The audience analysis company Attentional recently announced that it will now be publishing TV programme-related social media activity in the form of TPK (tweets per thousand viewers) on a programme-by-programme basis, as an additional metric alongside the BARB viewing data it already provides to its customers.
The ability to view social media activity related to programming content in this way is an interesting development, and one I applaud. I can also see how it might be a great commercial move by Attentional: it ticks many boxes in terms of the current fascination with “big data” as well as recognising the growing importance of “second screen” activity – viewers interacting with apps, social media, websites etc. whilst simultaneously watching a programme on TV.
Moreover, the idea of some kind of “viewer engagement” metric that can be viewed alongside traditional ratings has been a holy grail for decades. The impact of a programme could and should be measured in terms of more than the number of viewers, especially for public service broadcasters. It also intuitively make sense that there will be individual programmes with relatively small audiences, but whose viewers value, appreciate, and engage with that content more highly than they do with some “lowest common denominator” mass audience programming.
To this end, major broadcasters have subscribed to audience appreciation surveys for many years, but there are misgivings about this type of survey data, which tends to show comparatively low levels of differentiation between programmes. Perhaps this kind of social media metric has the potential to offer us a more discriminating measure of engagement. After all, viewers are, in this case, literally engaging with the programme by talking about it to their friends; and the fact that it is a behavioural metric rather than an attitudinal one (it measures what people actually do, rather than what they think they do or think) is surely another positive.
However there are number of potential pitfalls in this kind of approach…
Although measurement of social media activity is – on one level – quantitative, in that you can count the number of tweets (etc.), it is also qualitative in that a tweet can be an expression of either positive, negative or neutral attitudes; in this way it is fundamentally different to viewing data such as BARB.
Caution also needs to be exercised in respect of how representative of the viewing audience such data actually is. Although pretty much everyone has access to a television, and audience measurement panels are very carefully controlled to be demographically representative of the population, not everybody uses social media, and certain audience groups are without doubt more socially active on these platforms than others.
Different content, different audiences, different levels of social engagement
The press release from Attentional highlights two example programmes: EastEnders (TPK: 0.79) and The Only Way is Essex (TPK: 32.6).
Are we to deduce from this that viewers were more engaged in TOWIE than in EastEnders? Or are these figures more a reflection of the demographic profile of their respective audiences, and the demographic profile of Twitter?
Such a comparison could be very misleading (and, to be clear, I don’t believe that Attentional are necessarilyproposing that the data be viewed in this way). What would be more interesting is to compare social metrics for programmes whose audience profile in demographic terms is more similar – so, for example, TOWIE versus Made in Chelsea or Geordie Shore.
Not all social media engagement is equally positive
The other potential danger in this kind of analysis is to assume that social engagement has a unitary value (i.e. that every interaction reflects an equal and standard measure of positive engagement on the part of an audience member). This is clearly not the case.
Consider these 2 examples… ￼ ￼
Social media activity can reflect a positive or negative level of audience engagement, or indeed (as is the case with the second example here) originate from someone who has probably not viewed the programme at all.
We therefore need to be a bit wary about approaching this kind of metric from a traditional audience measurement perspective. Perhaps a more helpful contextual framework would be reputation tracking in PR, where analysts seek to quantify the positive or negative net impact of a piece of media coverage on an organisation’s or brand’s (or, in this case, programme’s) reputation.
Let’s look at a hypothetical example based on Twitter reaction to the current season of The Voice UK…
Because the programme is so active on Twitter, we could reasonably look at social media reaction to the four judges (Sir Tom Jones, Jessie J, Danny O’Donohue and will i am), as one means of evaluating which is contributing best to overall levels of audience engagement with the show.
Based on sheer volume of social media activity, we might conclude that Judge X is contributing most to audience engagement with the show, on the basis that he/she is the subject of a disproportionate share of the social media activity around the show. Maybe a pay rise is in order before the next season…
However slightly more detailed analysis of the data might reveal a different story. The majority of the social media mentions for Judge X might be negative, whereas the mentions for judge Y, though fewer in number, are overwhelmingly positive. Who deserves the pay rise now?
Social media metrics are a fantastic addition to the armoury of data available to us for audience understanding. However they need to be handled particularly carefully because they are a complex hybrid of qual and quant. Moreover, it is important to remember that social media engagment and audience engagement in the broader sense are two different things. Confuse the two at your peril!