What AI Won’t Tell You: 3 Rules to Turn Data into Compelling Insights

Don't forget to share this post!

blog-post-example-media

Today, artificial intelligence plays an expanding role in social listening and marketing. To better understand customer attitudes and behaviors, companies around the world are turning to AI to help break down hundreds of millions of social media exchanges every day. 

But as helpful as these AI tools are, they’re not perfect. You still need human expertise. 

Recently, Linkfluence’s Chief Evangelist Benjamin Duvall gave a series of presentations at Social Media Week in New York and Los Angeles on how to strike a balance between AI and human insight in analyzing social data. You can check out the full presentation here.

Linkfluence Chief Evangelist Benjamin Duvall at Social Media Week

Linkfluence Chief Evangelist Benjamin Duvall at Social Media Week

In this post, we’ll take a look at some of the uses and limitations of AI in analyzing social data, and will outline our three rules for how brands can stay ahead of the pack by turning detailed data into compelling market insights.

Let’s start with an example of the value of human insight over AI.

What color is this?

Take a look at this photograph. What color is the sky?
Beijing ‘Olympic blue’ sky

Beijing ‘Olympic blue’ sky

Most people would say it’s blue, or those who are more observant, sky blue. A well-trained AI would probably come to the same conclusion. 

In fact, this particular shade of blue is known as ‘olympic blue’, a term coined by Chinese Netizens to draw attention to the fact that a blue sky over Beijing is only produced through state intervention to temporarily close factories and take cars off the roads ahead of major events. 

Before social media, criticizing the government about the environment was interpreted as a threat to social harmony, which often came with repercussions. But do to largely to activism on social media platform Sina Weibo, improving air pollution has become one of China’s main objectives.

But what AI won’t pick up on is the disruptive social movement this photograph represents, nor explain how it impacts culture or the wellness movement in China.

This is a great example of the limitations of AI. Ask a program to detect perfect photographs of a clear sky over Beijing, and no doubt it’ll include this one in the results. Ask a program to explain the deeper meaning and insights behind the photograph, and you’ll come up short.

And given the challenges of digital transformation, brands need all the insights they can get. 

The challenge of digital transformation

The rise of social media has transformed many aspects of society and culture, including the way brands interact with fans and customers. 

And this rise has sorted most businesses into two categories: winners and losers.

While some brands have lost hundreds of millions of dollars due to social media crises (for example, Fonterra, KFC, and Apple), others have taken social media with sophistication and deliberate planning (for example, Maybelline, SK-II, and Nike). 

In fact, the Harvard Business Review predicts digital transformation as one reason why the majority of today’s Fortune 500 companies will be replaced within the next ten years. 

Like every other aspect of marketing, understanding digital transformation begins with attitudes and behaviors. Companies need a way to make sense of hundreds of millions of daily social media posts, from millions of sources, in dozens of languages. AI is an indispensable tool here. 

AI Series - Measure your influencer impact with celebrity identification

Source: AI Series: Measure your influencer impact with celebrity identification, Linkfluence

For example, by using computer vision, social listening software can automatically process and categorize complex images, including identifying scenes, objects, and people. 

However, even the most sophisticated computer vision can’t explain the deeper context and nuance behind a simple image. 

To do that, we need humans to connect the dots.

Context is everything

Don’t get us wrong: AI is an incredibly sophisticated tool when it comes to gathering, sorting, and analyzing huge amounts of information. 

That’s why most brands are already using social listening software with AI capability to track social media performance and highlight emerging trends. 

But while social listening tools allow us to process and interpret massive amounts of data, it can only give us observations, not real insight. 

For example, social listening could tell you who in the world tends to discuss a particular brand of sneakers, as well as where they are, and some basic demographics. These are great observations to have on hand.

Observation and insight, spotting risks and opportunities

Observation and insight, spotting risks and opportunities

However, these observations won’t tell you any insights about why particular celebrity endorsements have generated such a sales bump in a particular country, or the nature of the risks to brand reputation arising from using non-sustainable materials.

Observations provide guidance for incremental improvements. Market insights, however, allow brands to spot the next big risk or opportunity. Spotting these risks and opportunities can make all the difference between doing what everyone else is doing, and leaving them far behind.

So, how can you find these insights?

Three rules to turn data into valuable insights

As we’ve seen, AI is an indispensable tool. However, it isn’t everything: you need a way to turn data into valuable insights. 

Fortunately, we’ve got three rules to help you out. 

Rule #1: Be a marketer first, and a geek second

In 2011, social listening was getting a lot of attention. In that year alone, one social marketing company was acquired for $340 million, and another for $689 million. Those are some big numbers.

At that time, the focus of social listening was on simply tracking brand performance, and not on strategic metrics like brand equity. 

To see what we mean, take a look at one of our old reports:
Diagram

This example focuses on simple metrics like post volume, share of voice, and net sentiment score. What seems obvious to us now is how detached these simple KPIs are from core marketing metrics like awareness, familiarity, affinity, and purchase intent. 

For an example of these more nuanced and useful marketing metrics in action, let’s take a look at energy drinks brand Red Bull. 

Red Bull extreme sports hot air balloon brand strategyRed Bull extreme sports hot air balloon brand strategy

Red Bull’s brand strategy is to associate their product with risk and extreme sports, and they need to shape attitudes and perspective to achieve this. 

For Red Bull, social listening is focused on measuring the outcome of this strategy, and the extent to which this brand personality is perceived by fans and consumers. This involves using human analysis to connect the dots between simple metrics and market insights.

Linkfluence Social Brand Equity ADPR framework

Linkfluence Social Brand Equity ADPR framework

To help connect these dots, we’ve developed our ADPR framework to measure social brand equity, focusing on brand awareness, desirability, proximity, and relevance. 

For example, to measure brand relevance, we can identify the terms and sentiments most commonly associated with a brand online, then use specialist researchers to determine the extent to which these are aligned with the desired brand equity

So, as helpful as the data is, sometimes it helps to focus on the marketing side of the equation.

Rule #2: Use a hybrid model to move from observation to insight

Today, there are very few AI applications that we trust to work without human expertise. For example, it will be some time before AI is able to predictably solve complex creative and cognitive problems, such as ethical or moral questions. 

However, by combining the power of AI with human expertise, we can already solve complicated questions. What’s more, this is only likely to increase over time.

A helpful way to think about the increasing application of AI is through the AI maturity curve:

Competitive advantage diagramCompetitive advantage diagram

This curve helps us to think of AI in terms of the output it provides: descriptive, predictive, and prescriptive. Traditional, or legacy, social listening, is stuck in the sensing (descriptive) stage, with some innovations into predictive analytics. 

For example, we can now use AI in social listening to group a Twitter account’s followers according to their collective interests:

AI in social listening

 AI in social listening

However, while this kind of analysis can be helpful for community management and consumer targeting, this kind of analysis is not as applicable for advanced consumer insights

For an advanced example, take a look at this analysis of ingredient trends and popularity: 

Analysis of ingredient trends and popularity

Analysis of ingredient trends and popularity

Here, AI is being used to identify a subtle growth in the popularity of flavors and ingredients. This is not simply the tracking of already observable trends but is potentially getting ahead of these trends by understanding and interpreting unmet consumer trends and needs. 

And to do this, you can’t rely solely on software - you need a hybrid model putting the human brain to work.

Rule #3: Audit your digital maturity, and plan for transformation

Even if your social listening tools meet the current needs of your social marketing teams, your business could be exposed in the long run if you aren’t able to adapt to rapidly changing consumer behaviors and the wider digital ecosystem. 

Every brand needs a way for their social listening solution to evolve over time, enabling your people to adapt social intelligence to their needs. 

The first step in doing this? Gauging your own digital maturity.

Digital maturity model

Social intelligence maturity model

This graphic is a condensed version of our social intelligence maturity model. 

At the earlier stages, we begin with typical applications like content marketing and community management. At the more advanced end of the spectrum, we have consumer knowledge and insights, and executive decision-making. 

Every business needs to know where it is on the maturity scale. That way, you can grow your abilities steadily over time, and avoid the false confidence that your company is more digitally savvy than it really is. 

AI can only do so much - use human experts to bridge the gap

As we’ve seen, social listening tools powered by AI have become an invaluable part of the modern marketing landscape. Using these tools, brands can access insights from millions of social media exchanges every day and can find better ways to connect with customers.

However, as powerful as AI can be, it isn’t everything. 

To get the most out of social listening tools, you need a real-life human in the driving seat, helping to connect the dots and provide the context necessary to truly understand the risks and opportunities. 

This is why our three rules are so important. When it comes to social media intelligence, you should:

  • Be a marketer first, and a geek second
  • Use a hybrid model to move from observation to insight
  • Audit your digital maturity and plan for transformation

We’re hosting a webinar with Forrester on Thursday 27 June to discuss how brands and businesses can combine AI’s powerful data processing abilities with human expertise to get actionable business insights. To join us, register here

We hope to see you there! 

Forrester Webinar CTA-1

Don't forget to share this post!