The four-day Worldwide Readership Research Symposium held in Valencia, Spain in October discussed many topics of great relevance to magazine publishers. The 2009 Symposium, which takes place every two years, attracted 150 delegates from 28 countries.

Several members of FIPP’s Research Committee participated in the event, and met during the final day to discuss the issues which had arisen which were of particular relevance to magazines. The main topics picked out were as follows.

Multi-media market modelling

Multi-media market modelling is probably the most persuasive form of evidence about the effectiveness of magazine advertising – but it is one of the most costly too. It involves analysing real campaigns, and measuring sales or other desired effects, media exposure, and other factors – ideally on a week by week or daily basis. Modelling can then estimate the contribution to effectiveness made by each medium employed. Magazine advertising can be expected to show up very well in the great majority of campaigns.

This kind of modelling is growing in importance, especially in the US where there are half a dozen commercial modelling companies as well as several media agency models.

It is vital that high quality data is fed into these models. For magazines, readership accumulation needs to be a vital element of this, so that the delivery of magazine exposures can be correctly attributed to each week, reflecting the time period over which an issue of a magazine accumulates its readers. Only in this way can the week by week link between exposure and sales be highlighted. The old working assumption that all exposures generated by an issue occur in the first on-sale week is not good enough. For example, a key reason why the UK study ‘Sales Uncovered’ produced such good results (magazine advertising increases sales by an average of 11.6 per cent) was that it was the first UK study to use readership accumulation data.

Online access panels as a data collection method

Several Symposium papers illustrated the many biases in online access panels used for omnibus surveys. Access panellists are opt-in volunteers, self-selected. The recruitment process is so indiscriminate that the true response rate is of the order of 1 per cent - a figure suggested by one major online company as generally applicable to all access panels.

Access panels are therefore not suitable for currencies such as national readership surveys, and their use should be confined to measures which complement the currencies, i.e. where representativeness is not so crucial.

A Canadian study compared four of the large commercial access panels, and an American study compared two access panels. Readership questions were run on each panel, and within both countries large differences were found between panels in readership levels, and in profiles of readers, even after weighting. The differences tended to be particularly large for specialist magazines. The conclusions were not only that none of the panels are representative of the population, but also that different panels have different biases – or rather, panels vary in which of the common biases are the more prominent.

Online panels recruited by random methods

Online panels specially recruited for a specific purpose and using random sampling methods are much better than access panels which more or less accept anybody willing to participate. Even so, a randomly sampled online sample will still contain some bias.

Design of online questionnaires

An illuminating series of controlled experiments reported in several papers showed how sensitive readership figures are to small details of on-screen layout, visuals and design, as well as question wording. It underlined the importance of very careful piloting.

Measuring ad impact

Growing emphasis is being placed on measuring the impact of individual magazine advertisements, especially in the USA, where Affinity Research offer its VISTA service and MRI has introduced its AdMeasure facility. Online panel members are sent copies of magazines, and in an online interview they are shown ads and asked about recognising the ads, brand favourability, actions taken, etc. The performance of a particular ad can be compared with norms.

In measuring the proportion of readers of the issue who saw a specific ad, this is bringing magazines more into line with TV, radio and online where the currency is ads seen, rather than merely vehicles seen. However an MRI paper showed that something like 80 per cent of ad recall in magazines is due to the creative work and product interest, and ad recall is not a sufficient measure of a magazine’s value to advertisers.

More relevant are the other measures such as brand favourability and actions taken.

Repeat reading

A big strength of magazines is that readers typically pick up a copy more than once and see many of the ads more than once. This repeat reading is considerable but is not usually taken into account in any statistical way. This should be forced back onto the industry’s agenda, and built into the media planning models.

Methods of measuring repeat reading already exist, such as APX in the US and PEX in the UK. There is some controversy around these measures and they are not universally accepted. Perhaps some modified form of measure can be devised.

Multi-platform media planning databases and systems

Another growing development is the creation of media planning systems which incorporate audience data for all the major media. TouchPoints in the UK is a prime example, and many media agencies have their own systems. The problem for magazines is that they do not include adequate data reflecting magazines’ qualities and effectiveness.

Publishers need to define what further information is required, and provide it. A measure of repeat reading would be a good start, along with qualitative factors (engagement et al).

Time-based media planning

A particular problem for magazines with the multi-platform planning systems is that they encourage a tendency to attribute too much importance to time spent with each medium. It is easy to produce a graph of media exposure through the day which shows high ratings for television and online while the line for magazines crawls along the baseline. This is damaging for magazines because it makes them look a very minor medium and it does not reflect the way magazine advertising works. Publishers need to consider how qualitative factors can more readily be brought into the equation.

A study for Condé Nast in New York was a reminder that magazines deliver more ads in a given time than TV or the internet. In a 30 minute period spent with each medium, users saw 65 magazine ads but only 28 online ads, and only 16 TV ads. This is not a complete answer, and it raises some other questions (e.g. clutter; relative attention levels), but it makes a useful point.

Symposium website

All 54 papers from the Valencia Symposium are available online from the Symposium website. In addition, the papers from all previous Symposia, back to the first event in 1981, are also available at the site.

Guy Consterdine