The manager's ultimate objective is to gather psychographic insights or "why" certain content exchanges are correlated with corresponding consumer behaviour and to translate these into their long-term implications for brand management.

A recent PricewaterhouseCoopers (PwC) survey concluded that a majority of US companies are having trouble measuring their ROI (return on investment) on social media; add measuring consumer sentiments on-line, and the difficulty is compounded.

How did this come about and why are such attempts at this form of measurement problematic? Here's what we see: emotions are at the foundation of consumer decision-making and an in-depth understanding is seen as the pathway to consumer loyalty. In the past, brand managers often relied on sample surveys to elicit the needed data but now more are mining for insights from the content and context of social media exchanges, as the former take longer to administer, running the risk of being outdated when completed and more expensive as well.

The shift in methods is based on the immediacy, transparency and the emotion-laden context of social media. The manager's ultimate objective is to gather psychographic insights or "why" certain content exchanges are correlated with corresponding consumer behaviour and to translate these into their long-term implications for brand management. In the words of a marketing manager: "Data-driven insights are just as important for the "why" of a brand as they are for the "what" of marketing."

Rightly, they have sought to drill down into the IoT (Internet of Things) and its devices and internet platforms and extract deeper meaning than is often captured and conveyed by sample surveys or customer relationship tools... they wish to mine "feelings". This quest does not appear, on its surface, to be totally quixotic, but if we dive deeper into its concepts and methodologies, we often discover that the quest is challenged by several key business and research concerns, among them are:

1. Can human emotions such as those driven by passions be captured and measured?

2. What are the pitfalls of transforming qualitative remarks into quantitative data?

3. Do concepts such as "brand popularity" stand the test of "KPI Time"?

4. Does a concept such as "brand temperature" add value to a manager's toolbox?

5. How does the Internet of Things impact the gathering of such data?

So, let's get started.

?     Measuring human emotions such as love or hate, desire or despair, are part of the holy grail that social media managers are searching for. This has become all the more urgent by the ascendency of the brand as the primary differentiator and driver of businesses operating in highly competitive markets.       

Let's take love. Research has shown that brands are often literally consumer love objects-the deeper the love the greater the loyalty. But if love is driven by such intangibles such as "chemistry" and "fulfilment," how can we drill down deep enough in terms of measuring these concepts? And if we can get a read, think about the challenge in confirming the intensity and depth so we can measure love's staying power and health and therefore track when the "lovers" and the love object are in a meaningful "compatible" relationship. This requires deep dives into the psychodynamics of each social media respondent and given that the social media exchanges occur at warp speed and the actual "lovers" are usually aggregated into normal curves and averages, the waters become even muddier and getting to the bottom of things more obscured.

Accessing the sub-conscious becomes a Herculean task. But without this the love / loyalty axis gets called into question and any long-term relationship (AKA loyalty) becomes a fling and with it a failure to manage the brand and the budget and to measure the ROI.  

?     A recent new algorithm seeks to track consumer's perception of brands on Twitter, providing real-time insights into consumer preferences by tracking not what they post but rather which brands they follow. This is a departure from the usual approach which is to do textual analysis of what consumers are saying and catalogue the content. In this fresh approach, the algorithm steps in and identifies correlations between complementary brand sectors (say, automotive and environment) and can determine repeated "following" patterns say between brands such as Prius and Sierra Club. So far, so good.

But then the detour: the model assigns a quantitative value to the number of occurrences; the higher the greater value and the lower or less likely to be associated in the consumers mind (say, discount stores and luxury) the lower the score. Here is a classic example of where a qualitative decision (which brands to follow) is elevated to a quantitative value but for which there is no emotional evidence save repeated patterns. We don't know the "why" and without this the data does not have the necessary validity to be useful for serving a specific business objective. Seems like we have a vanity metric. Here's the test: now that we have the information, what can we do with it? 

?     The concept of brand popularity has also entered the mainstream of social media metrics. In and of itself, it's an interesting metric; but as to its usefulness as a business development tool, less so. The concepts used to capture the data include such soft variables as "share of voice," "social conversations," and in one such ranking system, "share of search interest" and "web traffic"- all of which do point to the popularity of a brand, i.e, how much the "populace" finds the brand of interest, a seeming measure of the depth of feeling. However, if the object of a business is identify, find and nurture a customer, the popularity of a brand is only the beginning of assessing if a strategy is on the right path.

The more compelling metrics must then be the KPI goals of a company and whether the popularity metric correlates with sales, profits, market share and loyalty and the latter's subset, customer acquisition and retention costs. In the instances where popularity rankings are posted, this connection tends not to be made. This makes popularity assessments vanity metrics.

?     Brand temperature is a more recent metric to enter the social media space. We all are aware that social media conversations can heat up and with influencers, postings, tweeters and re-re-tweet, etc, things can get pretty hot. Various indexes such as the Lyst Index have emerged which again as with popularity ranking, attempts to measure the "temperature" of a brand driven by how consumers online feel about them and communicate accordingly, resulting in a ranking.

Using counts such as Instagram buzz and brand engagement, the method seeks to chart the temperature (up or down) and correlate this with the topselling items of the hottest online brands. This is definitely going in the right direction, but the relationship between the heat online and the sales of particular items requires a more rigorous statistical connection. Correlations, unless they are confirmed by statistical validity and reliability standards, can be specious (the fact that an occurrence in one place seems to happen at the same time as another occurrence) and correlations are not necessary causalities. Finally, even if we are able to establish the right connections, the feelings or the "why", the most important variable for connecting with consumers, continues to be missing.

?     The Internet of Things (IoT), the myriad of devises that will connect us in a lifestyle changing eco-system, will play a profound role in determining if human sentiments can be measured in a way that serves both the consumer and the corporation.

Issues of privacy aside in this piece, IoT will enable the assessment of how we feel about an ad or a product or service as well as our propensity to purchase to be instantly sought and obtained and the natural additions of AI and machine learning will become more sophisticated so that facial expressions, body language, tone of voice and virtual reality will be available on all of our devices and translate these into "whys". Geo-spatial access will mean surveys and opinions can be obtained in real time eliminating the tendency for sample survey results to become outdated especially given the warp speed with which decision-making and data gathering occurs in social media. Nonetheless the human equation will still be a challenge to the objectivity index of opinions and sentiments and marketing validity. Measuring sarcasm, intensities, gaming the system, opinion vs. behaviour and measuring the unspoken, will remain.