73% of Brits were targeted by scammers last year, losing millions. O2 identified an urgent need to protect customers, improve anti-fraud credentials and fight back.
INSIGHT
The best way to tackle a scammer is to find them and arrest them. But many operate from anonymous locations, using burner phones. They’re unreachable and untraceable–and we’re a British telco, not the police.
We needed to attack them in the next best way: wasting as much of their time as humanly possible by deploying someone–or something–with all the time in the world.
IDEA
Enter Daisy: a 78-year-old granny who loves to chat. She tells anecdotes about her grandchildren and late husband Norman, forgets important information like her bank details and continually tries to share her lemon meringue pie recipe
The twist? She’s not a real granny. She’s a sophisticated AI with all the time in the world to scam scammers. The innovative technology behind Daisy can move at pace and at scale, without the need for human intervention.
We’re playing scammers at their own game. Scammers are using AI technology which is getting smarter – many people have fallen victim to sophisticated voice doubling tactics. Daisy uses the scammers’ own weapon against them and fights back with a more sophisticated AI solution; an AI solution which doesn’t need to eat, sleep or rest.
WE HAD FIVE KEY OBJECTIVES
INTRODUCING DAISY THE VIGILANTE
While each tool required to bring Daisy to life is bespoke and impressively realistic, the more technology layers you add, the more time it takes for the AI to respond. Using unconscious bias associated with the elderly helps mask technical limitations–we are more forgiving if their speech is slightly slower or they lose their train of thought.
Plus, Daisy is an unlikely hero–one of the factors behind her internet popularity.
MAKING IT ENTERTAINING
As our objectives included mass reach to drive awareness and change perceptions of O2, we prioritised cut-through. We couldn’t just educate people on the dangers of scamming–we needed to right a wrong while making it entertaining.
MAKING DAISY HUMAN
Key to Daisy’s success would be her authenticity and her charm.
Daisy’s voice was developed to sound as grandmotherly as possible. We worked with a colleague’s real granny (anonymised to protect her identity) to give Daisy an authentic personality. Thanks to this human input, she was built to mirror the voice, behaviours and quirks of an elderly woman. All in real time, and in a way that is indistinguishable from a real person.
MAKING DAISY BULLETPROOF
Daisy is composed of different AI models (defining what she says, what she sounds like, how she speaks and what she looks like) which all work together to become our protagonist.
Her scambaiter persona was trained using real anonymised scammer data–she can respond to any scamming situation, whether being told her bank account is being hacked or that she owes the council money.
INFILTRATING THE SCAMMER NETWORK
Once we had designed Daisy, one of the biggest challenges we faced was making sure scammers would actually call her. We used a combination of ‘number seeding’ and ‘breadcrumbing’ to get her number onto the scammers’ databases. To date, she has received over 1000 scam calls.
LAUNCHING DAISY
A key part of Daisy’s success would be making her famous. We wanted to get her as much PR as possible and get her on as many breakfast TV shows as possible, using our unlikely hero to spread the message that anyone could be the victim of scams.
As it was a PR-led campaign, we also needed a human spokesperson. We recruited ex-Love Island star Amy Hart as an authentic, relatable face of the campaign (she previously lost £5000 to a sophisticated scam). Amy acts to humanise ‘Daisy’ and remind people that this type of fraud doesn’t just happen to our grandparents – it can happen to anyone.
We used the familiar “scambaiting” content structure to launch Daisy to the world. The humorous film shows Daisy wasting scammers’ time in different locations in her home. When asked if her profession is “bothering people” she calmly replies that she’s just “having a little chat”. In follow-up content, she asks a scammer whether they’re talking about “hackers” or “snackers”.
The campaign launched on O2 and Amy Hart’s channels, as well as being sold into British publications.
The campaign received unprecedented coverage.
1. DRIVE MASS REACH
Daisy received over 1800 pieces of coverage in 9 markets, with 1 billion+ earned impressions.
We achieved 17% unprompted recall (the highest of 2024).
She was covered by everyone from Jeremy Vine to Whoopi Goldberg, from The Hindustan Times to the New York Times. By famous YouTubers, AI magazines and technology specialists. And everywhere from San Antonio to South Korea.
This equates to a total Advertising Value Equivalent (AVE) of £36m, from just £20k paid media.
2. IMPROVE O2’S REPUTATION
Satisfaction with O2's efforts to tackle fraud has risen 10 pp from 24% to 34% since April. (This outpaces competitors: EE and Tesco Mobile each only saw 3% increases in the same period. There have been no other ‘anti-fraud’ campaigns during this time.)
3. DRIVE POSITIVE SENTIMENT
The campaign drove the most significant positive sentiment in 2024, with a +6.7% uplift in favourability among those who recalled it. (O2 would consider anything above 5% “strong”). Independent analysis suggests media coverage is 100% positive.
4. INCREASE AWARENESS OF HOW TO REPORT SCAM NUMBERS
Awareness of 7726 has risen significantly since April, from 18% to 26%. (There have been no other major campaigns during this period, from O2 or its competitors.)
We saw an 8% month-on-month uplift in reports to 7726 during the campaign.
5. WASTE SCAMMERS’ TIME
A combination of ‘number seeding’ and ‘breadcrumbing’ got Daisy’s number onto the scammers’ databases – to date, she has received over 1000 scam calls.
