The 11-3-6 rule of friendship, and where the numbers actually come from

A tidy formula has been going round: eleven meetings, three hours each, six months. I went looking for the study behind it. What I found was a cough sweet campaign, a rounded number, and a peer-reviewed figure nearly three times higher.

If you have seen the 11-3-6 rule, it probably arrived without a source. It is the kind of thing that spreads well precisely because it sounds like research: three specific numbers, an implied study, no citation attached.

So I traced it. The numbers are real and they do come from a real survey. But almost everything about how they are being repeated is wrong, including the arithmetic, and the peer-reviewed answer to the same question is nearly three times higher.

What the 11-3-6 rule of friendship says

The 11-3-6 rule claims that turning an acquaintance into a real friend takes about 11 separate meetings, of roughly 3 hours each, spread over about 6 months. Multiply the first two and you get 33 hours of time together as the price of a friendship.

You will also see it written as the 11-6-3 rule, which is where it starts to come apart, because that ordering is usually attached to a completely different claim. More on that below.

The numbers come from a survey commissioned by a cough sweet company

The source is a study commissioned by Fisherman's Friend, the menthol lozenge brand, and published on 5 October 2022 to support a marketing campaign called "Whatever the day throws at you".

It was not a small or shady piece of work. The fieldwork was run by Walr, a Market Research Society member, among 2,001 nationally representative UK adults aged 18 and over, between 12 and 14 September 2022. The results were then analysed by Robin Dunbar, Emeritus Professor of Evolutionary Psychology at Oxford and the person behind Dunbar's number.

Here is what the release actually reported:

Figures as published in the campaign release, October 2022.

Source: Richmond & Towers, "What's in a friend?", 5 October 2022, the agency release for the Fisherman's Friend study. Online survey conducted by Walr among 2,001 nationally representative UK adults aged 18+, fieldwork 12 to 14 September 2022. Analysis by Professor Robin Dunbar. Note that this is brand-commissioned market research rather than peer-reviewed academic work, and it is a UK sample.

So the "6" is a rounded 5.5, and the release itself sold it as a formula

Line the survey up against the rule and you can watch the compression happen. Eleven interactions became the 11. Three hours and four minutes became the 3. And 5.5 months got rounded up to 6.

There is no six among the three headline figures, and 5.5 does not sit well in a hashtag.

The numbers do not quite reconcile either. Eleven interactions of 3 hours 4 minutes is 2,024 minutes, or 33 hours 44 minutes. The release headlines 2,040 minutes and 34 hours. That is a small gap, but it is the release's own arithmetic, and it is why you will see the rule quoted as both 33 and 34 hours depending on who is repeating it.

What I expected to find, and did not, was that the prescriptive framing had been added by the internet. It had not. It came from the release. The subheading reads "New study commissioned by Fisherman's Friend reveals the ultimate formula for friendship". The body says Brits "require" 34 hours of commitment, that each interaction "needs to last on average 3 hours and 4 minutes to be mutually beneficial", and asks "what is the ultimate formula for finding true friendship?" before answering that 11 interactions are "required".

So the rule-shaped thing was there from the start. That matters, because it moves the criticism from the people sharing it to the people who published it.

And the criticism is a change of kind, not of arithmetic. A survey can tell you what friendships averaged. It cannot turn that average into a requirement. "People who became friends averaged eleven meetings" says very little about what happens if you deliberately schedule eleven meetings with someone, in the same way that knowing the average marriage involved a certain number of dates does not hand you a procedure.

One caveat on my own reasoning here. The release publishes no question wording, so I am inferring that respondents were estimating retrospectively on friendships that had already formed. Figures of this kind are difficult to produce any other way, but I have not seen the questionnaire and cannot confirm it.

There is a second version, and it matches nothing at all

Search for the 11-6-3 rule and you will often get something different: the claim that a healthy social life means 11 casual friends, 6 good friends and 3 close friends.

This one has no origin study, and the pages pushing it broadly admit as much. The most prominent of them credits it not to the Fisherman's Friend survey but to Dunbar's layers, describing itself as a practical distillation of them.

That is where it falls apart, because the distillation does not follow from the thing it claims to distil.

• Dunbar's layers contain no 11 and no 6. They run 5, 15, 50 and 150. You cannot get 11 casual friends and 6 good friends out of that sequence by any route the pages explain.
• The Fisherman's Friend survey found 5 close friends, not 3.
• Dunbar's own layers are 5, 15, 50 and 150: about five intimate friends, fifteen good friends, fifty social friends and a hundred and fifty meaningful relationships in total. There is no 11 and no 6 anywhere in that structure.

There is a third reading in circulation too, which has 11 interactions over 6 months leading to a friendship that lasts 3 years. Three incompatible versions of the same three digits is a reasonable sign that nobody is working from a source.

The point worth holding onto: no research anywhere produces 11, 6 and 3 as counts of friends. If you have been measuring your social life against that version, you have been measuring against nothing.

Where it conflicts with the peer-reviewed number

There is an actual academic answer to this question, and it is not 33 hours.

Jeffrey Hall, then an associate professor of communication studies at the University of Kansas, published the most cited modern attempt to put hours against the stages of friendship in the Journal of Social and Personal Relationships. His university billed it as the first study of its kind, though the paper itself cites a 1965 study that had already estimated casual friendship at around 60 hours. He ran two studies: a survey of 355 adults who had moved to a new city within the previous six months, and a longitudinal study of 112 university freshmen, first surveyed two weeks after they arrived and then followed up four and seven weeks after that.

These are the bands Hall states in the paper's own discussion, drawing on both of his studies. You will more often see them quoted as 50, 90 and 200, which are roughly the midpoints and come from the University of Kansas press release. His individual study estimates vary a good deal either side: the acquaintance to casual friend transition came out at 43 hours in one study and 94 in the other.

Set them side by side and the gap is hard to miss. The rule puts "friend" at 33 to 34 hours. Hall puts the casual friend band at 40 to 60, so the rule's total does not quite buy you the first rung, and the milestone it claims takes 80 to 100 hours.

One caveat I should be straight about, because it cuts against the neatness of that table. Hall's bands are tied to time windows, roughly the first three to six weeks after meeting, and he notes that hours accumulated after about three months stop improving the odds of becoming casual friends. The survey's figure sits inside an average of 5.5 months. So this is not quite the like-for-like comparison the two columns imply, even though both are anchored to the same acquaintance to friend milestone.

Source: Jeffrey A. Hall, "How many hours does it take to make a friend?", Journal of Social and Personal Relationships, summarised by the University of Kansas, March 2018. US sample.

Which number should you actually use

Hall's, but with less confidence than most articles quoting him suggest, including the ones on this site.

The reasons to prefer it are straightforward. It is peer reviewed. It is two studies rather than one, and the second is longitudinal, following the same relationships as they formed rather than asking people to remember. It reports its own confidence intervals. And it was not commissioned to sell lozenges.

But Hall is markedly more careful about his own numbers than his popularisers are, and it is worth borrowing his caution. He notes three limitations on his first study: hour estimates get less accurate the longer people have known each other, the hours kept accumulating after the transition had already happened, which inflates the estimates, and only friendships that survived were available to study in the first place.

That last one deserves attention, because it is the same survivorship problem the Fisherman's Friend survey has. Neither piece of research can see the friendships that quietly failed. Both are, in that specific sense, describing the winners.

What both versions leave out, and it is the part that matters

Neither version says anything about what kind of time it is counting. Hall does, and it is the most practically useful thing in his paper.

Hours spent working together just don't count as much.

That line is the University of Kansas summarising him rather than Hall's own wording, but it is a fair summary of the finding underneath. In both of his studies, the proportion of time spent together at work or in class predicted less closeness, while time spent hanging out, joking around and sharing leisure activities predicted more. Hours are the headline number, but not all hours pull in the same direction.

One thing to be precise about, since almost everyone gets it wrong and I had it wrong in my first draft of this article: Hall's 40 to 60 and 80 to 100 hour figures are total hours spent together, not leisure hours only. The activity finding is a separate result about which hours do the work. It does not turn the thresholds into a leisure-hours budget.

Still, the direction is clear enough to be useful. You can sit near a colleague for three years, share something like 6,000 hours of the same room, and remain colleagues. And eleven three-hour work meetings will do far less for a friendship than eleven three-hour afternoons would, which is exactly the distinction the rule leaves out.

This is also why the rule's one genuinely good instinct is the three-hour block. Long, unhurried, unstructured stretches are exactly the shape of time that Hall's data rewards. A rushed coffee is not a third of a three-hour afternoon in terms of what it does for a friendship.

What I would take from all this

The instinct behind the rule is sound even though its numbers are not. Friendship needs repetition, it needs decent stretches of time rather than fragments, and it needs months. Eleven meetings over half a year is a reasonable picture of what showing up looks like.

What the rule gets wrong is the promise. It implies a countdown, and that if you do the eleven you have earned the friend. Hall's numbers put the real figure at roughly two and a half to three times that, and add that the type of time matters as much as the total.

The honest version is less shareable. Making a friend as an adult takes something in the region of forty to sixty hours together before it feels casual, and eighty to a hundred before it feels real, with the leisure hours doing most of the lifting. That is a lot of Saturdays with the same people, which is most of the reason it is hard, and it has very little to do with whether you are likeable.

It is also the reason we build Circles rather than run one-off mixers. The same four to six people, meeting repeatedly, is a structure designed to accumulate exactly the kind of hours Hall's data says count. A single event, however good the night is, cannot get anyone near forty.