Tagged: Theory

24/02/2013

Why Social Media needs qualitative research – not numbers

Social media (and the wider Internet for that matter) is full of numbers that appear to be explaining things… from 57% of users of network Y doing X to 46% of users of website B coming from C. The problem with all these numbers is that, although they seem to explain what is going on – they mean we forget to understand why it is going on. This is especially true in the case of social networks: many researchers (and marketing “gurus” at the same time) seem to treat social networks, and by that I mean anything from the classic social networks to forums, groups, everywhere where people interact online, just like any other mass media channel. Talking big numbers, rather than small stories.

The problem is, that, for the most part, these researchers forget that actually twitter users are not the same as people who watch ITV1, while facebook users are not the same as people who listen to Capital Radio. What many people forget is that social media networks are, for the most part, made up of circles of friends (or people sharing a common interest) who are using social media technology to do what they want to do. As Hart put it in 1992: The internet is a “growing community of communities”, or as more modern authors would call it “Tribes”. That makes social media very different from a traditional media outlet, even a traditional media channel for the most part. Understanding that “tribe members” are there with a purpose means that in many ways, social media is probably more and more like person to person conversation, or telephone conversations. Yes, social media can be measured more easily than say telephone conversations (Wouldn’t it be fun, if we could say “37% of telephone users mention brand X when speaking to their friends”) – but really, social media is and remains a big conversation between distinct networks of friends. When I talk with someone on twitter it is nothing like a TV advert, but it simply is a conversation that is public… that may be between me and all followers, or it may be between me and just one friend. And the ultimate dynamics that are governing the conversation are not numbers and probabilities, but quite unique relationship dynamics. Dynamics that may, span across different social networks: For example, I may be responding to something a friend wrote on LinkedIn by means of sending a @tweet. Or in fact, conversations may be strongly influenced by relationship dynamics which go beyond social media sites, when taking to someone I know in real life. Yet, all the quantitative tools and amazing numbers that we have won’t account for that. Sentiment analysis simply doesn’t account for irony or “in jokes”.

Of course, quantitative numbers aren’t useless… far from it. But when it comes to human interactions they explain some, but not all of the dynamics that are at work. And more often than not, what they do forget is that social media is actually not one big audience – but simply circles of audiences, tribes – or simply a few humans speaking to each other using this technology. And many conversations and relationships online are just as unique as those in real life. So we need to understand, more than simply “numerise”, what is actually going on, if we are trying to make sense of consumers, followers, friends, contacts and whomever else we talk with…

14/11/2012

Dogs and Adverts: Classical Conditioning

Have you ever asked yourself why adverts that are saying, seemingly, pretty little ….actually seem to work? Well, the theoretical underpinning of this phenomenon that has been observed in many different scenarios is actually an almost ancient theory (from the 1890s), linked to… training dogs – called classical conditioning.

Classical conditioning combines two potentially arbitrary and unrelated stimuli, in the classical example the ring of a bell and the arrival of food. What happens is that both stimuli get “mixed up” and inherently connected, so that after a while even if only one stimulus is presented, for example hearing a bell, the dog (or person) reacts in the same way as if food was arriving – although it isn’t. Pavlov showed this originally by measuring the saliva of dogs. Of course, the same “habituation” also works with other stimuli.

Advertising, especially advertising that says little, uses classical conditioning to transfer positive (or negative) emotions to a brand image. For example, by simply showing an attractive model (positive stimuli) and a brand (unrelated stimuli) the adverts hope that the positive emotions towards the model “transfer” to the brand.

An interesting example, attempting a negative emotional connection, is the current PETA campaign. It connects a negative stimulus (disgust at the untrimmed bikini line) with fur (quite unrelated really). The campaign obviously has some critics, but I’d argue that (disregarding the arguments made in this article, and yes, I agree with what is being said, and that it can be seen as offensive…. but), the advert has at least the potential to be associate negative emotions with a stimulus – because either way, if it creates disgust or offense, then that would hopefully transfer to “wearing fur”. Though this is what I’d argue thing are a little problematic with the ad: the disgust/offence may actually transfer to the brand (PETA) as there isn’t a clear fur stimuli in the advert… In other words, it may not be immediately obvious what the two stimuli are that are being connected, and thus, remembering that the connections are made pretty much subconscious, the negative stimuli may actually transfer to the most obvious other stimuli – which might be the brand (rather than the message).

In short, classical conditioning is an easy and powerful tool – but it needs to be used carefully: the connection between the two stimuli is made subconsciously, and therefore they need to be reasonably clear and not requiring much elaboration. Moreover, adverts relying on classical conditioning, need to be repeated many times for the conditioning to take effect – thus they are by far not a cheap option.

11/11/2012

Alcohol Regs Updated: or why current regulations are not working

The Portman Group, the self-regulation body of the alcohol industry in the UK, has updated it’s code of practice for alcohol marketing. The updated code contains a number of changes making the code more “comprehensive”. It also extends the code to cover social media where the ASA codes are not currently applicable.

While making the code more comprehensive is certainly something to be welcomed, the code (along with other codes) still suffers from a major problem: policy makers assume, that by regulating what advertising cans “say” and “show”, they can effectively regulate the potential harm done by “harmfull” products.

Unfortunately, the view of advertising is both outdated as well as wrong. Advertising does not (or at least no longer) simply work by providing simple product information. Most advertising does not try to do that. Simply put, what advertisements try to do is to change brand attitudes and develop emotions towards brands. The emotional bonds between the brand and the consumer then leads to purchase decisions – more so than simple product information which is unlikely to be recalled correctly at the point of purchase.

Ultimately, therefore, many adverts give little information – and rely much more on building a brand image and emotional connection. Even more so, in the case of social media. Take the example of Peroni – not only on Facebook, but also in their adverts. If you check out their Facebook page (here), you’ll see how little product information they actually give. In fact, much of what they are talking about has nothing whatsoever to do with the beer: It’s opera, it’s fashion, it’s what it means to be Italian… This is building a brand image in the mind of consumers of Italian sophistication. Thus, by creating this image, and the associated emotions, the marketers hope to bind the emotions to the product. Thus, when consumers see or use Peroni, they “feel” the Italian sophistication.

This means, focusing on regulating what can and can not be said (like, for example, saying you can say it has low alcohol, you can not say it makes you sexually attractive) makes actually little sense. Because most advertisers don’t actually “say” much at all about their product – they rely on implicit, brand creating claims (e.g. opera = sophistication). Thus the focus of current regulation has to change drastically if it is supposed to be successful. Because as long as it regulates what is explicitly said, especially product information, it won’t cover implicit, brand building advertising – which is ultimately more likely to create powerful purchase intentions.

6/07/2012

Don’t promise what you can’t keep: Expectation Theory

Did you happen to watch the opening “laser spectacular” of the Shard yesterday? Were you one of thousands of spectators who felt distinctly underwhelmed? Maybe you even posted on social networks how disappointing the event was? Well, whatever you did, you were probably not alone. But what does the backslash (and critical press) of the Shard to do with marketing? I’d say a lot, especially with marketers who did not pay attention to even the most prolific of marketing theories!

The image on the top right sums it up (taken from the TimeOut London facebook page): The top image was the image that PR and marketing professionals from the Shard distributed to the press. No wonder, many people expected something quite extraordinary – and extremely spectacular. The lower picture is how it looked in reality. Quite a let down. This raises the question: was the show THAT inadequate? The answer is: probably not, but it depends on what you expected – and it most certainly was not what the organisers promised. Zeithaml, Parasuraman & Berry (1985) discussed the problem in detail in their Model of Service Quality, particularly the Gaps Model in Service Quality.

Take a look at the pictorial summary of the model on the left. It provides an excellent way to explain what happened when the Shard laser show got on the way. On the top, we have Word of mouth (people talking about the expected spectacular show, journalists talking about it etc) and past experiences (similar events at Canary Wharf during New Year, maybe visits to Hong Kong, etc…). In other words, expectations were likely to be quite high based on “external sources” already. On the provider side, we find “external communications” – in this case the pictures released before the show, showing a spectacular show across all of London (upper picture above), further fuelling the high expectations.

Of course, what followed was nowhere near what the external communications suggested; let alone what WOM and personal expectations were suggesting. Thus, the problems occurred around the “Gap 5″ (circled in red), which means that actual expectations and perceived experience were no match for most people. Further, there may have been further gap-issues, though it is likely to be hard to find clear evidence of this. For example, it may be likely that the management thought that the show was indeed what the “customers” expected (Gap 1). Or maybe the management expected better, but these expectations were not clearly specified (Gap 2). Even when specified clearly, the expectations may not have been met “on the front line” (i.e. Gap 3). Or, maybe the marketing department thought the event would certainly look like what they promoted – although it was not planned to be ever anywhere near as spectacular (Gap 4).

Whatever the various gaps may have been, the outcome is a PR disaster for the Shard (which certainly could do with some good PR!). And what can we practically learn from it? Firstly, of course, there is nothing as practical as a good theory – and secondly, there is nothing as good as an applied cogent theory. Just imagine if the expectations would not have been so hyped? I’m sure today’s headlines would have been more about the outstanding engineering and the tallest skyscraper in the EU…

14/03/2012

Emotional and Rational Advertising

Developing the idea of product involvement further, Puto and Wells (1984) suggested that advertising can use one of two strategies: Either a rational, informational strategy, relying heavily on transmitting information and facts – or alternatively a “transformational” strategy, characterised by trying to make emotional connections between the brand and the consumer. Taylor (1999) expanded the dual strategy choice from Puto and Wells in his “six-segment message strategy wheel”. The wheel illustrates three types of message strategies for both emotional/transformational and informational advertising.
Ego strategy messages focus on the “this is how I am” messages, emphasising ego, while Social strategy messages emphasise receiving recognition or admiration from others as a result of buying the product (for example as a gift). Sensory messages are associated to feelings, i.e. how does it feel to taste the product etc.. All these message strategies are primarily emotional or transformational – as opposed to the three informational strategies on the other side of the wheel, which are:
Ration(al) strategies, based on clear arguments, USPs and convincing consumers of the product benefits. Acute Need strategy is more typically associated for products where there is a sudden need – and limited time to gather information. Thus, building brand familiarity is important for this category. Finally, Routine strategies are related to items often bought frequently and habitually without much deliberation, using elements such as exaggerated claims.

The “wheel” can be used to create strategies, i.e. it can be used as a way to “prompt” thinking about a product. Different styles can also be combined – for example: The delicious feeling of a chocolate (sensory) as a reward for a long week on Fridays (routine). Further, the strategies can be used to analyse advertising or marketing messages – for example to see how strategies change during an economic crisis (Lee, Taylor & Chung, 2011) or how company websites try to “sell” their message (McMillan & Lee, 2003).

8/03/2012

Advertising: Step by step – or all at once?

In a previous post, I discussed McGuire’s Persuasion Matrix as a way of looking at how persuasion works. The model consists of 12 steps that consumers complete to be persuaded and is more extensive than easier models often taught (such as AIDA = Awareness – Interest – Desire – Action/Think – Feel – Do etc.).

While the argument that consumers follow a series of steps to be persuaded is easy to understand, intuitive and logical, over the the last few years, both practitioners as well as fringes of the advertising research community seem to challenge this assumption. Vakratsas & Ambler (1999) looked at the various models used to explain advertising effectiveness in the literature. They identified 7 models, of which 6 are based on a hierarchy of effects.

Firstly, Market Response Models relate advertising (or other marketing measures) directly to purchase behaviours. I.e. the assumption is “exposure=effect”. Vakratsas & Ambler give the example of measuring loyalty directly by looking at repeat purchases – and therefore not taking into account different cognitive and mental effects.
The second set of theories are sumarised as Cognitive Information Models, assuming a (mostly strict) rational consumer behaviour, where the role of advertising is mainly to inform consumers of product attributes, which are then processed rationally and acted upon.
Pure Affect Models assume the opposite of cognitive theories: they emphasise a largely affective (liking, feeling, emotional) response to advertising messages leading to purchase behaviour. I.e. having a positive affective response (e.g liking the spokesperson) to an advertisement leads directly to a purchase. This can include familiarity of a product being established via advertising, leading to purchasing.
Persuasive Hierarchy Models are probably the most widely used models of persuasion – with McGuire’s model – or indeed the Elaboration Likelihood Model falling under this category. The basic assumption is that cognition is followed by affect which results in behaviour, rational deliberations result in attitude development that results in purchase.
Challenging the active form of cognition assumed in the Persuasive Hierarchy Models are Low-Involvement Hierarchy Models, which assume that (a low level of) cognition, i.e. merely being aware of – rather than actively deliberating about a brand, leads to an experience followed by affect and attitude formation. For example, Ehrenberg’s “Awareness-Trial-Reinforcement” model would be a typical example of such a model, addressing the processing of information particularly in low involvement situations. The important point is that these models assume that attitudes towards the products are created after product trial – while in the previous category, (positive) attitudes lead to product trial – and are therefore assumed to be formed prior to a purchase.
The last hierarchy based category combines Integrative Models, i.e. models that assume some hierarchical and sequential occurrence of cognitive, affective and experiential processes – but which argue that the sequence is based on the context in which they occur. For example, the FCB grid (Vaughn, 1980) would be a typical example of assuming product-category involvement to influence the sequence of processes.

The final category is probably the smallest in terms of research endeavors, but offers the most eclectic view of how advertising works. Hierarchy-Free Models don’t assume that any hierarchy of effects: Advertising is not seen as directly persuasive, decision making is not rational and advertising is seen as simply contributing to a “brand totality”. Taking a post-positivist/postmodern view, researchers in this category reject the idea of trying to measure advertising as a direct influence factor, focusing more on in-depth analysis of advertising, using, for example, feminist theory, semiotics or anthropological approaches to explain advertising (and its effect). While from a strictly rational perspective, such attempts may be seen as not practically helpful (e.g not providing practitioners with a way of measuring effectiveness), it is interesting that most practitioners support the view of advertising effectiveness as hierarchy-free when asked about their own theories (Nyilasy & Reid, 2009). Moreover, neuroscience appears to suggest that the brain itself is not processing information hierarchically – but rather interconnectedly and thus hierarchy-free. On the other hand, one could argue that the practitioners’ view of advertising itself may be contributing to the largely elusive effectiveness of advertising (Lodish et al, 1995, Sethuraman et al, 2011 etc). Of course, proponents of the hierarchy-free models would counter-argue that these measurements themselves are flawed.
What do you think? Can we measure advertising effectiveness? Are effects largely hierarchical (and sequential)? Or eclectic and concurrent?

6/03/2012

I’m (very/not) committed – Involvement

Involvement is one of those frequently used notions in marketing – that seem easy to understand, but that can get quite complex once they are looked at in more detail. The problem is that it can mean different things in different scenarios – and it’s therefore important tho clarify what type of involvement is being talked about: I.e. we can talk about product involvement, media involvement or message involvement. Let’s take each of these in turn:

Product involvement, is probably the classic idea of “involvement”, conceptualised as the consumers’ perception of the significance of the product category, based on their innate needs, values, and interests (Mittal 1995). This is the closest type of involvement to the original Robertson (1976) idea of involvement. This is the suggestion that some products, for example expensive products, have a high consumer involvement – and therefore consumers want a substantial amount of information before committing to a purchase. Conversely, for low involvement products, often inexpensive, everyday goods, consumers require a lot less information. Robertson’s idea was further developed into the FCB grid which helps to script advertising messages.

Media involvement on the other hand can also play a role in how a message is perceived (Krugman, 1965). For example, radio (in the background) can be seen a very low involvement media type – while reading your favourite magazine (the one you read from cover to cover as soon as it arrives) could be seen as very high involvement media type.

The final type is message involvement, i.e. the extend to which someone cares about specific message features. This type of involvement is likely to be highly personal, for example, someone who is interested in the environment may pay close attention to advertising messages that imply ecological product features. Conversely, someone who does not care about the latest technology may not pay any attention at all to messages about technical specifications of a mobile phone.

So what drives involvement? Well, this probably depends to some extent again on which one: Message involvement is most likely to be driven by interest in a subject matter and highly personal; media involvement is likely to be contextual and personal (e.g. some people will pay close attention to certain programmes, others won’t). Product involvement on the other hand is likely to be more complex than the often promulgated high/low price definition. Blair & Eagly (1989) use three types of motivators for involvement: (1) Value-relevant involvement, based on enduring values a consumer holds, (2) outcome-relevant involvement, where consumers are focused on the achieving a defined outcome and (3) impression-relevant involvement, where the consumer/scenario is driven by the impression they make on others.

A final point: Although involvement is often generalised as being particularly positive for a marketer – this may not always be the case. Imagine for example the case of people who are highly involved in ethical trading practices (i.e. they have a high value-relevant involvement). If a message contains relevant clues (e.g. hinting at fair trade deals etc) – then these consumers are likely to become highly involved in the message – and by consequence paying especially close attention to everything that is being said. Take the example of Starbucks, who very vocally advertise their commitment to fair trade coffee – and the criticism from message involved consumer groups (e.g. in 2000, 2007, 2011 …).

5/03/2012

I know you – you know me: Persuasion Knowledge

Developing the ideas of Attribution Theory further, Friestad & Wright suggest that commercial persuasion can be understood similar to a game play: Marketers try to “hit” the consumers, while the consumers in turn develop mechanisms to cope. Over time, both the marketers and the consumers develop “Persuasion Knowledge”, either made up of some knowledge about the product (for both sides) – and for the marketers knowing what works to persuade consumers – and for the consumer knowing how to cope with a persuasion episode. In other words: For the marketer (in the terms of the model called “persuasion agent”), persuasion knowledge derives from “topic knowledge” (e.g. product information) and “target knowledge” (e.g. use of celebrities will make target audience buy something) – while for the consumer (the “persuasion target”), persuasion knowledge derives from “agent knowledge” (e.g. this company is usually reliable) and “topic knowledge” (e.g. the product is long-lasting).

Some interesting studies have tried to shed more light on both target and agent knowledge: For example Nyilasy & Reid examined target knowledge and identified two widely held beliefs: “break through and engage”, i.e. cut through the clutter with creative, engaging advertising – and a “mutation of effects”, i.e. consumers will become resistant to advertising over time.
On the other side, being familiar with a product seems to moderate persuasion knowledge in consumers, as does knowledge about the agent itself (e.g. the company or organisation).

Many studies have looked at the effect of activated persuasion knowledge – especially focusing on the negative effects (from the point of view of the marketer). For example, knowledge about sponsored search results resulted in a reduced click through rate for internet searches, and frequently presented product placements resulted in negative brand attitudes if persuasion knowledge was high.

A hotly debated aspect of persuasion knowledge is how (and when) it develops – and how it’s perceived negative effects (and it is not always negative!) can be limited. For example, Mallinckrodt and Mizerski showed that for young children (aged between 5-8) persuasion knowledge had no effect on brand preference, suggesting that very young children have not yet developed coping mechanisms. However, college aged consumers coming from the millennial generation were shown to develop more negative perceptions when persuasion knowledge was high. Looking at advergames, Waiguny, Nelson & Terlutter showed that being very engaged in a advergame (and therefore having relatively little “spare” cognitive capacity) resulted in reduced persuasion knowledge activation.

However, as previously mentioned, persuasion knowledge does not have to be always negative: In an interesting study looking at product placement, Cowley and Barron reported that only people who liked a programme used for the study developed negative brand attitudes – while for people who did not like the programme the inverse was true. Equally, in a charity context, persuasion knowledge may actually increase the intention to donate.

Although there are many studies proving the existence of persuasion knowledge and coping mechanisms, there is a surprising lack of studies addressing exactly how coping mechanisms work (i.e. dismissive of all attempts? Nuanced?). There are, of course, many anecdotal examples – mostly focusing on what happens if consumers find out after the persuasion episode about the commercial nature – and then feel deceived (as for example in the fake Sony PSP blog episode). Similarly, McDonald’s got mocked when using sponsored tweets – resulting also in a “counter-tweet“.

So while on the one side persuasion knowledge (on behalf of the consumers) can be seen as particularly negative point for the marketers if it inhibits persuasion – in other ways it can actually be seen as positive – for example in the case of charity fundraising – or as a way of “shielding” the consumer from unrealistic or overly positive persuasion attempts by marketers. However, there is still a need for more research to understand the role of different aspects of persuasion knowledge – and how and when it develops.

1/03/2012

Is it me – or is it it? Attribution Theory

Continuing from the previous post about persuasion, a different way of looking at persuasion is from the perspective of the receiver – and what goes on on their “side” of the process. Late in the 1960s, “Attribution Theory” emerged as a way to understand what motivations are ascribed to senders by receivers – and how receivers “figure out” which factor causes the sender to say something (i.e. attribute a motivation). Basically, the theory recognises that if, for simplicity let’s say, a strange person talks about a product or service, then this can be attributed either to a personal reason (e.g. the person is making a profit from it) or it can be attributed to something about the product (for example, it’s a really excellent product).
Rather than following a simple decision rule (i.e. being a heuristic model), Attribution Theory proposes that individuals evaluate what the stranger says using three types of “cues” – or “covariant information”, derived from the receivers perception of the context in which this message is being sent. Taking the example of a restaurant, the three variables are:
Consensus – i.e. how far the message being received is coherent with what other people are saying (e.g. are they all saying the restaurant is good).
Consistency – i.e. is the stranger always saying the same thing (e.g. is the stranger repeatedly praising the restaurant on several occasions) and
Distinctiveness – i.e. is the message about different to other messages from the same sender (i.e. does the sender always talk highly about every restaurant – or are they usually critical, and this time it’s a different story).
Once received, the information is evaluated against these covariation information variables, and and the judged to be either high or low.

For example: If other persons have talked well about the restaurant (high consensus), the stranger does not normally talk well about other restaurants (high distinctiveness) and the stranger talks repeatedly well about the restaurant the receiver is likely to attribute the motive of the sender to be linked to the restaurant (i.e. something about the restaurant is good).
Conversely, if other people talk negatively about a restaurant – but the sender talks positively (low consensus), the sender always talks positive about every restaurant (low distinctiveness) and the stranger varies what he says about the restaurant (low consistency) then it is likely that the receiver will attribute the positive restaurant evaluation to something related to the stranger/sender (e.g a profit making interest) – and not to be the result of a particularly good product.

In many situations of course, the receiver may lack some of the contextual information, and therefore may not be able to evaluate the message using the three covariation variables given above. However, the receiver is still likely to attempt to figure out what causes the sender to say something before being persuaded using whatever cues are available to construct what could be regarded as a “mini-theory” explaining what causes the stranger to say something.

Taken as a model, Attribution Theory (Kelley, 1967) provides a basic explanation for negative reactions towards advertising in general (e.g. every advert always raves about the described product; other people may not say that the product is good – even if the advert says it is the best ever etc…= it is something in the ad that is causing the positive stance, not the product that is actually good) and of course to the model cynicism towards, for example, social marketing campaigns when they are run by the government (e.g. “they only tell us this to save money” etc). A typical example of using attribution theory in practice is the (not really very ethical practice) of some department stores to hire “random” customers, to compliment customers considering a purchase seemingly randomly on their good choice, or saying something positive about the product the person is considering to buy. As the contextual clues indicate that the other customer is not driven by a reward (such as a sales assistant), attribution theory shows that the customer considering the purchase is likely to attribute the praise to the product (and therefore more likely to be persuaded that the product is indeed good). A more modern form of using attribution theory would be the attempt to generate word of mouth, for example online reviews by “unbiased” customers.

Attribution theory is also at the basis of many regulatory principles, i.e. take the example of advertising, where the persuasive attempt is made clear by giving contextual information (“this is an advertisement”) or showing information identifying the advertiser. One of the criticisms of product placement for example are because of the lack (or even falsified) covariation information, making it hard to determine if it is the writer/director/actor who placed the item there – or the company trying to sell the product ( or indeed, like in this example, for anti-product placement, i.e. competitors placing a competitor’s products to make the competitor look bad).

To summarise, attribution theory tries to describe the process of “attributing” what causes someone to say something. In the context of persuasion, this means the receiver of a persuasion attempt will try, based on contextual clues (or covariation information), to work out if the message is motivated by something internal to the sender (e.g profit, reward) or by something related to the product/service being talked about.