A companion piece to Episode 9 of the BMD marketing case-study series. Episode 8 was about a company that changed who it was talking to. This one is about a company that had stopped being trusted, and decided to stop arguing about it. The video tells the story in minutes. This is the full autopsy: what Cadbury had already closed before the famous ad ran, what else was running beside it that autumn, and why one campaign carries at least four different sales figures.
Contents
The ninety seconds everyone remembers
On 31 August 2007, in the ad break of the Big Brother final on Channel 4, Cadbury ran ninety seconds of a gorilla sitting at a drum kit. Phil Collins builds in the background. The animal breathes, flexes its jaw, waits out the whole long intro, and then plays the fill from In the Air Tonight. Chocolate appears in the last two seconds.
No voiceover. No product claim. No taste shot, no ingredient, no celebrity, nothing a marketing director could defend in a meeting by pointing at it. The film came from Fallon London, directed by Juan Cabral, and Cadbury released it under a banner it had invented for itself: Glass and a Half Full Productions.
The story people tell about it has hardened into a single sentence. Cadbury was dying after a salmonella scare, it ran a brave ad about nothing, and sales jumped. Be emotional, be brave, and the market forgives you.
Two things are missing from that sentence. Everything Cadbury had finished doing before 31 August, and everything else it did that autumn.
What actually broke
In June 2006 Cadbury told the public that a leaking waste-water pipe at a British plant had dripped bacteria into the chocolate crumb used across several bars. More than a million bars were recalled in the United Kingdom and Ireland, at a reported cost of around 30 million pounds. The Health Protection Agency later linked a salmonella outbreak affecting around 37 people, most of them children, to the contamination.
That is the worst shape a trust failure can take for a food company. Not a taste problem, not a price problem. A safety problem, with children in it, at a brand whose entire meaning is childhood.
On 16 July 2007 the company pleaded guilty in Birmingham Crown Court and Herefordshire's own prosecution, and paid fines totalling around a million pounds plus costs. It had already spent tens of millions on preventing a repeat. Six weeks later, the gorilla sat down at the drums.
Here is the number the popular story skips. At the start of 2007 Dairy Milk was entering its third year of decline. The fall began around 2005, before the pipe leaked. Salmonella made an existing problem acute, visible and legally expensive. It did not start it, and a campaign that only answered the recall would have been answering the smaller half of the problem.
Why they stopped arguing
The instinct after a safety failure is to answer it. New standards, new testing, a quality message, a line about how seriously we take this. Every one of those sentences repeats the accusation. You cannot argue your way out of a trust failure, because the argument itself reminds people what they are being asked to forget.
So the brief went the other way. Not a better claim about the chocolate. No claim about the chocolate at all. Restore the thing Dairy Milk used to carry before any of this, which was joy, and say nothing else.
It was not an easy sell inside the building. Cadbury's marketing director at the time has since described it as the hardest idea he ever had to get approved, and the ad came close to never being made. That is the part of the story that gets told as courage, and courage is the least useful lesson in it, because courage is not a process anyone can copy on Monday.
What made the idea defensible was not bravery. It was sequence. The pipe was fixed, the prevention money was spent, and the company had stood up in court and taken the fine six weeks before it asked anybody to feel warm about it again. The fault was closed, and closed on a date you could point to.
31 August, and what ran beside it
The gorilla went out in the Big Brother final, which in 2007 was still one of the few places in Britain where a large number of people watched the same ad break at the same time. It spread from there on its own, which was new at the time and is the reason the campaign is remembered as an internet event rather than a television buy.
Five weeks later Cadbury did something much less romantic. On 7 October 2007 it brought back Wispa, 23 million bars at 42 pence, after a campaign by its own customers on social networks to resurrect a bar it had discontinued.
When Cadbury Schweppes reported its 2007 results the following February, it said the British business had grown 5%, and credited the Wispa relaunch and the gorilla advertising campaign. Both of them, in one sentence, from the company itself.
The market was also moving underneath all of it. Chocolate was recovering against a 2006 base that had been pushed down twice, once by the recall and once by a hot summer, which is the weather that most reliably stops people buying chocolate. The gum market grew 16% that year on the back of a separate launch.
So the year everybody attributes to a gorilla contained a bar relaunch with its own fan campaign, a category recovering off a doubly depressed base, and a company that named two causes when it reported the result.
The scoreboard
| Metric | Figure |
|---|---|
| The brand before | Dairy Milk was entering its third year of decline at the start of 2007, so the fall had begun before the contamination did |
| The failure | June 2006: a leaking waste-water pipe contaminated the chocolate crumb at a British plant; more than 1 million bars recalled across the UK and Ireland at a reported cost of about 30 million pounds, with an outbreak affecting around 37 people linked to it |
| The legal close | 16 July 2007: guilty pleas and fines totalling around 1 million pounds plus costs, six weeks before the ad ran |
| The ad | 31 August 2007: 90 seconds, a gorilla, Phil Collins, no chocolate until the last two seconds. Fallon London, directed by Juan Cabral, released under Cadbury's own Glass and a Half Full Productions |
| The result | Dairy Milk value sales up 5.0% year on year in the month the ad broke, against a 7% decline earlier that year. Figures of 7%, 9% and 10% circulate for the same campaign |
| The brand measure | 20% more people viewed Cadbury favourably after the campaign, per YouGov |
| The effectiveness claim | The IPA paper puts Gorilla's return at 4.88 pounds of incremental revenue per pound spent, and the platform's master-brand payback at 171% above previous campaigns |
| What else was running | Wispa returned on 7 October 2007 with 23 million bars at 42p. Cadbury's own 2007 results credited the Wispa relaunch and the gorilla campaign together for 5% growth in Britain, against a 2006 base depressed by both the recall and a hot summer |
| The awards record | Shared the 2008 Cannes Film Grand Prix with Halo 3; the IPA effectiveness paper won a Silver in 2010, not the Grand Prix. Cadbury's IPA Grand Prix came in 2022 for a different campaign |
The one figure with a stated window and a stated baseline is the 5.0%: Dairy Milk value sales, year on year, in the month Gorilla broke, against a 7% decline earlier in 2007. The 7%, 9% and 10% versions circulate without either, which is why they belong in the record and not in the verdict.
The autopsy: one result, four figures
"Cadbury made an ad with no chocolate in it and saved the brand." The ad is real, the recovery is real, and the sentence is still wrong in three places.
First, the disease came before the symptom. Dairy Milk was in its third year of decline when 2007 opened. The recall was an acute event on top of a brand that had already been losing its meaning for two years. Which means the ad was not treating food poisoning. It was treating a slow loss of relevance, and the recall is simply the moment that made the slow loss impossible to ignore.
Second, it did not run alone, and Cadbury never said it did. Wispa came back five weeks later with 23 million bars behind it. The comparison base was soft because 2006 had a recall and a heatwave in it. The company's own results statement credits Wispa and the gorilla together. The single-cause version of this story was written by the advertising industry, not by the client.
Third, and this is the one worth carrying into your own company: the lift has no agreed size. At least four numbers circulate for the same campaign. Dairy Milk value sales up 5.0% year on year in the month the ad broke, against a 7% decline earlier that year. A 9% rise over the twelve weeks of airtime. A 7% figure in the Cannes coverage. A 10% figure in later trade retellings. Only the first one comes with a stated window and a stated baseline. The other three are the same event measured by people who never agreed on the measure.
The award record drifted the same way. Gorilla shared the 2008 Cannes Film Grand Prix, because the jury gave two that year and the other went to Halo 3. The IPA effectiveness paper on the campaign, the one titled how a drumming gorilla beat a path back to profitable growth, won a Silver in 2010. Not a Grand Prix. Cadbury's IPA Effectiveness Grand Prix came in 2022, for a different campaign, under a different agency, fourteen years later. If you have been told Gorilla won the Grand Prix, you were told a number that grew in transit, which is exactly what happened to the sales figure.
What survives all of that is the part nobody quotes. YouGov had 20% more people viewing Cadbury favourably after the campaign. The IPA paper puts Gorilla's own return at 4.88 pounds of incremental revenue for every pound spent, and the wider platform's master-brand payback at 171% above what previous campaigns delivered. Those numbers were submitted to a process built to attack them. The round sales figures were not, and the round sales figures are the ones everybody repeats.
One more thing the legend hides. Mark Ritson's reading is that Gorilla had no legs, and he is right: the follow-ups could not extend it. Trucks and Eyebrows arrived and went nowhere, because a gorilla playing the drums is an execution and not an idea, and an execution has no second episode. The thing that eventually did have legs was the line underneath the production company's own name, a glass and a half, which is what won the Grand Prix in 2022.
What feeling can and cannot do
Two different failures get treated as one, and they need opposite answers.
A demand failure means people know you and are not choosing you. Argument works there. Proof, price, a better product, a clearer offer, a comparison. You are in a negotiation and you can win it with evidence.
A trust failure means people have decided something about you. Argument makes it worse, because every sentence you write about the incident is a sentence that puts the incident back in the room. This is why the apology post and the quality-assurance campaign so often make the week worse than silence would have.
Feeling works on the second one, and it works for a narrow reason. It does not change what people believe about the incident. It changes what comes to mind when they see your name, which is a different thing and a slower thing, and it is the only thing available once arguing is closed to you.
Three conditions have to hold before it works. The fault has to be closed and carry a date. Nothing in the campaign can refer to the incident, not even cleverly. And the product has to be back on shelf in a form somebody wants. Cadbury had all three on 31 August 2007. Run that same ninety seconds in July 2006 with the recall still live and it becomes the most cited disaster in this series instead of its most cited triumph. Same film, different date.
The right order: close it, baseline it, then change the subject
Close the fault and give it a date. Say what went wrong and what changed, once, in the least emotional language your company owns, and then stop referring to it. Cadbury fixed the plant, spent on prevention and took the fine in open court before it asked for any goodwill back. Anything you spend on feeling before that date is spent against you, because it reads as a company buying its way past a question it has not answered.
Write down the baseline before the campaign runs. One measure, one window, one owner. What number, over what period, compared to what. Cadbury's recovery exists in four versions because nobody fixed that in advance, and fifteen years of decks have been quoting whichever version suits the slide. This is the whole difference between a clarity metric and a vanity metric, and it costs one meeting to settle before the work starts and is unrecoverable afterwards.
Then change the meaning rather than the argument, and check that the idea can run a second time before you build a year on it. Ask what the next three executions are. If the honest answer is another gorilla, you have an execution and not a platform, and you will be back in this position in eighteen months. Gorilla could not be extended. A glass and a half could, and did, fourteen years later.
None of this is a British television budget. A restaurant here after a closure notice, a clinic after a complaint video goes around, a delivery company after a data leak. The default sequence in all three is the same: an apology post, then silence, then a discount. The apology repeats the accusation, the silence reads as guilt, and the discount tells people the only remaining reason to come back is money. The order that actually works is the one Cadbury used: fix it and date it, agree the one number you will judge the recovery on, then spend on being liked again for a reason that has nothing to do with what happened.
The takeaways
A trust failure and a demand failure need opposite answers. Argue with a demand failure. Never argue with a trust failure, because the argument puts the incident back in the room.
Feeling restores meaning. It does not fix the fault, refill the shelf, or repair a comparison base. Cadbury closed the plant issue and paid the fine six weeks before the gorilla ran, and Wispa came back five weeks after it.
Check what was already falling. Dairy Milk was in its third year of decline before the recall, so the famous ad was treating a slow loss of relevance and not a food-safety incident.
If you do not fix the measure before the campaign, the market will pick one for you. One result, four circulating figures, and only one of them carries a window and a baseline.
Ask for the next three executions before you approve the first. Gorilla had no second episode. A glass and a half did, and that is the one that won the effectiveness Grand Prix in 2022.
Frequently asked questions
Did the gorilla ad really increase Cadbury's sales?
Yes, and the size is disputed. The figure with a stated window and baseline is Dairy Milk value sales up 5.0% year on year in the month the ad broke, against a 7% decline earlier in 2007. Versions putting the rise at 7%, 9% and 10% also circulate. The campaign did not run alone either: Wispa returned five weeks later with 23 million bars, and Cadbury's own 2007 results credited both. The most defensible numbers are the brand ones, including 20% more people viewing Cadbury favourably per YouGov.
Was the salmonella recall the reason Dairy Milk was declining?
No, it accelerated a decline that had already started. At the beginning of 2007 Dairy Milk was entering its third year of decline, which puts the start of the fall around 2005, before the contaminated crumb was discovered in June 2006. The recall made an existing loss of relevance acute, visible and legally expensive. That distinction matters, because a campaign built only to answer a recall would have addressed the smaller half of the problem.
Did Gorilla win the IPA Effectiveness Grand Prix?
No. The IPA paper on the campaign won a Silver at the 2010 IPA Effectiveness Awards. Cadbury's IPA Effectiveness Grand Prix came in 2022, for a different campaign, under a different agency. At Cannes in 2008 Gorilla shared the Film Grand Prix, because the jury awarded two that year and the other went to Halo 3. The awards record grew in retelling in the same way the sales figure did, which is a useful reminder to check a famous number at source before it goes in a deck.
How did an ad with no product in it work at all?
Because the problem was trust, not information. After a safety failure, every sentence a company writes about quality returns the audience to the incident. Cadbury stopped arguing and rebuilt the association instead: the feeling the brand used to carry, with no reference to what had happened. That only works once the fault is genuinely closed. The plant was fixed and the fine was paid in court six weeks before the ad aired.
Should a mid-market company run an emotional campaign after a public failure?
Only after three things are true. The fault is fixed and carries a date you can point to, the campaign makes no reference to the incident at all, and the product is back in a form people actually want. If any of the three is missing, an emotional campaign reads as a company buying its way past an open question, and silence would have cost less. The order is not stylistic. It is the whole difference between the Cadbury outcome and a second news cycle.
What should I measure during a recovery campaign?
Fix one measure, one window and one owner before the work starts. Pick the sales line you will judge it on, the period you will compare, and the baseline you are comparing against, and write all three down where the rest of the company can see them. Add a brand tracker question, because a trust recovery shows up in perception before it shows up in revenue. Cadbury's result exists in four versions today precisely because that was never settled in advance.
This article is part of BMD's marketing case-study series. Episode 9 is about what feeling can and cannot repair. Episode 10 is canned water with a heavy-metal name and a 1.4 billion dollar valuation. That is Liquid Death.
Sources and further reading
The IPA Effectiveness Awards case study on the campaign, Cadbury's Dairy Milk: how a drumming gorilla beat a path back to profitable growth, a Silver Award winner in 2010, for Dairy Milk entering its third year of decline at the start of 2007, value sales up 5.0% year on year in the month the campaign broke against a 7% decline earlier in the year, Gorilla's return of 4.88 pounds of incremental revenue per pound spent, and the Glass and a Half Full platform's master-brand payback of 171% above previous campaigns. YouGov brand tracking showing 20% more people viewing Cadbury favourably after the campaign. Contemporaneous reporting of the June 2006 contamination and recall of more than a million bars across the UK and Ireland at a reported cost of around 30 million pounds, the Health Protection Agency's link to an outbreak affecting around 37 people, and the guilty pleas and fines totalling around a million pounds plus costs on 16 July 2007. Trade coverage of the 31 August 2007 debut in the Big Brother final on Channel 4, the 90-second running time, Fallon London, the director Juan Cabral, and the Glass and a Half Full Productions banner. Coverage of the Wispa relaunch on 7 October 2007, 23 million bars at 42p, and of Cadbury Schweppes' 2007 results in February 2008 crediting both the Wispa relaunch and the gorilla campaign for 5% growth in Britain against a 2006 base depressed by the recall and a hot summer. Cannes Lions 2008, where the jury awarded two Film Grand Prix, to Gorilla and to Halo 3. The 2022 IPA Effectiveness Awards Grand Prix for Cadbury's There's a glass and a half in everyone, a separate campaign. Mark Ritson in Marketing Week on why Gorilla lacked legs and why the follow-up executions failed to extend it.
About BMD
Most companies don't have a marketing problem. They have a marketing department that was never built. BMD is a boutique consultancy that installs structured, measurable marketing departments inside mid-market companies across the GCC. We don't run your campaigns, and we don't hand you a strategy deck and leave. We build the operating system: the structure, the measurement, and the ownership that turn marketing into a function leadership can rely on. The method is the BUILD framework, published and practiced: a book, an online program, a community of Gulf founders and marketers applying it, and diagnostics that replace assumptions with measurement. Delivered in Arabic and English, founder-led.
Redha Alayesh
A marketer with a software engineer's discipline and a scientist's mindset. Across 40+ marketing departments in the GCC, he built the BUILD framework to solve the problem he kept finding: capable marketers trapped inside companies that never built them a department.