On this page
Key takeaways
- The BLS survival table used here gives no national bar-only failure rate, and 34.4 percent of 2013 accommodation and food services establishments had closed three years later.
- Confident figures like sixty percent in the first year are repeated widely, usually without a traceable bar-specific source.
- Failure rates are hard to measure because closure is not failure, the denominator moves, and surveys only reach venues still trading.
- Four mechanisms recur: the business is a person, margin leaks faster than it is measured, knowledge leaves continuously, and growth exposes what was never written down.
- The common factor is a business that cannot see itself quickly enough to correct while correction is still cheap.
What percentage of bars fail? The published federal survival series does not give a national bar-only answer. The Bureau of Labor Statistics' 2013 establishment cohort shows that 65.6 percent of accommodation and food services establishments were still operating in 2016, three years later. That leaves 34.4 percent no longer operating. This category also includes restaurants and lodging, and a closure is not necessarily a business failure. It would be misleading to call 34.4 percent the bar failure rate.
Search for how many bars fail and you will also find confident numbers: sixty percent in the first year, eighty percent within five. These figures get repeated across trade press, vendor blogs and business advice sites, usually without a traceable bar-specific source. We will not substitute one of them for a measurement. What follows is an argument about mechanism, which is a weaker kind of claim than a measured failure rate and is labelled as such.
Why are bar failure numbers unreliable?
Three problems make bar failure rates hard to measure honestly.
Closure is not failure. A venue that closes because the lease ended, because the owner retired, or because it was sold and rebranded appears in the same column as one that ran out of money. Most datasets cannot separate them.
The denominator moves. Counting failures requires knowing how many venues existed to begin with, across a category with no agreed boundary. Whether a restaurant with a good bar counts changes the answer substantially.
Survivorship shapes who is asked. Surveys of operators reach operators who are still trading. The ones who closed are the population you actually wanted, and they are the hardest to reach.
None of this means bars do not fail at a meaningful rate. It means the confident percentage is doing rhetorical work rather than analytical work, and you should be suspicious of anyone who leads with it, including anyone selling you software.
What can we say with more confidence?
The mechanisms are better established than the rates, because they are visible from the inside of a business rather than requiring a census of a whole category. Four recur.
1. The business is a person
In most independent bars the standard is held by an individual. That works well while they are present and degrades when they are not, which is not a character flaw in anyone but a structural dependency.
The failure mode is slow. The owner takes less time off than they should. Quality dips when they do. They conclude they cannot step back, which is correct given the arrangement, and the arrangement never changes. Three years of that is exhausting, and exhaustion closes more venues than any single bad quarter.
2. The margin leaks faster than it is measured
Bar margins are thin enough that a few percent of variance is the difference between a good year and a hard one. The measurement of that variance typically arrives monthly, as one figure covering four weeks.
A business losing money at a rate it can only observe monthly can lose a meaningful amount before the observation arrives, and cannot attribute it once it does. That is not the same as being badly run. It is being run with a slow instrument.
3. Institutional knowledge leaves continuously
Hospitality turnover is high. Every departure takes some amount of unwritten knowledge with it: which supplier is unreliable, which regulars matter, what went wrong last time this happened.
One case study client, a Houston nightlife group with three venues, reported a 73 percent reduction in staff turnover, which they attributed to LearnTAP. That is a client reported figure from a single group and we would not present it as an industry number. What it illustrates is that retention and knowledge retention are the same problem, and most bars have no mechanism for the second one at all.
4. Growth exposes what was never written down
The second site is where an implicit standard becomes a visible problem, because the person holding it can no longer be in both rooms. Groups that expand on the strength of one excellent venue frequently find the second one is merely acceptable, and that the first has got worse because attention moved.
What actually reduces the risk
None of the four mechanisms above is solved by working harder, and three of them are made worse by it.
The common factor is that the business cannot see itself quickly enough to correct while correction is still cheap. A monthly instrument on a nightly problem produces a business that is always responding to something that finished weeks ago.
Making the loop shorter is not a technology argument in itself. A disciplined operator with point of sale exports and a weekly hour can do most of it. The reason it does not happen is that the assembly work has to be redone every week, by someone who is also running a bar.
That is the case for tooling, and it is narrow. It removes the assembly step so the routine survives past week four. It does not supply the judgement, and anything claiming to is overselling.
The honest summary
Bars fail for reasons that are visible from inside the business well before they become terminal, and that are usually observed too slowly to act on. That is a claim about mechanism, argued rather than measured, and you should weigh it as such.
The research this company has actually done is on the research page: 13,063 guest reviews across 41 Texas venues, classified twice by separate models and checked against state alcohol tax filings. It is a narrower claim than a failure rate, and it is one we can show you the working for.
Frequently asked questions
What percentage of bars fail?
The published federal survival series does not give a national bar-only answer. In the Bureau of Labor Statistics 2013 cohort, 65.6 percent of accommodation and food services establishments were still operating in 2016, leaving 34.4 percent no longer operating. That category includes restaurants and lodging, and a closure is not necessarily a failure, so it would be misleading to call it the bar failure rate.
Is it true that most bars fail in their first year?
Figures like sixty percent in the first year or eighty percent within five get repeated across trade press, vendor blogs and business advice sites, usually without a traceable bar-specific source. The article does not substitute one of them for a measurement. Be suspicious of anyone who leads with a confident percentage, including anyone selling you software.
Why are bar failure statistics unreliable?
A venue that closes because the lease ended, the owner retired or it was sold appears in the same column as one that ran out of money. Counting failures also needs an agreed count of how many bars existed, and the category has no agreed boundary. Surveys reach operators who are still trading, not the ones who closed.
What actually causes bars to close?
The article argues four mechanisms recur, labelled as argument rather than measurement. The standard is held by one person, margin leaks faster than monthly reporting can show, institutional knowledge leaves with every departure, and a second site exposes standards that were never written down.
How can a bar owner reduce the risk?
Shorten the loop so the business sees itself while correction is still cheap. A disciplined operator with point of sale exports and a weekly hour can do most of it. Tooling helps by removing the weekly assembly step so the routine survives past week four, but it does not supply the judgement.