James Walker – Does Size Matter…And Position, Colour And Content?
James Walker, joint managing director, Advanced Techniques Group Worldwide, JWT
So… Project Cosine – this has been all about creating cardinal indices to compare against cost premia in the UK national press market, for the effectiveness of different sizes of advertisements, and then to go on to quantify the impact of colour versus mono, positioning in the paper, solus versus clutter…., and left hand versus right hand pages, etc.
However…, for this paper – in keeping with the spirit of the conference (facing up to the challenge of media research in a rapidly changing commercial environment) I thought it worthwhile to present this study in the widest possible context… including why we carried out the project…, how we are using it…, and what we are going to do next.
So…. this presentation has six sections:
– Firstly… setting out the context of the study….., ….then going on to explore:
– our objectives
– the methodology
– show some brief examples of the results
– explain it’s practical application within the body of the JWT Media Dept
– and finally I’ll speak at the end about our next steps.
So let’s begin with asking the question, “why did J Walter Thompson commission Project Cosine ? ”
The answer is very simple. J. Walter Thompson as a World-wide company recognised the growing importance of print advertising to us… We didn’t do this work out of charity, but because of specific commercial reasons.
And I think there are a number of factors at play here:
– Firstly, future growth in advertising expenditure is likely to be from sectors other than the fmcg TV heartland.
– Secondly, around the World, it’s TV that is first hit by advertising restrictions – thinking of, for example,….tobacco and alcohol.
– Also, there’s an increasing realisation of the growing polarisation of television audiences, and from that, the necessity of mixed media strategies…
….so more of our television activity has to be accompanied by an element of print.
As this chart shows (using a real life UK example) by creating sextiles ranked by weight of viewing, you can see that 400 Hws ratings (giving an average of 83% 1+ cover at 4.8 OTS) actually delivers 750 ratings against heavy viewers and a mere 140 GRPs against the lightest viewers. And this light viewer/heavy viewer polarisation is unambiguously getting worse.
(And if there is anything that I can claim to have achieved in my one year with the JWT London Media Department as Media Development Director, then I think it was in the area of mixed media, and persuading our clients to think beyond television)
A further real life problem, right now, across media is declining coverage ceilings – and this is especially acute with television, where we predict that over a ten year period the coverage ceiling for 400 Hws ratings on ITV and Channel Four will have fallen by 10%tge points…. from 86% 1+ in 1991 to 76% in 2001.
And whilst the national press is not immune to the same problem, by combing a number of media together, overall coverage declines are greatly reduced….. So we have a strong imperative to switch money out of television (and this is true of JWT World-wide).
So a corner stone of our global research and development programme is learning more about press…
Hence this project.
Additionally, just quickly:
- Overall, the advertiser mix is changing. In London, the bias of our existing clients is changing…. in favour of such sectors as, for example, financial services… so that’s a positive for press also.
- And, we’re realising that against the back drop of client demands for increased media accountability. Cosine (and it’s TV equivalent Tangent) have an important role to play.
- Thirdly, the way we’re planning at JWT is changing, and it’s worthwhile stating that Cosine is a product of our new media planning process…. so the use of Cosine and similar research tools is written into the new process manual of JWT Media World-wide.
- And finally, as our clients’ structures have regionalised and become more international, we have created a global media research agenda – executed by the Advanced Techniques Group, based in London. So… work done in London has to have a replicable methodology across countries, and indeed Cosine is being extended to a further 4 European countries this year, and has become an explicit global project (although in truth the methodologies will end up varying greatly around the World)
Cosine is an organic project (a 3 year commitment), that was always intended to grow beyond the results that I’m presenting this afternoon.
So this year in London, we’re looking at weekend newspapers and consumer magazines (depending on client and media owner co-funding), and the whole programme is being rolled out across the major European markets.
It is important also to set out that this is part of wider work in our project Atlas to better understand the comprehension of newspaper ads versus other media, and Cosine has to work within the context of our Atlas and ComAssessor mixed media research for the world-wide company.
So. That’s the context.
In terms of objectives, we were intending to learn three things from the research.
Finding out more about:
– Normal reading behaviour (so that’s: frequency, motivations, sections etc)
– Page traffic
– And…, noting scores to create the indices for the value of size, colour, positioning etc.
As far as the indices are concerned, it was our intention to create a 10 by 10 matrix of factors, for which we would have cell scores enabling us to look at any combination of factors.
So for example, be able to combine factors to give us a score for: a full page ad, front third, colour, right hand page, in a mid-market title, etc.
There were some gaps in the matrix (which we used modelling to fill)… so, what we wanted to end up with was a giant spreadsheet, and a simple model, enabling us to compare any combination of factors against rates in the market.
So they were our objectives.
Moving on to the methodology.
Working with Jennie Beck, we were looking at:
- National press, Monday to Friday
- A sample size of 500
- Face to Face interviews
- Conducted in the South East, last Autumn
So…. now, the results.
For the purposes of this presentation, I’m just going to talk about 5 sets of results, as examples. You will understand why we want to keep the bulk of results confidential.
Really just for illustration, I’m going to talk about:
- When read
- Time spent reading
- Ranking of interest in editorial content
- The relationship between reading and noting (which is very interesting)
- And… major on our indices for quantifying the effect of colour, size, positioning etc. (take a long pause)
‘When read’ gave some interesting results, that on a couple of pieces of business have already helped us with planning around the idea of chasing a target audience through the day as they work their way through their ‘Media Diets’. Not a new idea, but this research has been useful for us.
Actually looking at ‘when read’, what we saw were very different results by demographic groups, and very different results across newspaper segments.
I particularly note some highlights, just for example, here:
- 29% reading qualities at breakfast
- 26 % reading mids before dinner
- 33% qualities read in the evening.
Reading time finally answered a question I personally had always struggled with in my previous job (when I was Director of Media Consultancy at the Henley Centre, and worked with several national newspapers), which was…
“Does increasing the content of newspapers lead to longer reading times ?”
The answer is that incremental pagination is not marginal, in the sense that reading time is more content dependent than time constrained. Days of high/low pagination give high/low reading times.
And I very much look forward to the UK weekend study, which should be very interesting in this regard.
Editorial reading preferences vary hugely by demographic group. (I was slightly scared that this study could give potentially ‘flat’ results’, but that has not been the case at all).
Staggeringly, 22% of readers under 25 don’t look at all at foreign news, and indeed for all readers, celeb interviews and gossip outscore foreign news.
Something we’re going to use in models from future studies, is a really interesting, and totally linear, relationship between reading and noting……. Whereby, ad noting increases 1.4 times faster than reading.
So, and I guess this actually pretty obvious, it’s very important to be on the well read pages, and the other thing is…, that ad noting scores are very easy to predict from ‘reading’.
Next is factor hierarchy – so we’re on to the scores now.
The order for the individual factors we looked at, by greatest influence, is:
- Size
- Colour, rather than mono
- Front third position
- Solus, and…
- Right hand page
The chart shows indices where the ‘true’ average ad (which is obviously just an abstract construct) scores 100.
In the meantime, I’ll just run through some of the raw factor scores.
Not surprisingly, size is the most important factor…
… And this is an incredibly misleading chart, because a casual observer might actually read the chart as presenting the exact opposite of the reality….
Which is in fact that, measured by column centimetres, you actually see very strongly diminishing returns for greater size. So against market rates, a 25 by 4, actually scores optimally.
Colour….. Colour is best thought of as making a mono ad as effective as ‘the next size up’. So adding colour to a mono 25 x 4, makes it as effective as a mono 38 x 6. Add colour to a mono 38 x 6 and it’s as effective as a page mono (in this broadsheets example). And the benefit of colour declines, with increasing size…. but it always beats the market premium that you pay.
In this chart we can look at, in a bit more detail, the relationship between the indices generated by colour and mono sizes (in this case, again from the broadsheets aggregate) and the ‘benefit’ of adding colour.
What this chart very clearly shows is that the ‘benefit’ of colour diminishes with increased ad size (adding about 50% effectiveness to a full page rather than plus 100% to a 20 double). Therefore, to unravel the effect of colour… what is required is a non-linear model to explain the size/colour benefit interaction.
And this is what we are doing at the moment, and we have a non-linear model completed shortly… although that said, the linear model we are using at the moment is perfectly serviceable.
So if now we can go on to talk about position.
Across both broadsheets and tabloids, the pure scores are clear – in broadsheets, a front third position out performs the back third by a factor of almost four.
However, as I have previously alluded to, there is a problem with this chart, in that these scores are only from ‘the edge’ of our factor matrix, and that combinations of sizes and colour (for example) with position will give different effects. So the broadsheet example here, for instance, will be polluted by the high number of full pages in the first third of book….
So these scores aren’t too useful to you.
And so finally in terms of the factor scores, the much talked about right hand page versus left hand page issue. In the tabloids, there is very little difference between the two, whereas in broadsheets, right hand pages score about 40% stronger than left hand pages (and these scores are largely unaffected by the other factors).
And so now in terms of results, what makes sense, I think, is to show an illustration of how we have unravelled the individual factor scores.
What it is possible to do, is to create a multi-dimensional matrix (in this example, just a two dimensional matrix… allowing us to look at combining just two factors) plotting the different factors against one another.
Taking care to weight the cells by the number of ads in our sample, it’s a simple piece of regression to unravel each factor in isolation… assuming a multiplicative , but linear (which we know is a slight simplification), model.
Also, I can now show you the isolated factor scores (excluding for this presentation, size, which we wanted to keep confidential).
There are a couple of problems with this analysis, that I won’t seek to obfuscate:
- The first is, as I have said, we are working with a linear model so far – which works fine, but is a bit of a simplification.
- Secondly, for presentation purposes, I have assumed ‘symmetry’ within the factors to give a base of 100 – which is slightly unrealistic….
So, the isolated scores put a front half position (excluding all other factors) at three times more effective than a back half, colour at twice as effective as mono, and so on……
By using these values, we can predict scores for any combination of factors, and so optimise size, colour position etc against market rates. Which is pretty neat really.
Moving on to my fourth section – which is ‘application’.
What we are hoping to do with our whole research agenda, both in London and now for the World-wide company, is to affect a cultural change in the agency. And what we have been seeking to do is two things:
- Firstly, enhance the discipline within our Media Departments (creating a more analytical culture), and,
- Secondly, using research as a catalyst for media creativity.
And I think this project allows us to both, in that:
- The scores have brought more discipline to our press planning and buying, and,
- The broader aspects of the study have worked as a spark to ignite more creative thinking in our use of press – and you can see a very good example of this with our Jaguar business, where smaller sizes, and many different creative executions, and use of colour, are working together well, at the moment.
Again, just to emphasise, that we see this project as a calibration of a planning process that we have already set out as a vision for Media World-wide. So this type of analysis is written into the new process manual for JWT, and so this is not just research that will appear at conferences and on the pages of ADMAP…., but will actually be used.
Thirdly, it’s also worth saying that what’s really interesting about this project is that, if you start generating theoretical optima for sizes, use of colour etc., against budgets, it begs some questions as to: how many insertions you need, what the burst pattern should be, how the creative executions should be phased, and how do all the elements combine best.
And also I think, this project as enabled us to have a more informed dialogue with the rest of the agency – the Creative Department in particular. And again, what this project has done, is to make us question how the different elements of press advertising fit together, and so it has had far more of a profound impact on our thinking than merely looking at the noting scores versus market premiums.
However, this project… has… given us a set of norms to frame our discussions with clients (and open discussions with media owners). And whilst this is absolutely not meant to be a straightjacket discouraging creativity or topicality and tacticality in press advertising, clients do seem to like having some numbers the starting point.
Finally, next steps.
What we are doing with Cosine, and as part of our separate ComAssessor research, is to explore – though still very tangibly – the actual impact of perceived creativity (and how, for example, it might compensate for small sizes).
Next steps for Cosine in the UK include studies looking at consumer magazines and weekend national newspapers (all co-funding is welcome).
In France, Italy, Germany and Spain…. we are proceeding this year with the full Cosine programme of work, combining all three components (again, for all co-funding, we’re happy to talk)… And indeed, Cosine has now become a World-wide project for JWT as part of the ‘Flight Deck’ programme of work that we at the Advanced Techniques Group is working on (reflecting the demands for consistency, from our global business partners).
So that’s it.
I’ve been through:
- Why we did the project
- How we did it
- The results
- How we are using it
- And what we’re going to next
So….. thanks very much for listening – I hope you found it interesting.