A couple of weeks ago, you saw how customer engagement is evolving, from carrots and sticks to sticky carrots. We followed up with a look at how different media affects the consumer in different ways—in The Winter of our (Dis)content—and continued with some metrics for Millennials, riffing on the Ultimate Question from the famous Net Promoter Score. This blog completes the set—with some ideas on how we measure ad effectiveness in the age of the digital consumer . . . and, perhaps, some insights into seeing the signal in the noise.
Let’s get away from advertising for a moment
With political stuff like the Snooper’s Charter—sorry, the Investigatory Powers Act—making the news, an associate of ours at Scorch has a radical idea for policing. And a bit of back-of-envelope research shows it could work equally well for locking up consumer attention.
Here’s the hypothesis. Democracies make great play of treating everyone equally. And that’s a good thing. But policing everyone equally takes a lot of money. And most of it is wasted, because most people don’t break the law.
(Do you know that a hard core of just 5,000 habitual criminals are responsible for half of all reported crimes? That’s just one in ten thousand people. And if you raise the sample size to 100,000, over nineteen out of twenty crimes are accounted for.)
Now, there are over 100,000 police officers in England and Wales. So what if every police officer’s first day on the job entailed being given one person’s name—one individual with anti-social tendencies—with instructions to keep him closer than their own shadow?
Boom! Instant 95% drop in crime. It’s hard to get away with jimmying doors or twocking cars when PC Plod has one hand on your shoulder.
In this model, the police aren’t counting the criminals they arrest. They’re arresting the criminals who count.
That’s our associate’s idea. To cut crime, you don’t need to put a population of 65m under surveillance. You just need to keep an eye on the ones proven to cause trouble. That’s less than one percent. (Actually a lot less.)
Could we look at advertising effectiveness the same way?
That’s our cue to get back to advertising.
To measure your campaign, don’t measure your campaign
The equivalent for consumer audiences, rather than cowed populations, is your “superfans”: the consumers who talk about you a lot and spread your messaging further and wider than any others. We all know an Apple acolyte. Or an Arsenal fanatic. But these are outliers. With most brands, most consumers aren’t superfans. (In the same way as most citizens aren’t habitual lawbreakers.)
Let’s look at the status of superfans for two premium consumer brands: ecstasy inducer—sorry, chocolatier—Green & Black’s, and cereal killer Jordan’s. First, though, a bit of statistics.
The Gaussians are coming
Whether you call it a Bell Curve, Gaussian, or Normal Distribution, you’re probably talking about the same thing. It’s a way of organising a set of data whose values cover a broad range, by arranging them in terms of how far they are from the average (mean) value.
Statistics nerds divide the distance from the mean (the midpoint of the x-axis) into chunks called Standard Deviations. Whatever the spread of your data, “normalising” its distribution like this will always mean 68.3% of your values fall within 1SD, 95.5% within 2, and 99.7% within 3 SD’s.
Why these numbers? The simple answer is that they cover almost every value. Just 0.3% of all values will fall outside 3 SD’s: a value 4SD from the mean is one of very few outliers. Almost everything you need to know about a dataset will fall within 3 Standard Devs on its Normal Distribution.
This means it’s a great way to shake out consumers who are a bit . . . different. Remember the people who scored you 9’s and 10’s on the Net Promoter Score in our last blog? They roughly coincide with the second Standard Deviation. But the Third SD and beyond—the hardcore superfans—get all Spinal Tap on us. They’re the ones who take it up to 11.
Stirring up our brands
There’s a reason for our delicious chocolatey and crunchy brand picks. Both brands are consumed by connoisseurs, or at least people whose tastes are a bit above the average. And both have a lot of cheaper, lower quality competitors in the same marketspace. Which gives the data a bit of context.
Why do we need context? Because of something we’ve riffed on in previous blogs: consumers are complete people, not demographics or data points. To understand them, you’ve got to look at how they interact not just with your brand, but with brands in general.
To explore this universe (and cut it down to size!) we looked at a sample of social media traffic in the last year, and applied four metrics. First, the overall volume of messages (posts, comments, tweets) mentioning the category (e.g. not just Jordan’s, but Kellogg’s; not just Green and Black’s, but Lindt) in a short period. Second, we shook out the number of people talking. Third, we counted the brand addicts, represented by the number of interactions each of their posts received—i.e. whether they were Liked, Shared, or retweeted. Fourth, the presence of a single word in these retweets: delicious. Let’s see the results . . . and see if we can draw any conclusions.
Green & Black’s Chocolate
Green & Black’s has an excellent profile on social media. Eyeballing the data, it punches above its weight against heavyweights like Cadbury and Mars. Principally because its biggest fans are voluble—exciting an average of twelve people to take action.
When normally distributed, there’s a twist. These superfans with twelve or more reshares are all in the top 5% of the graph, in the third SD. Rare, but still within the mainstream of the audience.
For all non-Green & Black’s posts, however—those about other chocolate—such superfans are in the far 0.3%. i.e. they are much rarer. All chocolate has fans, but Green & Black’s has more superfans. (As we could have told you from a look around the Scorch office.)
Is there an opportunity for G&B to take on mighty Cadbury’s and Hershey’s, with a programme of leveraging its superfans to take more actions? This small dataset suggests that if they did, it’d be very successful.
Jordan’s sell crunchy, fruity breakfast cereal. A smaller brand on social media, its Followers aren’t loyal to Jordan’s. They’re also fans of granola in general; of fruit for breakfast; of healthy eating. Some of them run cereal-appreciation websites. (Yes, there are such things.)
From our data it apppears that in the cereal sector, there’s a bit less competition to get a superfan. It takes just four reshares to catapult yourself into superfan status in the third SD from the mean. But unlike Green & Black’s, Jordan’s doesn’t have a clear advantage. While it’s not competing against cornflakes, it does share mindspace with lower-quality cereals. Our feeling is that Jordan’s could try harder.
For Jordan’s, a winning strategy would be to convert more cereal lovers to the delights of Jordan’s. There are a lot of consumers out there that with just a nudge—a personal approach? A sampling?—might start singing the praises of their granola. An opportunity so far unexplored.
Green & Black’s and Jordan’s superfans were all in the third Standard Deviation, the top 4.5% of the tweeting population. And this is what’s interesting about the Normal Distribution. Those percentages of 68.3%, 95.5%, and 99.7% aren’t random: they’re common in nature. Many datasets in the natural world also follow a Normal curve.
It’s intriguing that fandom on social media – the least natural dataset imaginable – follows the Normal pattern so closely.
So there we have it: to measure the effectiveness of your advertising, focus on what the superfans are doing. Win the superfans, and the rest will line up behind them.