22 Laws of Food & Drink Marketing You've Probably Never Heard Before - Rory Sutherland
Behaviour shifts by accretion, not decree: the intermediate possible — a pizza with meat as ornament, a heat pump sold as air conditioning, an alcohol-free beer that owns an occasion — beats the theoretical perfect because it starts where people already stand. Nobody switches sides after being humiliated, and beliefs are worn for the status they confer. Categories move when a lot of people change a little, and only once the default has quietly flipped.
Leave a CommentAbstract
A three-hour conversation ranging across food, drink and marketing, opening on comparison itself: bad news is fast and good news is slow, so an industry that looks beleaguered up close looks transformed at fifty years' distance. How healthy anything looks is set by what it is being judged against.
- The best marketing is divisive, because the alternative to dividing people is blandness nobody notices.
- Status cannot be surrendered the way money can, which is why no persuasion begins by humiliating anyone.
- Change happens through the intermediate possible rather than the theoretical perfect.
- Customers optimise against the worst outcome, so being reliably not-bad is what they are buying.
- A brand is defined by what it refuses, and scale kills the magic unless it is deliberately bought back.
- Delight is bought precisely where the spending makes no functional difference.
For someone building a personal brand, the hour reads as a catalogue of choices a solo operator gets to make directly. Nobody is averaging the opinion down, nobody is cutting the discretionary gesture, and nobody is handing the customer-facing decision to a function rewarded only for cost. The refusal, the frame, the one unexpected detail and the willingness to be divisive all survive to reach an audience intact — which is precisely what the organisations described here cannot manage.
Chapter summaries
00:40 - How healthy a market looks is set by the comparison it is judged against
How healthy a market looks is set by the comparison it is judged against.
"because You' got cost of living crisis cost of living crisis rates crisis rates crisis supermarkets not taking cost price increases"
— 00:40
Bad news is fast and good news is slow, so the short view registers only decline while the long view registers transformation. An industry in crisis up close looks unrecognisably better against the same sector fifty years earlier, when eating out was a wedding-anniversary event and you took sandwiches onto the train. The comparison is doing the judging.
09:51 - The best marketing is divisive, because the alternative is blandness nobody notices
The best marketing is divisive; the alternative to dividing people is blandness nobody notices.
"road opinion in other a qualified opinion with which most people could sensibly agree"
— 09:58
A moderate, qualified opinion gains no traction at all — and, stranger still, feels ridiculous even to type. The medium selects for the extreme, which means the alternative to being divisive is not being balanced; it is being invisible.
15:14 - People adopt beliefs for the status they confer, not because they think them true
People adopt beliefs for the status they confer, not because they think them true.
"the stability of your home background has a bigger determinant on your future success"
— 15:14
A belief that costs its holder nothing and signals membership of a high-status group spreads regardless of evidence, and is frequently not acted on by the people professing it. Family stability predicts outcomes more than wealth does — offered as statistics rather than destiny, which he says explicitly.
17:14 - Status cannot be surrendered the way money can, so a politics of status has no mechanism
Status cannot be surrendered the way money can, so a politics of status has no mechanism for changing minds.
"if your principal motivation is the redistribution of wealth and resources okay there are ways psychologically where you can get"
— 17:14
There are many ways to separate people from money and leave them feeling good about it — philanthropy, taxation, premium pricing. There is no equivalent mechanism for standing. Nobody has ever felt good about losing status, which makes a politics of status a dead end.
20:10 - Nobody ever switched sides because the other side mocked them
Nobody ever switched sides because the other side mocked them.
"great restaurants aren't always about food"
— 20:10
Diminishing somebody's status forces them to defend it, so no enduring persuasion has ever begun by humiliating the person being persuaded. A campaign that portrayed rival drinkers as apes was killed for exactly this reason — and rival supporters abuse each other endlessly and convert nobody.
28:57 - A behaviour change is permanent only once the default has flipped
A behaviour change is permanent only once the default has flipped.
"the point about the the kind of the extremism point we were talking about earlier about people jumping on"
— 28:57
Something that was the default and required a reason not to do becomes something that requires a reason to do, and that reversal does not revert. Office attendance is now consulted rather than assumed — some post-pandemic behaviours were a flash in the pan and this one was not.
34:59 - A new equilibrium cannot be decreed
A new equilibrium cannot be decreed: the intermediate possible beats the theoretical perfect.
"a huge amount of the reduction in meat consumption in the UK"
— 34:59
Evolution and adoption both proceed by what can be done next rather than by jumping to an imagined end state. Most meat reduction came from meat as garnish rather than from new vegetarians; a heating upgrade done one room at a time beats ripping out the system. Start where you are.
49:38 - Location no longer gates quality, so the interesting end forms where property is cheap
Location no longer gates quality: a category's interesting end forms where property is cheap.
"I don't think that the advantage that London confers on consumers relative to the rest of the country like smaller provincial towns"
— 49:38
Once consumption is geographically neutral, the advantage of the capital collapses and experimentation happens wherever rent allows it. Specialist food, film and equipment are available anywhere now, and a coastal strip can have the more interesting restaurant scene — employment opportunity remains the real advantage.
57:31 - Manufactured scarcity turns customers into the venue's distribution channel
Manufactured scarcity turns customers into the venue's distribution channel.
"there're two great creative tricks if you want to move to Suburbia and stay cool"
— 57:31
A very small number of seats creates a booking worth announcing, and the announcement is the marketing. Scarcity converts every customer into a distribution channel, which no advertising budget buys as cheaply.
1:03:47 - When an activity depends on its location, the landlord takes the gain
When an activity depends on its location, the landlord takes the gain.
"it's a strange economic belief which means that the principal sort of form of tax should be on the value of unimproved land"
— 1:03:47
Most of the value created by work performed in a place accrues to whoever owns the ground, so the escape is becoming somewhere people travel to. Luxury retail in an airport enriches the landowner; a derelict site with parking lets the operator keep what they make. The land takes the gain.
1:09:11 - Expectation is part of the product
Expectation is part of the product: words and price format change what people actually taste.
"that's not a great restaurant is it if I I mean measured by my form of"
— 1:09:11
The best meal anybody names is almost always one they ate exactly once, which is the tell. A place people return to repeatedly is being judged on the food; a place visited once may have been sold on the telling of it.
1:23:12 - Customers optimise against the worst outcome, so being reliably not-bad is what they are buying
Customers optimise against the worst outcome, so being reliably not-bad is what they are buying.
"it's like we we're waiting for that kind of what we expect and we and we're almost like speeding up to what we actually get"
— 1:23:29
Perception is mostly prediction corrected by the senses, so the frame a food arrives in becomes part of its flavour. Provenance and adjectives raise what people will pay; a citrus-looking sweet flavoured as strawberry produces a taste that exists nowhere in nature. Expectation is an ingredient.
1:28:47 - What sells and what sells repeatedly to the same person are different numbers
What sells and what sells repeatedly to the same person are different numbers.
"there is probably satisficing something"
— 1:29:09
The avoidance of a negative looms larger than the attainment of a positive, which makes predictability the product rather than a compromise. Chains succeed by never being bad, which is why people eat in them in cities full of better food — and why a sandwich in an unfamiliar town is a gamble nobody wants to take.
1:36:17 - Scale kills the magic, so growth comes from a category bounce or from a chain that
Scale kills the magic, so growth comes from a category bounce or from a chain that hides being one.
"shopping now has changed"
— 1:36:17
Volume alone cannot distinguish a hit from a habit, which is why loyalty data is worth what it is. A retailer keeps low-volume items because their repeat rate anchors an entire basket — somebody orders a hundred pounds of shopping to get the one thing they cannot buy elsewhere.
1:49:51 - Brands are killed by cost decisions made by people who own none of the consequences
Brands are killed by cost decisions made by people who own none of the consequences.
"counter now what a lot of supermarkets have done is they've gone God it's cheaper to actually have self-service checkout"
— 1:50:13
Either you take the brand into a scalable adjacent category, or you keep the sites and vary them enough that each reads as a local. A bakery moving into homeware rather than opening more shops; chains that make every outlet look different and source one item locally. Deliberate inconsistency works once established, not at the start.
1:55:48 - Customers care about status, certainty, autonomy, relatedness and fairness
Customers care about status, certainty, autonomy, relatedness and fairness — none of which economics measures.
"nobody expects you know you know a personal welcome in a Motorway service"
— 1:55:48
A function rewarded purely for saving money takes decisions whose customer cost lands somewhere else entirely. Car-park enforcement that fines people on a technicality; self-checkout rollouts whose consequences arrive years later. Enrage somebody three times in a row and you have lost them for good.
2:04:06 - Category-defining products are impositions rather than answers to stated demand
Category-defining products are impositions rather than answers to stated demand.
"I love and I think it's so applicable to marketing is is the audience comes last"
— 2:04:35
Each of the five — status, certainty, autonomy, relatedness, fairness — produces a reaction far larger than poor food does when violated. Being told an item is off after ordering rather than before; choosing your own table; recognition of past custom; who gets served first.
2:11:01 - The same food reads as a delicacy or as destitution depending on the frame it arrives
The same food reads as a delicacy or as destitution depending on the frame it arrives in.
"quite a lot of food is the beneficiary of associations rather than the food itself"
— 2:11:01
Averaging opinion produces something that appeals to nobody, because nobody is average. A fermented drink nobody would have endorsed; a fast-food format that imposed limited choice on a category built entirely around customisation. A group is valuable for dissent, not for its mean.
2:15:39 - Narrow focus signals competence, and the format follows the situation the eater is in
Narrow focus signals competence, and the format follows the situation the eater is in.
"you could revive that quite easily because again it's rather like my bacon sandwich idea if you just focus on one thing"
— 2:15:39
Association rather than the food itself carries most of the value, which is why the same dish reads as a delicacy or as destitution depending on where it is said to come from. Raw fish in rice described with and without its origin produces two completely different reactions.
2:19:16 - Skill improves at the speed of the feedback loop, so food teaches marketing fastest
Skill improves at the speed of the feedback loop, so food teaches marketing fastest.
"one thing so with your book and um is this is more for like kind of our listeners who who analyze data to make"
— 2:19:13
Doing one thing implies doing it well, and the format follows the circumstances the food is eaten in. Single-item concepts trade entirely on that inference — and a whole national tradition of portable food exists because of where people were when they ate it.
2:24:21 - Value lives in the combination, and one trivial defect can cancel the whole of it
Value lives in the combination, and one trivial defect can cancel the whole of it.
"there's a lot of parallels between cooking and marketing"
— 2:24:21
Comedy improves because laughter is immediate; a trade with slow feedback cannot be learned the same way at all. Restaurants against banking, where a customer-service change takes years to evaluate — which is why the slow business should be stealing from the fast one.
2:32:30 - Originality is mostly theft: a proven, unglamorous format made posh
Originality is mostly theft: a proven, unglamorous format made posh.
"you just have sell two things you sell wings and you sell Tinys"
— 2:32:30
Components that are unremarkable alone become remarkable together, and a single misaligned element cancels the whole. Broth as the thing that makes the separate parts of a dish cohere — and an excellent meal ruined by one bent fork. The combination is the product.
2:34:45 - Delight is bought precisely where the spending makes no functional difference
Delight is bought precisely where the spending makes no functional difference.
"that business of setting the expectation is really interesting"
— 2:34:45
Nearly everything that works is something else recontextualised, and the unfashionable version usually already had the proof points. High-end bakeries were built on a format ordinary towns had been queueing for all along — the innovation was making it posh, not inventing it.
2:41:52 - Taste is made of context, so a product taken out of its setting collapses
Taste is made of context: a product taken out of its setting collapses.
"there's basically every single every single touch Point there's there's there's there's a unique story to tell"
— 2:41:52
Basic attributes must not fail and performance attributes return linearly, but the unexpected third kind produces an emotional effect out of all proportion to its cost. A damped eject mechanism that changes nothing functionally; unusually pleasant spoons that were the only thing anyone talked about.
2:46:26 - A substitute wins by owning an occasion, not by being a lesser version of the original
A substitute wins by owning an occasion, not by being a lesser version of the original.
"people drive to a golf club so there's a limit how much you can drink"
— 2:46:26
Some foods and drinks are enjoyable only in the circumstances that produced them, which is why the import fails. A drink bought abroad that nobody can finish at home — and the real answer to why fish restaurants cluster by the sea is not distribution cost.
2:52:56 - Obvious innovations arrive late because the enabling part is being built in another industry
Obvious innovations arrive late because the enabling part is being built in another industry.
"technology is often dependent for its adoption on something else"
— 2:52:56
Two unrelated trends can reinforce each other, and a brand that claims an occasion rather than a category escapes the comparison entirely. Alcohol-free beer grew alongside sleep tracking; a wine positioned around one specific day of the week is selling the day rather than the drink.
2:59:52 - The blocker is almost always a marketing problem, not a technological one
The blocker is almost always a marketing problem, not a technological one.
"a lot of it's been done by the military which is M meals ready to eat which is the military category"
— 3:00:07
A technology waits for an unrelated advance before it becomes possible, which is why the simple thing so often follows the complicated one. Wheels on suitcases arrived after a man reached the moon, enabled by improvements made for something else — and the blocker is almost always framing rather than engineering.
Personal Branding Lessons
Everything in this episode that gets decided badly gets decided by someone who owns none of the consequences — the procurement manager who removes the till, the accountant who hates the unsellable local burger. Working alone spares you that, and replaces it with something harder: you are the one who would veto the idea on cost, the one who wants proof before committing, and the only thing on the table a competitor cannot buy a copy of. Nine things to do about that this week.
Post the sentence you feel ridiculous typing. The experiment in the episode was writing a reasonable, qualified opinion and getting nothing back — no replies, no argument, not even a reaction, and a distinct sense of foolishness at the moment of pressing send. You are competing in an attention market that pays out on disagreement or on allegiance, and the moderate middle supplies neither, so it is not that your balanced take underperforms — it is not entered. Take the post you softened last week, restore the version you actually believe, and accept the trade: fourteen likes from people who already know you, swapped for a comment thread you have to spend the evening in. 09:51
Ask the person you disagree with for a line in reply. Nowhere in the episode does humiliation persuade anyone: the beer ad portraying rival drinkers as apes was killed before it ran, and no football fan on record has ever defected to the club that waved money at them. So quote the strongest version of the view you are arguing against, credit whoever made it, and then ask them to respond — people who have done you a small favour tend to conclude afterwards that they rated you. Their answer lands in their own feed, where their readers meet you as someone worth replying to rather than someone worth blocking. 20:10
Bolt the new thing onto what already runs. Meat consumption fell not because people became vegetarians but because the meat turned into a few slivers of pepperoni on a pizza, and the workable heating fix was a two-thousand-pound unit installed alongside the boiler rather than a fifteen-thousand-pound excavation of the lawn. Your version is not the rebrand, the new name and the migrated archive; it is one new section inside the letter you already send, using the audience and the habit you already have. The test is whether the change survives to week eight, not whether it looks complete on the day you announce it. 34:59
Make yourself boring to predict. The global burger chain is not the best restaurant in the world, it is extraordinarily good at not being bad — the bun toasted for a minute and thirty-nine seconds, no surprises, exactly as it was last time. A stranger's first question about your work is not whether it is brilliant but whether it is safe to spend eight minutes on, and how they decide whether to believe you is largely settled before your argument begins. Fix the day, the length and the shape for three months, and the marker of success is someone opening the email without checking the subject line first. 1:23:12
Count the people who came back, not the people who arrived. Restaurants know what sells and not what sells repeatedly, which is the whole reason supermarkets want you carrying a loyalty card; the online grocer that knows every buyer runs two valuations and keeps lines the volume alone would not justify. Your view count is the first number and nearly useless — keep a running list of names that reply, forward or subscribe more than once, and read it monthly. It will tell you to keep the quiet essay thirty people return to and to stop making the thread forty thousand people scrolled past once. 1:28:47
Write down what you will never publish. The agency game for discovering what a brand is ran by deciding what it isn't — the mint could plausibly become chewing gum, and bubble gum was so totally inappropriate that the boundary appeared on its own. Do it on a page: three formats, three topics and three kinds of paid work you refuse, with a one-line reason for each that a reader could repeat. The payoff is the pitch you decline in a single sentence, and an audience that can name the thing you never touch. 1:43:29
Finish it to your own standard before you ask anyone. The record producer's line is that the audience comes last, and the reason is mechanical: research can only report preferences assembled out of what already exists, which is why the market-researched version of a category-defining act would never have been made. Write the piece you would want to read even if it gets buried, then publish it and watch, rather than polling first and building to the average. What you are looking for is the format nobody requested turning into the thing people introduce you by. 2:04:06
Overspend where it makes no functional difference. The eject mechanism that hisses out on a damper instead of going clack does nothing the clack does not do, and it is the part people remember, because the delight sits exactly where no effort was expected. Pick the one surface a subscriber touches after the value has been delivered — the reply you send to a first-time commenter, the page they land on after confirming, the last line of the letter — and make it unreasonably good, since the ending is one of the few slots memory reliably fills. The result you are after is a screenshot of your confirmation page turning up in somebody else's feed. 2:34:45
Assume the blocker is framing until something proves otherwise. Long-life food is technically solved — the preserving work was done for space agencies and armies, and it tastes delicious — and the freezer is still read as downmarket because it is cheap and efficient, which is why the episode lands on it being a marketing problem rather than a technological one. When nobody takes the free thing you built, the thing is rarely the fault: what matters is what accepting it says about the person accepting it, since the final question is what people think of them for doing this. Rename the free audit as something a senior stranger could admit to signing up for in public, and the difference shows up as names in a calendar rather than a better landing page. 2:59:52
Questions
Each answer ends at the moment in the recording where it is given.
Each answer ends at the moment in the recording where it is given.
Why do my balanced opinions get no engagement?
Nothing comes back because neither of the engines running that platform is switched on: there is nothing to argue with, and no tribe to signal membership of by sharing it. A moderate position is selected against in the same way a weak organism is, so louder and slightly ridiculous ones out-compete it. The flattening starts earlier than publishing, though — a qualified sentence feels faintly ridiculous while it is being typed, so most are never sent at all. 10:27
Why do people share beliefs they clearly do not act on?
Holding the belief pays in social standing, and that is the whole return; whether it is true or useful is secondary. Advocacy is cheapest for whoever is insulated from the consequences, which is why a position gets fashionable precisely among the group it will never touch. The practical limit: stated positions are a poor guide to what an audience will actually do, so anything built on what people say they believe should be checked against what they buy. 16:10
Does attacking a competitor ever win over their audience?
No recorded case exists of anyone defecting to the side that mocked them. Abuse across a rivalry is potent, but it is aimed at the in-group watching, not at the people being insulted — it buys applause at home while hardening everyone else. Specific behaviours can be stigmatised successfully; identity and founding myths cannot. The inverse move, raising someone else's standing, has a real mechanism behind it, where lowering your own has none. 19:09
Is it better to ask my audience for a small change or a big one?
Smaller, nearly always: the version that keeps the reward and removes most of the harm gets accepted where outright cessation gets refused. Vaping over quitting, a beginner's programme named for the sofa its recruits start from — the volume sits in moving people off zero, not in pushing the already-committed further. It works by accelerating something already being done rather than reversing it; holding up the perfect ideal mostly pushes beginners out. 47:32
Does being dependable matter more than being brilliant?
For most of the audience most of the time, yes. Beneath the choosing sits an old question about whether this is safe, and avoiding a bad outcome weighs heavier than winning a good one, so the thing that is reliably not-bad collects the return visit. What distorts the picture is that only the remarkable gets written about. The usable version: the number worth watching is repeat custom, not first-time volume, and almost nobody measures it. 1:25:10
Should I narrow my brand down to one subject?
Narrowness reads as evidence of competence before anyone can verify anything, because quality cannot be checked in advance and the scope of the offer stands in as a proxy. Claiming two specialisms triggers the opposite reading — both halves get discounted. Going one layer narrower is the safer move than going broader. The limit is that this is a caution rather than a prohibition: combinations do work, they simply carry a penalty to overcome. 2:15:46
How do I get better at this faster?
Skill improves at whatever speed the verdict arrives. Comedy is the extreme — laughter or silence within seconds — and restaurants sit just behind it, which is why operators learn from failures quickly while a bank waits years to find out whether a change worked. The length of the customer relationship sets that clock. So work done in the open, where reaction is immediate, compounds faster than the same work done privately and judged later. 2:20:12
Where should I spend effort once the basics are covered?
On the parts that make no functional difference at all. Three kinds of attribute exist: the ones that only avoid complaint when present, the ones with a roughly linear return on improvement, and the unexpected ones, which land an emotional effect wildly out of proportion to their cost precisely because nobody was anticipating them. The condition is order: a basic that fails cancels everything above it, and a missing expected element reads as being cheated. 2:37:12
Sources
22 Laws of Food & Drink Marketing You've Probably Never Heard Before - Rory Sutherland
The sections above follow the episode's own order. Each timestamp is the point in the recording where the idea is discussed, so the post reads as a map of the conversation rather than a rearrangement of it.
This recording runs over two hours, so the map is drawn at a coarser grain than a shorter episode would need: each section covers several minutes of conversation rather than one turn of it.
Quotations come from the episode transcript with spoken filler ("okay", "right", "you know", "um", "actually", "kind of", "sort of") and false starts removed. No words are added: square brackets mark anything inserted for sense, and an ellipsis marks any cut. British spellings are restored, and speech-to-text errors were repaired against context only where the intended word is unambiguous. Where a line appears both in a cold-open teaser and in its place in the conversation, the timestamp cites the second — the point at which it is actually said.
Quotes are stamped with the moment they are said rather than with a speaker's name.