By using the PRSV algorithm, or any part of it, the user is agreeing to the following:

Creative Commons Licence
PRSV for Media Coverage by Crescendo Consulting is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License. Based on a work at http://crescendoconsultinguk.com/prsv-open-source-algorithm/.

If you have any questions or feedback please contact markwestaby@crescendoconsultinguk.com

 

PRSV has been designed and developed specifically to reflect the business benefits that PR activities deliver, ie the larger the PRSV value the greater the business benefits.  It does this by reflecting the impact of PR – primarily but by no means exclusively, media coverage – on audience behaviour in key decision-making processes, including purchasing.  Here we present the algorithm used to apply PRSV to media coverage.

The fundamental principle on which PRSV for media coverage is based is the probability, or likelihood, that audiences will both see and engage with the coverage. In this context ‘engage’ is defined as “the emotional, cognitive and behavioural connection that exists, at any point in time and possibly over time, between a user and a resource” (Towards a Science of User Engagement).

Such engagement matters hugely in a world dominated by clicks, likes, impressions and shares, none of which truly reflect the impact of marketing on audience behaviour.  For instance, a Columbia University study revealed that 59% of consumers who share an article through social media have not read it and that sharing content and actually reading it are poorly correlated.  The Columbia study inevitably concludes that likes and shares are not a meaningful measure of content popularity.

In addition, our own work at Crescendo has revealed, first, that only a small fraction of people who ‘like’ a Facebook page ever go on to engage with it; and, second, that out of hundreds of thousands of ‘clicks’ through to a website, only a handful might actually engage with it.  In short, while ‘big numbers’ look impressive they are meaningless.

As the PR industry faces increasing competition from digital marketing activities that enable ROI to be measured and tracked, so it must embrace metrics that reflect the true business benefit – and ROI – that it delivers.  The good news for PR is that it is playing an increasing role in delivering real business benefits, including the crucial one of purchasing.

For further details we strongly recommend our PRSV White Paper, which can be downloaded here.

PRSV has been developed to provide a robust and credible alternative to the flawed and discredited Advertising Value Equivalent (AVE), which should never be used to measure the ‘value’ of media coverage

Some important points to note before using PRSV:

  • Unlike advertising value equivalents, PRSV is not designed to produce the biggest ‘number’ possible
  • PRSV is based on the fundamental premise that real value can only be realised when audiences find and are engaged by media coverage
  • If you have carried out a ‘valuation’ of media coverage based on AVE it is therefore highly likely that PRSV will return a lower financial ROI.  Similarly with audience figures. PRSV reports these based on the probability that audiences see and engage with coverage
  • PRSV is not a media monitoring service.  The media coverage analysed is based on what the majority of an audience is most likely to have seen and been engaged by over the period in question, not the number of pieces of coverage
  • Volume of media coverage does not equate to audience reach, exposure or engagement and does not reflect true business value or benefit.  Basing any decisions on volume of coverage alone carries a very high risk
  • The real impact of media coverage typically follows the ‘80/20’, or ‘Pareto’ rule.  Put simply this means that 80% or more of business benefit is generated by 20% or less of media coverage.  This is the main reason why decisions based on volume of media coverage alone carry a very high risk
  • The PRSV algorithm for media coverage is presented as a set of instructions and formulae, which are designed to be as straightforward and easy to use as possible, with a minimum of technical expertise

 

PRSV for Media Coverage Algorithm, Part 1: determining ‘viewing’ and ‘engagement’ probabilities, which are defined as the probability, or likelihood, of an audience member viewing and engaging with coverage, respectively.  Part 1 is itself split into two parts.  Part A determines viewing and engagement probabilities for national and general interest media; and Part B determines viewing and engagement probabilities for trade and specialist media.

The following sources provide useful background to the algorithm.

  • Share of interest by type of story (research conducted by Crescendo Consulting)

 

Part A: calculation for national and general interest media (including national broadcast media such as the BBC)

i. Determine the viewing probability of the piece of coverage by defining the type of coverage from the following table

Type of story

 

General news

Viewing probability

18.77%

Sport 16.79%
Tv/showbiz 11.79%
Science/tech 10.53%
Food & drink 9.77%
Life and style 9.70%
Business 9.68%
Education 9.65%
Media 9.63%
Travel 9.56%
Money 9.56%
World news 9.50%
Politics 9.45%
Local 9.43%

 

ii. Determine the linguistic subjectivity of the piece of coverage using the following formula:

Linguistic subjectivity = number of sentiment adjectives in piece of coverage divided by number of other adjectives in the same piece of coverage

Note: if the number of ‘other’ adjectives in the coverage is less than 10, the linguistic subjectivity automatically defaults to 0.01

A list of sentiment and ‘other’ adjectives can be downloaded as an excel file below

 

Sentiment and other adjectives~Jul 2017

 

iii. Determine the engagement probability of the piece of coverage using the following formula:

Engagement probability = Viewing probability multiplied by linguistic subjectivity

 

Part A example: we have a piece of coverage that appears in May about mobile phones on a UK national media site.  The number of sentiment adjectives in the coverage is 11 and the number of ‘other’ adjectives is 99.  From Part A (i) the viewing probability = 10.53%; and from Part A (ii) the linguistic subjectivity = 9/99, or 0.091

The engagement probability, therefore = 0.1053 x 0.091, or 0.0096

Note: this is clearly a small number, which is not untypical and it reflects the reality of media coverage being seen and engaged with.  It should, however, be remembered that the total audiences for many general interest media are very large; and as we shall see in Part 2, even a small engagement probability can result in a very significant audience.

 

Part B: calculation for trade and special interest media

i. Determine the viewing probability of the piece of coverage. In trade and specialist media this will be high

Viewing probability for trade and specialist media = 90%

 

ii. Determine the linguistic subjectivity of the piece of coverage using the following formula:

Linguistic subjectivity = number of sentiment adjectives in piece of coverage divided by total number of other adjectives in the same piece of coverage

Note: if the number of ‘other’ adjectives in the coverage is less than 10, the linguistic subjectivity automatically defaults to 0.01

A list of sentiment and ‘other’ adjectives can be downloaded as an excel file above

iii. Determine the engagement probability of the piece of coverage using the following formula:

Engagement probability = Viewing probability multiplied by linguistic subjectivity

 

Part B example: we have a piece of coverage that appears in February about video games on a UK media site devoted to gaming.  The number of sentiment adjectives in the coverage is 25 and the number of ‘other’ adjectives is 100.  From Part B (i) the viewing probability = 90%; and from Part B (ii) the linguistic subjectivity = 25/100, or 0.25

The engagement probability, therefore = 0.9 x 0.25, or 0.225

Note: although still quite small, this is clearly a much larger number than the engagement probability for national and general interest media.  Again this is not untypical and reflects the reality that media coverage in specialist media are much more likely to be seen and engaged with.  It should, however, be remembered that the total audiences for many specialist media are relatively small; and as we shall see in Part 2, even a large engagement probability can result in a relatively small audience.

 

PRSV for Media Coverage Algorithm, Part 2: having established the ‘engagement’ probability for a piece of coverage, we now need to determine the financial ROI and size of audience for it.  In order to do this we make use of traffic and traffic cost for the URL (domain) where the coverage has appeared. You can find this – free of charge up to a limited number of searches and for a small monthly charge beyond this – by going to semrush.com.  We use SEMrush because it is widely recognised as the best tool in the marketplace for this type of data.

Once on SEMrush type in the URL (domain) of the site you are searching, eg dailymail.co.uk, and both ‘traffic’ and ‘traffic cost’ will come up. ‘Traffic’ is the monthly audience for the site being researched while ‘traffic cost’ is the amount of money that people are bidding on the open market to appear in results for search terms that send audience traffic to it.

Note: remember to click on the country for which the traffic cost is being determined, which is shown towards the top of the SEMrush page.  It is also important to check the currency for traffic cost and, if necessary, to convert this to the currency you require.  If you are using the SEMrush API to gather traffic cost for bulk domains please note that SEMrush usually delivers these in US dollars regardless of the country of the Google database from which the data is taken.  If in doubt, please check with SEMrush

Now that you have the traffic cost for the piece of coverage you need to determine its traffic value for the PRSV. The calculation to determine traffic value is as follows:

Traffic value = Traffic cost x engagement probability / (divided by) number of days for the month in which the coverage appears

Example: using our piece of coverage about mobile phones that appeared on a general interest media site in May, for which the engagement probability was 0.0096.  From SEMrush we determine that the (monthly) traffic cost is £37.1m while the traffic (audience) is 53.3 m.

PRSV Traffic value = 37,100,000 x 0.0096 / 31

PRSV Traffic value = £11,500

Where traffic cost is £37.1m, engagement probability is 0.0096 and the number of days in the month (May) is 31

The overall PRSV Traffic Value is the sum of PRSV Traffic Value for each individual piece of coverage

Note

  • If the coverage is negative the PRSV Traffic Value is zero, but PRSV Audience Traffic for negative coverage should still be reported
  • PRSV Audience Traffic is the size of the audience that has a significant probability of seeing and engaging with coverage
  • PRSV Audience Traffic is determined by repeating the calculation for traffic value but using the ‘traffic’ figure from SEMrush rather than ‘traffic cost’

 

Example: using our piece of coverage about mobile phones that appeared on a general interest media site in May, for which the engagement probability was 0.0096.  From SEMrush we determine that the (monthly) traffic cost is £37.1m while the traffic (audience) is 53.3 m.

PRSV Traffic audience = 53,300,000 x 0.0096 / 31

PRSV Traffic audience = 16,500

The overall PRSV Audience Traffic is the sum of PRSV Audience Traffic for each individual piece of coverage