Wayne P. Eadie, Magazine Publishers of America
Dr. Jim Collins, Mediamark Research & Intelligence*
Worldwide Readership Research Symposium Valencia 2009 Session 1.10
Introduction:
In a world where digital media are gaining in importance and advertising accountability is increasingly essential, a key question is: how do other media work with digital media to improve the odds for advertising success? This paper will explore what we learned from available data about the value of using print magazines with websites, and specifically addresses the following:
Is it important to have print magazines alongside a digital buy?
Is there a difference between adding a print magazine to a non-magazine-branded website versus a magazine-branded website?
What insights can be learned by looking at product usage that is endemic to the magazine genre?
Executive Summary:
Given that advertisers want to raise the odds for advertising success, better targeting those most likely to buy can increase the likelihood that those consumers exposed to advertising will act.
Comparing product incidence/usage based on composition for consumers who had visited a website in each of three specific categories and not read a print magazine in the same category relative to those who had read a magazine in the same category, we found the following:
- Overall, in each of the three categories examined, the addition of magazine reading increased the incidence of product usage.
- In every situation analyzed for all websites and non-magazine websites, a significant lift was achieved when reading a print magazine was introduced.
- For magazine branded websites, increases were seen, although to a lesser degree.
- When looking only at endemic products, services and attributes pertinent to the given categories, the relationships were even stronger and held up equally for both magazine-branded and non-magazine websites in the two categories looked at. (Celebrity did not have enough endemic cases for examination).
Methodology Overview:
Data Overview and Fusion Methodology:
In order to answer these questions, we employed the MRI and Nielsen Online’s Net//MRI fused database. The selection of this database was largely driven by four factors:
-
- It contains ratings quality magazine audience measures from MRI’s National Study
- Nielsen Online’s NetView is a random-digit-dial sample-based, meter-measured internet site audience measurement standard comprising approximately 25,000 Home and Work internet users.
- NetView includes Home and Work audiences for thousands of sites and is revised on a monthly basis
- MRI includes thousands of consumer behavioral and attitudinal measures
In short, Net//MRI includes extensive ratings-quality measures of the two media of interest in this analysis (magazines and internet) along with extensive consumer behavior and attitude ones.
*Elizabeth Fisher and Lancey Heyman of Mediamark Research & Intelligence were instrumental in doing the extensive data runs that provided the foundation for this paper.
Among potential alternatives several were considered, but each was substantially compromised with respect to the ambitions of the work at hand. For example, MRI’s National Study alone includes measures of internet behaviors, but they are relatively limited when considered against the number measured in NetView. Similarly, Nielsen Online’s @Plan product measures magazines along with internet sites, but 1) the number of titles is relatively limited and 2) it is a qualitative/directional rather than a ratings-quality metric. Moreover, @Plan’s internet visitor estimates are themselves conformed to NetView’s.
While the Net//MRI database is the best available source of data for this analysis, the question remains, “Does the use of fusion to integrate MRI and NetView bias the results of this analysis?”
In developing an answer to this fundamental question, the mechanics of the fusion and especially the repertoire of data employed offer reassuring insights. In brief, fusion as a matching process attempts to match records from multiple databases (in the case at hand, MRI respondents with NetView panelists) on measures common to both data sources. In the Net//MRI fusion, the extent of the common measures large (over 200) and multi-dimensional (demographics, generic internet behaviors and site-specific measures). Moreover, while the Net//MRI fusion matching process cannot guarantee that all MRI respondents and NetView panelists are matched perfectly one with the other, the fusion technique incorporates a dynamic model-based matching scheme wherein a matching hierarchy is developed based on the relationship between the common demographics and internet behaviors on the one hand (independent variables in the model) and total amount of time spent on the internet – a broad metric of internet behavior. Thus, while the Net//MRI database is not a single source measurement, the relationships which are analyzed in the course of this analysis emerge in the database as a result of very strong and extensive linkage.
Analysis Procedures:
The intent of this analysis of available data was to take a first step directional approach to the value of adding a print magazine to a website in general, and more specifically, to non-magazine and magazine-branded websites. A more in-depth evaluation employing more advanced statistical measures may be appropriate to further extend these initial findings.
Three distinct consumer interest categories were chosen to explore. They were selected to represent sizable categories for magazine advertising where sufficient web presence was available both for magazine-branded and non magazine-branded sites. In addition, both print and web properties needed to be present in the MRI and Nielsen Online database.
Furthermore, the ad categories represented across these magazine genres needed to represent male and female targets and be differentiated sufficiently to allow for the identification of differing results to emerge. Once selected, broad demographics were applied as a filter to focus on likely buyers and help highlight product usage. The analysis centered on using composition to determine the presence of active category users within each of the three data sets.
Categories of Interest Target
Food/Cooking Women 25-54
Celebrity Women 18-44
Sports Men 18-44
Since all data included in the MRI survey were fused with the Nielsen Online database, the MRI code book was accessed to choose the best selection of gateway product usage (users of a product category, e.g., consumed domestic wine in past six months, consumed imported wine in past six months) and attitudinal questions that would likely be stable and suitable to provide the needed insights for each category. Many were chosen to be endemic to the category of interest; others were selected to represent the advertising categories most likely to advertise in these magazine genres. The choice followed set procedures and was done without regard to which products would likely score well. The goal was to best represent the three categories with pertinent product usage and attitudinal items. A summary list of products included for each of the three categories of interest can be found in the appendix.
Product usage data were eliminated for any products where the majority of the data accessed were unstable. In the cases where some of the data were stable and some not, it was the practice to use only data where it was stable for both cells being compared.
For each of the three categories, a three level analysis was conducted as shown in the table below:
- Addition of …Magazine Reading to Visiting Any …Website (Food, Celebrity, Sports)
- Addition of … Magazine Reading to Visiting Magazine Branded …Website (Food, Celebrity, Sports)
- Addition of…Magazine Reading to Visiting…Website that is not a Magazine branded Website (Food, Celebrity, Sports)
For the analysis stage it was determined that at least a ten percentage point differential was necessary to show differentiation between cells being analyzed.
Visiting a website was defined as within the past thirty days (30 day net home/work). A six month average issue audience was used for magazines.
Specific Findings:
Women 25-54: Visiting Food & Cooking Websites and Read Food Magazine (Ref. Exhibit 1)
The addition of reading a food magazine to visiting any food and cooking website boosted the incidence of product usage for each of the three website scenarios: overall, magazines branded and non-magazine branded.
Two-thirds of the cases (104 of 166 relevant product usage items looked at) showed a difference at the level of at least ten percentage points in index or greater. 94% of these cases showed increases in incidence of product usage with the addition of magazine readership.
The addition of reading a food magazine to visiting a food magazine website showed much the same lift as evidenced over any food or cooking website. In this scenario in the two-thirds of the cases where a difference was seen, 88% of those (all but eight cases) favored the addition of magazine reading.
When looking at those who accessed food and cooking websites but not magazine branded food and cooking websites, much the same holds true. Of the 103 cases where there was a difference, 94% showed a lift in incidence of product usage when reading a food or cooking magazine was introduced.
Exhibit 1.
Women 25-54 Food/Cooking
Base: 166 Cases
2
| Cases d Product Usage Attribute ltems | Addition of Reading Food/Cooking Magazine Тo isiting Any Food/Cooking Website | Addition of Reading Food/Cooking Magazine Тo isiting Food/Cooking Magazine Website | Addition of Reading Food/Cooking Magazine Тo isiting Food/Cooking Website Тhat is Not a Magazine Website |
| Cases with Sufficient Sample Size for Analysis | 166 | 103 | 165 |
| Cases Where Sufficient Difference Shown | 111 | 68 | 103 |
| % of Тotal Cases | 67% | 66% | 62% |
| Cases Where Addition of Magazine Reading Boosts lncidence of Product Usage | 104 | 60 | 97 |
| % of Cases with Sufficient Difference | 94% | 88% | 94% |
Women 18-44: Visiting Celebrity Websites and Celeb Magazine Reading (Ref. Exhibit 2)
The addition of reading a celebrity magazine to visiting any celebrity website substantially boosted the incidence of product usage in the cases where a difference was shown.
For 186 relevant product usage items looked at, in the 42% of cases (78 of 186) where at least a ten percentage point difference in product usage index was shown, on average, reading a celebrity magazine increased the incidence of product usage roughly two-thirds of the time.
The addition of reading a celebrity magazine to visiting a celebrity magazine website for this category did not show the same lift as evidenced over “any celebrity website”. In this specific case, roughly half the cases showed no difference, and for the other
half the differences were evenly distributed when looking at reading or not reading a celebrity magazine against visiting a celebrity magazine website. A similar pattern to all other findings was seen for men 18-44 accessing sports magazine websites and is explored in the “Observations” section..
For those who accessed celebrity websites excluding magazine branded celebrity websites, much the same relationship was found as in overall celebrity websites. Of the 76 cases where there was a difference at ten percentage points, 70% showed a lift when reading a celebrity magazine was introduced.
Exhibit 2.
Women 18-44 Celebrity
Base: 186 Cases
1
| Cases d Product Usage Attribute ltems | Addition of Reading Celebrity Magazine Тo
isiting Any Celebrity Website |
Addition of Reading Celebrity Magazine Тo isiting Celebrity Magazine Website | Addition of Reading Celebrity Magazine Тo
isiting Celebrity Website Тhat is Not a Magazine Website |
| Cases with Sufficient Sample Size for Analysis | 186 | 150 | 183 |
| Cases Where Sufficient Difference Shown | 78 | 68 | 76 |
| % of Тotal Cases | 42% | 45% | 42% |
| Cases Where Addition of Magazine Reading Boosts lncidence of Product Usage | 51 | 33 | 53 |
| % of Cases with Sufficient Difference | 65% | 49% | 70% |
Men 18-44 Visiting Sports Websites and Sports Magazine Reading (Ref. Exhibit 3)
The addition of reading a sports magazine to visiting any sports website boosted the incidence of product usage in more than half the cases where a difference was shown.
More than half the cases (89 of 184 relevant product usage items looked at) showed a difference at the level of at least ten percentage points in index or greater. Of the cases showing a difference, 61% (54 of 89) favored sports magazine reading being introduced.
The addition of reading a sports magazine to visiting a sports magazine website showed that the increases in product usage were equal for the cells where magazine reading were introduced versus where it is not. Commentary on why this might be is included elsewhere in the paper under “Observations.”
Again, when looking at those who accessed sports websites but not sports magazine websites, we find that print magazines, in the 99 cases where there was a difference, 64% (63 of 99) showed increases in product usage indices when reading a sports magazine was introduced.
| Cases d Product Usage Attribute ltems | Addition of Reading Sports Magazine Тo
isiting Any Sports Website |
Addition of Reading Sports Magazine Тo
isiting Sports Magazine Website |
Addition of Reading Sports Magazine Тo
isiting Sports Website Тhat is Not a Magazine Website |
| Cases with Sufficient Sample Size for Analysis | 184 | 170 | 166 |
| Cases Where Sufficient Difference Shown | 89 | 96 | 99 |
| % of Тotal Cases | 48% | 56% | 60% |
| Cases Where Addition of Magazine Reading Boosts lncidence of Product Usage | 54 | 48 | 63 |
| % of Cases with Sufficient Difference | 61% | 50% | 64% |
Analysis of only those products/attributes endemic to the categories of interest:
Exhibit 3.
Men 18-44 Sports
Base: 184 Cases
3
To test further whether targeting based on product usage/incidence is a valid proposition; additional analysis of only those products endemic to the category for two of the categories, food/cooking and sports (celebrity was deemed not to have enough cases of endemic products for meaningful examination) was conducted. We wanted to see whether the findings held for endemic products to the given category to see if the same relationships existed for those products most connected with the category, and relationships not only held up equally for both magazine-branded and non-magazine websites in the given categories, but they were even stronger with virtually all scenarios showing a lift.
Exhibit 4.
Women 25-54 Food/Cooking – Products Only
Base: 105 Cases
1
Exhibit 5.
Men 18-44 Sports Related – Products Only
Base: 27 Cases
5
| Cases d Product Usage Attribute ltems | Addition of Reading Food/Cooking Magazine Тo isiting Any Food/Cooking Website | Addition of Reading Food/Cooking Magazine Тo isiting Food/Cooking Magazine Website | Addition of Reading Food/Cooking Magazine Тo isiting Food/Cooking Website Тhat is Not a Magazine Website |
| Cases with Sufficient Sample Size for Analysis | 105 | 71 | 105 |
| Cases Where Sufficient Difference Shown | 69 | 45 | 65 |
| % of Тotal Cases | 66% | 63% | 62% |
| Cases Where Addition of Magazine Reading Boosts lncidence of Product Usage | 65 | 42 | 61 |
| % of Cases with Sufficient Difference | 94% | 93% | 94% |
| Cases d Product Usage Attribute ltems | Addition of Reading Sports Magazine Тo
isiting Any Sports Website |
Addition of Reading Sports Magazine Тo
isiting Sports Magazine Website |
Addition of Reading Sports Magazine Тo
isiting Sports Website Тhat is Not a Magazine Website |
| Cases with Sufficient Sample Size for Analysis | 27 | 27 | 27 |
| Cases Where Sufficient Difference Shown | 22 | 19 | 14 |
| % of Тotal Cases | 81% | 70% | 52% |
| Cases Where Addition of Magazine Reading Boosts lncidence of Product Usage | 22 | 18 | 14 |
| % of Cases with Sufficient Difference | 100% | 95% | 100% |
Observations and Further Considerations:
For two of these categories, Celebrity and Sports, the relationship between magazine reading and visiting magazine branded internet sites was not stronger than for visiting non-magazine branded internet ones. In contrast, for the Food category this relationship was stronger. Why might this be happening?
While necessarily conjectural, an answer may lie in the distinctive nature of the content among these three subject genre. Celebrity and sports may be thought of as being more headline-driven and food could be viewed as service-oriented which may serve to impact the way readers are drawn to the websites.
Duplication may be greater between magazines and their respective magazine websites than between the magazine and a non- magazine-branded website for a given category. If so, duplication may ameliorate the lift in product usage indices that would otherwise be expected. Large, overlapping audiences and long tails may be at work here. Respective website universes may also play a role.
Separately, the current analysis was limited with respect to sites included.
First, E-Commerce, in contrast to advertising-supported sites are generally excluded from the standard NetView database, as its primary purpose is providing site metrics in support of planning and transacting internet advertising. The inclusion of e- commerce sites in subsequent analyses may well offer additional insights into the dynamics of magazine print and internet behaviors.
Secondly, while this analysis included the major advertising supported sites in each category, the finite size of the NetView RDD (random-digit-dial) panel (approximately 25,000 Adults Age 18+) excludes measurement of lower incidence sites. RDD does a good job of accounting for the larger sites which account for much of the category traffic. For the three site categories included in this analysis, Sports was comprised of 101 sites, Multi-Category Entertainment of 153 and Food & Cooking of 44. By way of contrast the NetView MegaPanel database (not available at the time of this analysis, but with approximately 140,000 panelists) includes 1905 Sports sites, 1660 Multi-Channel Entertainment and 537 Food & Cooking ones.
Were e-commerce and smaller sites included in this analysis some of the print/internet relationships might be (slightly) different.
Finally, the MRI consumer measures used in the evaluation do not account for volume of usage, simply whether or not the product was consumed. However, as volume of consumption is a primary measure of category/product involvement it might be expected that consumption volume would be positively correlated with related media consumption.
Conclusion:
For the most part this is about increasing the odds for success by targeting the greatest concentration of people likely to buy or use your product or service. It provides an opportunity to move beyond demographics alone and take advantage of the synergies provided in using magazines and websites, whether magazine-branded or other sites sharing a category focus.
The analysis showed that the addition of magazine reading significantly increases the incidence of product usage among readers also accessing web sites in the given genre. The print magazine clearly plays a strong role when used in combination with the web.
References:
McDonald, Scott and Collins, Dr. James (2007) “Internet Site Measurement Developments and Print.” Worldwide Readership Research Symposium XIII, Vienna. Session papers.
McPheters, Rebecca and McDonald, Scott (2009) “Opportunities to Influence: Increasing Ad Impact in a Multi-Platform World.” Expanded Paper, June 2009.
| Appendix: | ||
| Magazine titles included: | ||
| Celebrity | Food | Sports |
| In Touch Weekly | Bon Appetit | ESPN The Magazine |
| Life & Style Weekly | Cooking Light | Golf |
| OK! Weekly | Everyday Food | Golf Digest |
| People | Everyday with Rachael Ray | Ski |
| TV Guide | Food & Wine | Skiing |
| Us Weekly | Gourmet | Sports Illustrated Tennis
The Sporting News |
| Magazine Websites Included: | ||
| Celebrity | Food | Sports |
| People.com | Gourmet.com | ESPN.com |
| TVguide.com | MarthaStewart.com | SportingNews.com |
| TVguideMagazine.com
UsMagazine.com |
MarthaStewartnetwork.com | SI Digital |
Categories of MRI product usage items and counts for each that were included in the analyses: Advertising Category Food Celebrity Sports
| Apparel/shoes | 4 | 6 | 4 |
| Automotive | 0 | 21 | 39 |
| Books, music, movies | 3 | 24 | 9 |
| Buying styles | 13 | 3 | 0 |
| Children’s toys & games | 1 | 2 | 0 |
| Electronics (cameras, computers, phones, tv) | 0 | 33 | 53 |
| Financial, insurance, real estate | 8 | 7 | 8 |
| Food, beverage, dining out | 36 | 7 | 14 |
| Food preparation, cookware, table settings | 46 | 0 | 0 |
| Gardening | 5 | 0 | 0 |
| Health, personal care products, beauty aids | 29 | 51 | 9 |
| Home services/improvement/furnishings | 7 | 8 | 2 |
| Jewelry, watches, eyewear | 0 | 5 | 2 |
| Leisure activities, public service | 9 | 0 | 3 |
| Pets | 2 | 1 | 0 |
| Sports, fitness | 0 | 0 | 23 |
| Travel, theme parks | 3 | 18 | 15 |
| Tobacco | 0 | 0 | 3 |
| Total items analyzed | 166 | 186 | 184 |