Content on this page requires a newer version of Adobe Flash Player.

Get Adobe Flash player

TODAY´S NEWS

Guest Column: Power outage and TV ratings

Guest Column: Power outage and TV ratings

Author | Mallikarjunadas CR, Aarti Bharadwaj & Ashwin Ramas | Tuesday, Feb 04,2014 10:24 AM

A+
AA
A-
Guest Column: Power outage and TV ratings

Power outage is a huge reality in India. Do TV ratings reflect this? For instance, in January-March 2013, Tamil Nadu faced 6-8 hour power cuts across the state. Many amongst the community did not, however, ‘sense’ it from the measurement system. There is often a debate between the planner and client that goes like this:

Planner: Here are the actuals of last month’s media plan – 500 GRPs, 45 @5+. There is a variation of (-) 3 per cent, which is pretty good (Big Grin)

Client: Boss, my entire salesforce tells me the plan was not ‘visible’ (planner winces; that ephemeral word ‘visibility’ continues to haunt him). It seems there were eight hour power cuts. I don’t think anyone saw our ads.

and this debate continues…

Media practioners still have a Mumbai-centric view of power availability, while political parties such as Aam Aadmi Party have won an election on power shortage!

India has a large energy deficit, which peaks in summer and is lesser during November-December. In summer, there are power cuts because the electricity provider is unable to keep up with consumer and industrial demand. Layer this with the fact that the penetration of power backup or uninterrupted power supply for televisions is miniscule in the country.

Power is not one of the factors in weighting TV ratings. The current methodology of calculating TV ratings relies on the assumption that the pattern of power availability in the population will mirror itself in the sample. This belief, or heuristic, is called as the ‘Law of Small Numbers’ by Daniel Kahneman & Amos Tverksy in their ground breaking research, ‘Judgement under Certainty: Heuristics & Biases’. According to this misplaced belief that many hold, even small samples are assumed to be highly representative of the population from which they are drawn. In the specific case of power, the big assumption one has is that ‘the dispersion of power-cuts in the population will reflect itself in the sample’. A common example of this fallacious heuristic is the ‘gambler’s fallacy’ where one believes that nature is fundamentally fair, and things will even out even in the short run. For instance, suppose a fair coin were tossed consecutively six times. Six is a small sample, and our friendly neighbourhood gambler were asked which of two outcomes were more likely:

Event 1: A run of HHHTTT or
Event 2: A run of THHHHH

Our gambler, and most people, would answer that Event 1 is more probable than Event 2. That is because it seems fair that a run of 3 Heads is evened out by a run of 3 tails. On the other hand the second event is seen as patently unfair. It cannot happen would be the wail of our gambler as by the laws of nature that which goes up must come down. Statistically however, both events have exactly the same probability, 1/64!

In the long run, things do even out versus nature. Similarly, with a large sample, things might be more representative of the population. However, there is absolutely no basis to believe that a small sample will represent in its entirety, the population.

Coming back to the power study, we at Starcom MediaVest Group felt that this problem of power was big enough to investigate, and that too from several perspectives:

1. If power is indeed an issue, then the CPT of television is artificially more favourable than otherwise vis-à-vis other media. For instance, this could provide additional reasons to investigate the role of mobile, and online video as modes of audio-visual entertainment

2. Were there patterns to power-cuts? Is it worth it to punt on more prime-time, more weekend, etc. in a media plan?

In partnership with Mubble Networks, we initiated a power availability live panel in Chennai and 1 Million+ towns in Tamil Nadu, and that too at a minute to minute level.

There were several challenges we faced most notably the capture of power-availability and the sampling plan.

The sampling plan involved looking at publicly available information on TNEB, mining newspaper data of the past and the census. A big assumption made, as we were studying planned load-shedding, was that power availability was SEC agnostic and is driven more by location. The key to finding out this is to get information on the mapping of areas X transformers. However, we did not have that data readily available in the public domain. The daily load-shedding of power by areas is provided by newspapers. By clustering these on a map, one can figure out zones that are ‘clustered’ and likely to have power being supplied by the same transformer. Thus, by a combination of data mining of newspapers and websites of daily power-cuts, we were able to guesstimate ‘transformer zones’. We then went about distributing our sample as per these transformer zones.

The next thing was to capture power availability at a minute to minute level. This is where Mubble Networks devised an elegant solution. We used SmartPhones to capture power availability.

The representative household in each locality would be given a smartphone to be plugged in round the clock. Usually, in every HH, there would be a rarely used plugpoint, like under the wash-basin etc. An app on the smartphone triggers a message to a central database every time the power supply goes. Another message is triggered when it is restored. Therefore, this database has captured the availability of power for Chennai & 1 Million + TN at a minute by minute level.

Again, matching this data with the population of each locality gives us exactly what percentage of a target audience does not have access to power and therefore, could not be watching television. This percentage, therefore, leads us to the Power Deflation Index. This index, in essence, reduces the reach of a particular program by the proportion of population that did not have power during that time.

We integrated this index into SMG India’s proprietary Optimizer, TARDIIS. Thus we had live information on a minute to minute level. The entire process is illustrated below-

Our critical findings included that CPT of Television without power might be over-optimistic. When this is factored, other media, on sheer cost considerations, might not be so unfavorable. We feel this becomes even more critical as TV advertising becomes isolatable – especially in Hindi Speaking Markets with technology such as Amagi and what some of the other channel networks are planning. If one were to make a plan for say, Bihar, what would be the CPT of TV post factoring in power-availability versus mobile/ online video? As TV advertising transforms itself into video neutral advertising, it is even more critical for a marketeer to get as realistic about factors such as power that are fundamental to determining ROI.

We believe, this research is a major leap toward the new age in media research in India. The research used mobile phones and unconventional research techniques to capture accurate data and answer an age-old question regarding measurement in conventional media. We were severely limited by funds of course, and information and had to use data-mining to determine aspects that publicly available data could not fill. The objective of this study is not to stake claim as the most perfect study but rather as a first, albeit imperfect step, in a direction that is critically important.

Mallikarjunadas CR is CEO, SMG India.

Aarti Bharadwaj heads SMG India’s Analytics Centre of Excellence.

Ashwin Ramaswamy is CEO & Co-Founder, Mubble Networks.
 

Write A Comment