Melmed is a true New Yorker - so much so that when we sat down for this overview after the first day of last week's IMC, he ordered a Manhattan. That's what makes his insight into how our local industry ticks all the more impressive.
But that's not all - he also has deep expertise on quantitative and strategic analysis of digital marketing, so much so that he's served as a consultant for media litigation as well as an in-house expert for global corporates along the likes of Droga5, MediaMind and NBC. He also holds an MBA and a Masters in Quantitative Methods from Columbia University in addition to a Bachelor in Economics from Oberlin College. Luckily, Melmed has the type of smarts that translate ideas across well.
So well, in fact, that I simply had to share some of his thinking with you. Below, Melmed elaborates on points raised during his keynote presentation, namely that big data is only human as it's fallible and needs humans to drive it, as well as from his workshop session about just how to interpret that data to glean those useful business insights...
1. Let's kick off by setting the context. Tell us more about the attention-deficit society and the impact of ever-shortening attention spans on advertising.
Melmed: When we look back on this era of passive consumers, it will prove to have been aberration in the history of media. Something like this would never have happened before TV! People were selective in the content they received. Advertising was simply about new products and how to integrate them. Then, with the explosion of affluence and TV, the dynamic changed. From being so tightly controlled, you could now get away with almost anything. As a result of that, what advertisers could get away with became increasingly stunning. They actually assume people would sit through their messages and be engaged throughout, it's a strange blip in media culture. Engrossing, compelling experience was firmly in the advertisers' hands. Now, we're back to reality. Both advertisers and brands need to understand how they are relevant and truly add value in order to stand out and grab the consumer attention they need.
2. Delve deeper into the specifics of how South Africa uses big data...
Melmed: Not surprisingly, I've been asked this a few times. Simply put, South Africa does not have the scale of other countries, nor the homogeneity that exists elsewhere, especially for a country of its size. Lack of data volume is another barrier. That said, the advantage, benefit and value of big data in South Africa far exceeds that of most countries, largely because the audience is so diverse, which means you think of things in many different ways yet at heart we all want the same things. We're now using data to bridge cultural gaps, to bridge differences of people speaking different languages. As a data researcher, you have less to work with, but the analysis and implications are more powerful here - luckily the data catches up eventually.
3. Well then, if data is fallible, as was the topic of your keynote presentation, how can we trust it?
Melmed: The only answer to this day is that you have to look for transparency in terms of how you're collecting data, storing it and matching it to people you're interested in. That's the only way a buyer can truly vet their data. It's almost impossible for anyone who is not in the industry to actually audit what you're getting. For many, that the temptation is great to simply dilute what you have in order to get more reach or to approximate similar interests in order to meet client needs. It's actually the digital equivalent of a used car lot - if you're not a mechanic, you need to bring one along to get the job done properly.
4. You also mentioned that people tend to lie or exaggerate in surveys, especially about their age and earnings. What's the behavioural reasoning for that?
Melmed: What makes people want to appear better than they are? It's simply a powerful subconscious desire, to align with the expectations of those around us. There's an element of wanting to appear more interesting or powerful, it's a survival instinct we all have. Like it or not we feel our fortunes are tied into those of others around us. When you consider that, what's the incentive to actually tell the truth, especially to yourself?
5. True. That links to your point that with big data, context is everything. Especially with your example in the keynote of married men visiting dating websites around a Valentine's Day for inspiration and advice rather than to cheat - it's fascinating, tell us more...
Melmed: Context is something that only a human being can bring to the situation. In the case of the dating website it was poor at best - if we recognise that they weren't just looking at profiles we'd know they're not looking to cheat. There's no doubt we need to understand the underlying motivation or core benefit driving the behaviour as we only see how it's exhibited.
One example of this is to look at high-end headphones. There are widely different reasons these are bought - some pay for the image, some for the actual sound quality. If you were trying to diversify product line SKUs, it would be easy to discern the different buyer types. Factor in decision-making factors, like bringing prominence to price and build quality as being more important to fashion buyers. By getting a complete picture of the users you'd find that those who participate in yoga and have strong fitness interests tend to also have high spend on fashion, an interest in their outer appearance and holistic health. On the other hand, some just want to look great while doing it. So there really is a huge difference between finding a man shopping for lingerie on Valentine's Day and doing so in July, and usually a human has to step in and say 'this is why'.
Think of a strong cellular network. Look at who is buying the product. There's a difference between those who need the 'best of everything', even if it won't actually make a difference to their life at the end of the day, compared to a family based in a rural area that actually genuinely need the strong signal. A robot can identify the cohorts, but you would need to emphasize different things. So if you were going for the family-oriented 'must have it all' purchaser, you could target those furnishing new rooms or buying pregnancy things. It's about using data in a clever way and also taking a collaborative approach to find who your product resonates with. You can differentiate at the product level.
6. Interesting. Let's look at gender split in advertising as its women's month, you spoke of the olden days' advertising message that cigarettes were promoted as 'torches of freedom' to get more females purchasing the product, does that extrapolate to 2015? How can we flip that perception of women from weak to strong, as successfully done in the Cannes Lion-winning Always #LikeAGirl campaign?
Melmed: Gender in general in marketing is lazy now. Before, we didn't have the granular data that's at our fingertips today and we couldn't discern one woman from another. Before men were into sport and women were all about their image. Now we see what people are doing online and we realise that the stereotype is inflated. Men don't spend that long on sport at all. These descriptors are simply the lowest common denominator in the worst possible way, and advertisers and marketers alike relied on this information to target the smallest possible slice of the population. One way to turn this around? Think of the Boss Babe campaign in the US, where she's all about focusing on her career now in her twenties and fitting in dating later. It's bizarre that marketing is so far ahead on other sticky issues, like gay and race, but seemingly still stuck on gender. Ironically, that's the worst data we have.
7. Lastly, you mentioned the power of the emotion of advertising in your workshop session. Let's end our chat by understanding the importance of emotion.
Melmed: I found that the entire IMC Conference emphasised it, which was surprising. There was a movement almost a decade ago, calling for clever ways of using emotion to create brand connection along the lines of 'no more babies (and puppies), please'. These are very clichèd ways to elicit emotion and influence buying behaviour among consumers. I'm hoping the level of emotional discourse in this industry moves beyond this to something truly clever soon.
For more on Exponential and Melmed, click through to their press office. You can read Melmed's latest thought leadership there, on why data needs a human touch and how the truth hurts when you ask the wrong questions.
