The Three Blind Men and the Elephant


Lou Jordano, Chief Marketing Officer, Crimson Hexagon, in this article addresses the pitfalls brands experience by segmenting consumer research – and outlining how to leverage AI and ML technology to achieve a full picture of customers’ opinions and preferences

There’s an old parable, originally from India, about three blind men encountering an elephant for the first time. The story has many variations, but usually goes something like this:

Three blind men heard that a strange new animal, called an elephant, had arrived in their town. None of the men knew anything about the shape, size or appearance of an elephant so they decided to approach it and use their hands to better picture exactly what an elephant is like.

The first man felt the trunk and declared that an elephant was round and long like a snake.

The second man touched the elephant’s tusk and announced that the strange animal was smooth and sharp like a spear.

The third man felt the elephant’s leg and determined that elephants were thick and curved like a tree trunk.

None of these men were right, of course, but they weren’t strictly wrong either. From their experiences and research, an elephant was, in fact, like a snake, spear and tree trunk. The problem is that they were missing the holistic experience of the elephant. They were limited to only a partial understanding, each focusing on a single (accurate) facet of the elephant without understanding how those aspects exist together.

Blind research

What does this centuries-old parable have to do with modern consumer research? It provides valuable insight into how most companies approach customer research. It’s incredibly difficult for them to wrap their arms around and visualize their entire audience, so they resort to fractional understanding of their target audience — a few thousand survey responses here, a couple of focus groups there. Maybe an analysis of product reviews and support tickets if they’re especially data-focused.

But each of these offers only a partial understanding of customers’ opinions and preferences. Making strategic decisions based on survey results is akin to conceptualizing an elephant based only on its tail.

Combining data sources

Ok, so if you’re a company that wants to better understand the characteristics and preferences of its audience, what do you do? If surveys and focus groups aren’t enough, what is?

The short answer is the internet. Each day, billions of consumers post on social media platforms and other websites with their thoughts and opinions about every topic, from every possible angle. If you’re looking for the full picture of the elephant, the internet is where you should go.

The slightly longer answer is all types of public consumer conversations. It’s not just social media, it’s not just reviewing sites and forums. The internet is full of consumer conversations of all types, but each of these taken in isolation is simply a snapshot of one aspect of your audience.

Maybe this is hard to conceptualize, so let’s make it a bit more concrete. What exactly is a brand missing out on if they focus their customer and market research to a single or small collection of data sources? Let’s take a look.

The Big Picture

Here are three examples of how a limited consumer dataset could be leading your brand to the wrong decisions.

1. Will the real Grey Goose audience please stand up?

Everyone loves vodka, right? But who is the spirit’s main audience? Men or women? Young adults or older consumers? If you wanted to find out, one great place to start is the online conversation. If you could understand who is doing the talking about vodka in forums, on blogs, and on social media, you can get a pretty good picture of the vodka audience, right? Right?

Not so fast. Let’s use the vodka brand Grey Goose as an example. If we analyze the Twitter conversation about Grey Goose, for example, we find that the audience is significantly male (73%) and mostly over the age of 35. Using this information, Grey Goose may decide to target their ads at this demographic or even change their messaging to better appeal to what the brand now considers its main audience.

But consumers don’t always talk about the brands they love. Increasingly, they show their love by posting pictures of the brands and products they favor, often including no text reference at all.

Therefore, simply looking at the text-based conversations about Grey Goose may be unintentionally marginalizing an entire audience segment. If we include images of the Grey Goose product line or logos in our analysis, the audience shifts dramatically. Consumers including the Grey Goose logo in their posts are more likely to be women (+22%) and younger than what a text-only analysis would suggest.

2. Credit where credit is due

If you’re a credit card company trying to better understand consumer opinions about personal finances or credit card choices, your mind might immediately jump to social media. With billions of consumer posts every day, sites like Twitter and Instagram would seem the perfect venue for a huge flood of consumer insights about credit cards.

Again, you’d be right and wrong. Social media platforms certainly contain a massive trove of consumer conversations about credit cards. But it turns out that consumer conversations on social only comprise a small fraction of the total online conversation about credit cards. In fact, forums and Reddit account for nearly two-thirds of the overall online conversation about credit cards.

If you were a brand in the financial services industry making strategic business decisions based on just the social conversation about credit cards, you’d be missing a huge part of the conversation — and potentially coming to the wrong conclusions based on an incomplete dataset.

3. Baby and the bathwater

So far, we’ve given examples about how brands might be misunderstanding their audience by excluding images, forum data and social media conversations from their analysis. But is it possible for them to rely too much on social media and other online consumer data?

In certain situations, yes: there is a risk of basing business decisions solely on social media conversations. Because social media has such a broad user base, and because consumers use it to discuss brands they may not support and products they may not use, there is a risk of conflating the aggregate social media opinion with the opinions of your actual consumer base.

A recent scenario brings this risk into relief with the question, “Are some brands wrongly viewing social media as a sufficient guide for tracking consumer sentiment?” There are situations in which the overall sentiment of a conversation on social media diverges from (or even contradicts) the sentiment among active customers.

As an example, let’s look at to Nordstrom. Last year, the retail giant found itself in the center of a social media maelstrom as President Trump blasted the brand over TwitterOpens a new window for dropping his daughter Ivanka’s clothing line.

Within hours, Twitter users piled on the brand, calling for a boycott and demanding immediate action from Nordstrom. It looked like a bad time for the retailer, with consumer sentiment threatening to plunge overnight.

The only problem? Nordstrom didn’t see any of the negative effects of this social storm in the real world. In fact, Nordstrom’s shares rose – and its business outperformed many of its rivals in the following months.

The blind brand managers groping for audience insights would have seen the social conversations about the brand and assumed the entire fate of the company was in jeopardy. In fact, they were only looking at one small, potentially misleading facet of the brand’s reputation.

The elephant in the room

So far, so good, right? It’s easy to understand why a bigger data library leads to a more accurate and holistic understandingOpens a new window of whatever subject you’re analyzing. Understanding this isn’t the hard part. The hard part is finding a way to access all of these relevant datasets and analyze their (massive amounts of) data in an efficient and scalable way.

Luckily, rapid advancements to artificial intelligence and machine learning have given brands the technology they need to sort through the massive datasets that matter — social media, blogs, forums, reviews, internal call records and chat transcripts — to get a full picture of their customer, not just a fragment.