The document discusses the results of an experiment testing personalization in advertising. It tested personalizing ads for 2,000 people across 4 product categories and 3 levels of message intimacy. The findings show that personalization can give lower interest categories more charisma and make for a more confident and direct approach. However, personalization risks becoming creepy if the level of intimacy is too high or if boundaries are not set on what data is used. A human perspective is needed to ensure personalization remains useful and not intrusive.
18. % uplift vs. traditional advertising
G3. And which, if any, of the following descriptions would you say apply to this advertising for [BRAND]? Base: 47 for personalised ad (Behaviour) and 83 for traditional ad
19. F3c1. Now, thinking only about [BRAND], please indicate how much you personally agree or disagree with each of the statements below? Base: 108 for personalised ad (Behaviour) and 176 for traditional ad
% uplift vs. traditional advertising
A company that recognises customer loyalty
Good introductory offers
Provides simple, transparent and clear products
Is a brand or company I would recommend to others
Is an innovative brand
20.
21. G3. And which, if any, of the following descriptions would you say apply to this advertising for [BRAND]? Base: 41 for personalised ad (Behaviour) and 57 for traditional ad
% uplift vs. traditional advertising
It used to be that personalisation was about stitching a label with your name on it on your school jumper. But times have changed.
Starbucks’ attempts at stamping your name on your cup have been met with a lot of derision and laughter. This is from the ‘Starbucks Name Fail’ blog: the first cup belongs to a customer called Sarah, who at the time specified to the barista that her name was spelt with an ‘h’. The barista took the instructions verrry literally. The middle picture belongs to a guy called Edward. Edward went to the Starbucks in China, and the barista there, being unfamiliar with the name ‘Edward’, essentially wrote ‘foreigner’ in Chinese instead! The last one is really unfortunate, poor Stephen, who has been referred to as ‘semen’. No idea what the barista there was thinking!
Anyway, personalisation has been taken to various interesting conclusions by different brands – this is Coke’s ‘Shar a coke’ campaign that we’re probably all familiar with.
Even more familiar are the so-called personalised ads on sites like Facebook, which essentially base the ads off cookies and retargeting.
…and our colleagues at our sister agency Rocket did a more interesting version of this for a UKTV campaign promoting comedian Dave Gorman’s show ‘Modern Life is Goodish’ not too long ago.
They thought it would be compelling to take Dave’s view on personalisation through to the actual ads.
Here are two examples…as you can see it got very meta – and it worked. It got brilliant results…This is my favourite visual from the campaign actually. Imagine you’re sitting in the loo at a bar. And you see this.
And really got us thinking about how far you can push personalisation.
We already know personalisation is a powerful tool via our performance clients and the effects of optimisation
We don’t know how people really feel about it and if there are any casualties along the way. I spent a while looking for existing research on the subject, and there isn’t a lot.
So we thought we’d find out by creating an experiment. We surveyed 2000 people to find out their personal details – such as having kids to how quickly they paid off their credit card. Weeks later we the re-contacted the same people to fill out a website survey. But what we were really interest in was they reaction to 400 personalised ads peppered throughout the survey. Across 4 categories – travel, ents, retail & finance, with 3 different levels of intimacy
Low – household situation, interests, likes
Medium – lifestyle behaviours
High – addressable ads – name, sex, religion, politics
What we found can help brands connect better with their customers.
Of course, some brands don’t need any help connecting with customers. You never see Clooney with his nose in a self-help guide. If you have a brand aura as strong as Clooney, you can take the next 20 minutes to take a nap.
Back to our research: I have to confess here..that the experiment we ran resulted in the exact opposite of what we thought…we thought that personalisation would have the biggest effect on high interest categories.
But the exact opposite turned out to be true. Just like Clooney, these categories don’t need extra help to connect. As they are naturally more interesting anyway…
It was the lower interest categoriess that benefited the most from personalisation .We saw they biggest incremental effects for Mr Finance and Mrs Retail in terms of noticeability, impact & brand uplift
So to be crystal clear..
The good news for the rest (who aren’t Clooneys) is that personalisation helps brands to connect with people.
And punch above their weight…
Don't mince your words, get straight to the point. Get specific to get noticed. Make sure they know you are talking to them. And that they recognise their own personal traits in what you are saying. Avoid generalisations ‘you like beach holidays don’t you’. Yeah who doesn’t? By being direct and to the point, we saw a 40% increase in standout for personalised vs traditional ads, for these two categories.
Spell out exactly what you are bringing to the party. How you help make their lives a bit better/easier. In the finance category we saw a 122% increase for ‘helpful’ personalised messages vs straight ads. But if you can’t help, walk away, there's plenty more fish in the sea.
Here are finance examples of best performing messages that were both direct and useful…
This is what you can achieve if you follow rules one and two. Compared to traditional ads, personalised ads in the finance category were seen as more modern, informative, useful, eye catching – and far less likely to be perceived as ‘ordinary’
We also saw uplift in key brand metrics across the board compared to traditional ads
Here are retail examples of best performing messages that were both direct and useful…
Again similar results… their personalised ads were seen as more modern, useful and eye-catching compared to traditional ads, as well as clever and surprising.
It’s a core part of media planning that you advertise in places where your audience is likely to be. So if you want to advertise to ‘progressives’, namely upwardly mobile adults, the Guardian would be a good choice. Millennials, maybe Vice Media, and so on. That makes sense – but when it comes to personalisation, we found that context matters even more – in fact 3 times more. If the ads appeared on a website people already had high affinity towards, they were 3 times more likely to respond positively to personalisation.
Our experiment has a disproportionate effect on the under 35. They expect a certain degree of personalisation – and respond well if you get it right. But get this wrong at your peril – they will punish you!
Last year Facebook had to apologise after a seemingly innocuous feature it rolls out to its users shortly before Christmas every year. Called the Year in Review it automatically picked one particularly well-engaged photo to present to users, under the banner: “Here’s what your year looked like!” For many users, that will have been a happy memory, such as a graduation, wedding, or the birth of a child. But for some users, the algorithm forced painful memories back to the surface such as this lady’s ex-boyfriend’s which framed a picture of his house on fire. There were some much sadder stories as well – people being reminded of family members they’d lost, for example.
This demonstrates that not everything can be done algorithmically. Some things, it seems, need the human touch.
In reality, personalisation is carried out by machines…we all know this, it’s what programmatic advertising is about.
Computers essentially auto-generate millions of creative iterations, and if the feedback loop says yes, computer says yes.
So back to the experiment…we wanted to see what would happen if we acted like machines and took the message to the inevitable conclusion. We went into ‘no-mans-land’. We added name & religion…to the copy
The very concept of machines has changed through the years. During WWII, a machine like the Colossus, and of course the Bombe, which Alan Turing designed, is what helped us win the war by breaking German codes.
Today, we have machines like IBM’s Watson, that was able to win the quiz show Jeopardy a couple of years ago, and Deep Mind, which is being put to work by Google on ever more complicated artificial intelligence questions.
So going back to the advertising machines, so to speak…what happens when they are allowed to run on their own?
If you dial up personalisation without any regard for boundaries and let the computers do their thing without ANY interference, you become the proverbial creep. You *think* you are going all Clooney, but actually….you’re doing a Dapper
67% of people found extreme personalisation ads creepy and intrusive, this was across all categories. So what's happened? You’ve entered the ‘uncanny valley’. When machines try to act like humans in this way, becoming the stuff of nightmares.
Plain weird
We’re about to cross the line from auto-placement of ads by AI to auto-creation of ads by AI, spelling the end of A/B testing. But if you’re running thousands of auto-created ads at any one time……What you’ve got is a ticking time bomb in terms of victims of offensive messaging. So how do we prevent the inevitable? How do we stop the machine sense over-ruling common sense? Well the good news is that its in our hands. Here are three simple things we do at PHD.
Which brings us to rule number 5 of personalisation: A human must always be involved in the process, not just a computer. Remember who you’re advertising to, and for what. As an agency, we have a responsibility to our clients and to their customers that we take very seriously.
Right upfront, set clear parameters of what is and isn’t allowed. Some things are just a no go, such as personal religious beliefs. When we crossed that line in our experiments, it showed in the feedback.
Give yourself the opportunity to have a time out before the business of real time deployment begins in earnest
Especially at the early stages of a project that involves artificial intelligence. Include brand owners in this process.
This talk wouldn’t be complete without mentioning the future. Places like Microsoft Research and MIT are already looking at what the evolution of personalised advertising might be like. There isn’t a lot of research in this space, but what little there is, these are guys are exploring – and it’s mostly academics (PHD is an anomaly in this, in a positive sense). I think this is particularly interesting because of a trend that some advertisers think doesn’t matter but that I think is a pretty big thing: the increasing use of ad blockers by regular people. People block ads for various reasons but privacy is a big one. What researchers are calling client-side personalisation is one potential solution – namely, personalisation done by making use of data that sits on the user’s personal device at a local level and isn’t seen by the massive data servers being run by Amazon, Facebook and so on, other than an interaction to be able to serve the right ads. People tend to be more comfortable with personalisation if it is done in a way that they have control over.
CSP is in direct contradiction to the solutions provided by consulting companies like McKinsey, which break down a user into their various ‘parts’. This is McKinsey’s recent Digital Genome research, which neatly lays out the different touchpoints when you look at a shopper as a faceless, mysterious person – now these are useful points to consider but it crucially forgets one thing..
http://www.mckinsey.com/insights/marketing_sales/cracking_the_digital-shopper_genome
…that we’re talking about real people, like you and me.
So what have we learnt?
Personalisation elasticity is stretchier than you might think. This X is the current average for personalisation
You can go further than you think…
But if you take things a bit TOO far…this is a recent ad from Peugeot, tell me what you think