When I last worked in the ad industry, George W. Bush was president, tweets were something you heard in the park and Yahoo was – actually, Yahoo was the same in 2007 as well.

I left the business and went off to work in new product development, occasionally sitting in the same meeting rooms as my client’s ad agencies, and generally not paying a lot of attention to their presentations. Sorry.

Last year I started to get interested in ad land again, as I read more and more about programmatic advertising. In the NPD game, we’d been using clever algorithms to predict consumer behaviour for years. Wow, I thought. Now you can crunch all that data in real time on the viewer of any ad, you could do some truly amazing things. Agencies must be all over it.

My own online experience told me that maybe agencies weren’t. The same ad for a Garmin watch had been following me for a week. The same ad, whether I was on Facebook where I chat with friends, on Wired where I read about technology, or on Twitter, where I scream into the abyss. All that data about who I was, what I was doing or what platform I was on: clearly, none of it was affecting the ad.

I started to talk to friends who worked in programmatic. Most ad agencies, it seemed, were not interested in the data. Instead, they were just tossing stills from print and TV campaigns over the fence and asking the media companies to make them work online.

Why, I asked. Dunno, said the programmatic dudes, we can tell them 5,000 things about a single page view, but they don’t use any of it.

Five thousand things! I doubt I know 5,000 things about myself. My 10 years geeking out on vast amounts of data has taught me that 5,000 results are as good as none. Two are too few. Four to six – those I can work with. I wondered how my old friends who’d stuck it out in ad planning were doing. I asked a few of them, from global ad agencies to a couple of hot shops, to send me a blank briefing form from their agencies.

What is the brand’s tone of voice? What is the single most compelling message we can tell them? What supports this? I had to rush up to a few people as if I were Dr Who. I grabbed them by the lapels. What year is it? I asked in rising panic. It felt like I was back in the early 90s, when there were five TV stations in the UK and a bunch of poster sites at roundabouts.

Back then, when you had two campaigns a year at News at Ten and some slots in the Sunday supps, it was completely fine to bang away at being the ‘ultimate driving machine’ in the same Teutonic tone. Nobody got too tired of it. But when I wake up to your brand on Facebook, and scroll past it on Twitter, and see it on Instagram… it just comes across as repetitive.

Yet there are brands that are thriving online, both big and small. They’ve realised that it’s more important to be multifaceted than monotonous, to be surprising rather than consistent. Take Taco Bell, a chain that’s gone from being dubbed ‘Taco Hell’ five years ago to Gen Z’s default hangout. Its social media takes a lot of credit for this: it’s beautiful on Instagram, bitchy on Twitter, inspiring on its website, and its $40,000 TacoBot has already taken $10m in orders on Slack.

Movember has become a global movement online, rising from two Aussies in a pub to a phenomenon that’s raised over $300m. Their messages cover prostate cancer, male suicide and facial hair grooming tips. Their Facebook posts are funny, tear-jerking, surreal, handy, thrilling and heartwarming. Clearly nobody ever showed them an ad agency brief.

Nike, Rude Health Cereals, Victoria’s Secret even Victoria Beckham all seem to have discovered a way to thrive online, one that’s a million miles away from the USP or ‘brutal simplicity of thought’. They’re all multifaceted personalities, and they adapt their tone of voice to their audiences’ moods. Data allows you to do that, if you use it intelligently.

Last year, I teamed up with Torie Chilcott, one of the gods of programmatic advertising, to systematise the process. We started crunching data on what people loved online – not advertising, just… everything, from kittens to TED talks. This led us to a startling conclusion. There are dozens of kinds of content that people love, but they have four broad types. (Remember how I said you can work with four to six data points? Here they come…)

Great online content is either funny, useful, beautiful or inspiring. Great online brands do all four of those things. Victoria’s Secret is hilarious on Instagram. Rude Health is angry and ranty on YouTube. They bring surprise instead of consistency, and match their tone to the platform, rather than expect their audience to change emotional gear.

Data can help you broaden a brand’s emotional appeal. Our data shows that young women in London tend to find grandiose things beautiful, laugh hardest at dark humour and value authoritative opinion. BMW drivers aren’t inspired by social good or anything heartwarming. (Don’t you love it when the facts confirm your prejudices?) Data can also tell you where they’ll be most receptive to each of those tones of voice. To become a multifaceted brand, you’ll need to start thinking in a new way, at that 90s ad brief really isn’t going to help you.

Here’s the questions we think your brief should be asking.

If you can’t answer them, you can’t expect to sync with your audience’s emotions online.

We need 4 executions, not one.

We need to find our brand’s funny, useful, beautiful and inspiring.

Funny

What makes our target audience laugh?

How does that humour connect to our brand personality?

Inspiring

What kind of voice does our audience stand up and follow?

What in our brand could create that kind of rallying cry?

Useful

Where do our target audience go to learn?

What makes them lean forward and how can we learn from that?

Beautiful

What aesthetics turn our audience on and off?

What would our brand look like in that style?

By Brian Millar

Brian Millar is co-founder of the Emotional Intelligence Agency, a communications planning company that maps consumers’ emotional lives online. Follow him on Twitter @arthurascii