strongly on numbers related to how it will affect the UK economy and immigration as those two areas seems to of greatest concern to the Conservatives.

When it comes to education too, the right wing here is extremely preoccupied with numbers.

Given your proposed list of "right wing positions", it may be that what you're seeing is arguments for predominantly religious positions that are basically too extreme for most European right-wing partes that'd be my frame of reference.

Pro life, anti marriage equality, trickle down economics, abstinence-only education are all policies that are extreme fringe positions even on the right wing in most European countries (with the odd exception - e.g. stronger support for anti-abortion bills in the catholic countries, though that's crumbling even in places like Ireland), and of those only "trickle-down economics" is not tied to religion here. And trickle-down economics certainly does not survive any kind of focus on numbers, so the few people here who believe in that certainly won't drag out numbers.

War on drugs is the only policy on your list there's fairly universal support for on the European right.

Instead what we tends to see are arguments over the economy, taxes and costs of immigration, where in fact the right very often will be trotting out numbers.

(That was a horrible experience trying to view the content; anyway.)

If you don't want to fail, don't try, but be sure you realize that no one else is going to try for you; that is, not trying results in 100% failure rate for you.

Next, new businesses based on an existing business model do not have the same types of risks and failure rates of startups in search of a business model that has never been done before; that is, this data includes not just startups, but new businesses, and as it relates to understanding startup failure rates is questionable.

> Not trying results in 100% failure rate.

It is worth noting that you are comparing two failures with very different personal impacts.

While not trying to create a business certainly ensures that you fail to build a business, you do at least avoid the much greater failure of trying to build a business and it driving you to bankruptcy and depression.

I am sure there are many people that would happily swap a 90% chance of a high-impact failure for a 100% chance of a low-impact failure.

Actually, a lot of people will try for you. If you don't do something, someone else will. It's an interesting question as to why we humans want to be personally responsible for the major changes in the world. I think we want to leave our mark because we want our name to be remembered. We want to have made a difference because we are concerned about our biological death. There is a book called Denial of Death that explores this theme, I think.

After all, why does it matter toa particular person if they are the inventor or originator or driving force behind a new technology that would have come about anyway? If money wasn't an issue, say, like for Nicola Tesla.

Do we really think someone else won't do it? Actually we usually think the opposite.

I used to think this, but I am still waiting for a few things which I may need to just create. Ex: A friend wanted 'grade A human chow' for the better part of 30 years.

Some are just slow; 10 years ago I was thinking a turbine electric hybrid could work really well. Another is you can extract energy from hot air and turbines and gas engines have really hot exhaust, the issue seems to be how nasty cooling hot exhaust is. Just look inside a cars tailpipe.

Actually we don't. Because the likelihood of someone else executing as well as you depends entirely how far out towards the edge of the innovation bell curve you are.

Beyond a certain point you can't be copied because you're literally the only person or company in the world who can execute.

Obviously this depends on talent. Mediocre talent is much more widely available than exceptional talent. So if you're not exceptional, you'll be copied.

If you're talented you can be first to market, but you'll be copied sooner or later.

But if you're truly exceptional you can't be copied at all, because you'll be combining exceptional imagination, exceptional execution, and exceptional access to resources. If you happen to combine those and provide something enough people want, you win your niche.

At this point you can only be imitated poorly. Sometimes that's enough to for competitors to survive around you. But it's a precarious existence, and if you continue to innovate you'll always be ahead of the competition.

I don't follow this line of thinking.

Almost any idea in any industry can be copied. Facebook was copied (and the copiers executed well), despite FB having "exceptional talent". Groupon was copied. Most new products are just copies of past products.

> Not trying results in 100% failure rate.

Depends on how you define failure. I define business failure as "losing my money in a failed business", and by that measurement, not trying results in a 100% success rate.

EDIT: Actually, if you want to be pedantic (this is HN after all), by this definition, not trying results in a 100% "non-failure" rate.

Wasting hours of your time could be another definition.

I don't know the source of the claim in the referenced Forbes article, but in my recollection the claim that 8 out of 10 businesses fail applied to restaurants – tho even a a cursory review of Google search results related to that don't seem to support it particularly strongly.

I came in to say this as well. I only this figure apply to restaurants not businesses in general.

I think this issue/misunderstanding roots in the somewhat loose interpretation of the word startup. A lot of my friends claim to have one, when all they really have is a business idea and maybe a couple of lines of code. Does that mean they are a real startup? By its definition, sure, but in reality what they have is more of a playground than a real business.

I had never heard the 8/10 statistic and always thought it was about half of normal businesses fail within 5 years, and 90% of startups fail within 3 years. Looks like fleximize supports the 1/2 statistic, and I found the data for the 90% one [1]. I think it's important to note that not many people talk about how many companies continue to operate in zombie mode, or how many are just small side businesses that don't really generate much revenue, but continue to exist for tax purposes (ex. my friends and I pretty much all have a business we use for occasional contract work).

[1] https://s3.amazonaws.com/startupcompass-public/StartupGenome...

Why couldn't you just let the back button be for god's sake? It makes no sense to override it like that.

The stats must vary a lot depending on how you define starting a business and failing. I made an app for example and it didn't go anywhere but I wouldn't call that starting a business.

Of course the actual life of the business idea and team is broader and not 100% the same, but it makes sense to count the days you are actually incorporated. You sign the papers with your government that your corporation starts on day X and died on day Y. So this data can be used to calculate your business's life time.

I was thinking about this as well. But, I think that is a bit of a problem. I had a startup that failed, but I didn't close it for a couple years with the hope that I would jump back in. I ended up starting a totally different company instead, but that first one survived for 4 years while it was really only around for 2. Since taxes are really easy when you aren't making any money, I didn't really have any reason to officially shut it down.

You got to wonder how often the case is that someone delays a year because there wasn't any reason to make it official.

Now I get what you mean. Yes, it's likely that corps are dragged along for some time even if there is no business happening. It's true that it would be hard to find out about a company that was closed after 3 years but was already dead in the middle of the first.

Except in the United States you don't need to incorporate in order to start a business. You can start a sole proprietorship any old day, and don't need to report it to the government until filing your schedule C -- and if your business never made any money at all, you don't even need to file.

From the IRS's perspective, I've started and failed at creating half a dozen businesses in the last ten years.

The claim that failure rates are the same YoY on the third slide is wrong.

The original Forbes article didn't claim a constant failure rate and it stands to reason a business that fails in the first 18 months simply failed to find market fit. That being the case, the survival rate would likely go up each year rather than down. [i.e. 18 months? 80% of the original total 36 months? 90% of the original total. Obviously not real numbers but it gets the idea across.]

Nice presentation? What tool do you use to builded? D3.js from scratch or another one?

yep, it's built on d3.js with custom animations and transitions

This has worse problems with credible authority than the Forbes article they're debunking. Forbes is at least a journalistic publication, this is clearly content marketing with an agenda to promote.

The stats for the UK were taken straight from the referenced ONS report and used in correct context. It also seems like a bit of a stretch to describe Forbes as journalism anymore

EDIT: obviously the suggestion they make that the claimed 80% failure rate is every year rather than just the first one was spurious, but overall a near 60% failure rate over 5 years would be enough to put me off getting a loan, so I'm not sure any marketing spin they are attempting actually works

How a marketing piece gets to the top of HN? I wouldn't take it too seriously, as it is a commercial for the publisher, encouraging taking buisiness loans, telling people it is quite safe.

There's actually similar common wisdom in Germany that most entrepreneurs fail in their third year because of how taxation works (having to pay taxes for the first two years plus having to pay an advance on taxes for the current year) and many beginners forgetting that they will have to pay taxes from their income.

OT: Interesting UI - What is being used?

By looking at the page's source code: seems to be a custom visualization created using d3.js.

Experience could be improved by allowing to navigate between the "slides" using space bar and by scrolling [0]...

[0]: How To Scroll - https://bost.ocks.org/mike/scroll/

Looks like d3.js -> https://d3js.org/

This ratio has not much meaning because it really depends of the context which is unspecified. Consider software projects where people just throw ideas at the wall to see what sticks. The ratio could be much worse than that.

What's does it mean for a business to fail/survive? That's not clearly defined.

That super-shiny interface made me entirely skeptical of everything they said. To me, the visual experience screamed "marketing BS". Anybody else have the same reaction?

Reply on news.ycombinator.com