Marketing Communications

POST-TESTING OF ADVERTISING 303

Recall and recognition tests are useful and easy to carry out, but they have a number of

severe limitations ( Figure 9.7 ). In recognition tests, consumers can say what they like, they

can lie, exaggerate or guess. Th is makes the results of recognition tests far from reliable. An

ad that is part of a campaign of similar ads will be more easily recognised or recalled. In these

circumstances, it is very hard to isolate the eff ect of one single ad. Th e underlying assumption

that recognition or recall of the ad eventually leads to buying the product may be erroneous.

Recognising or recalling marketing communications may be a necessary condition to buy a

product, but certainly not a suffi cient one. Product involvement infl uences the results of the

test. A consumer who is very interested in a certain product category will perform better

in recognising ads for this product category. Th is does not imply anything about the eff ec-

tiveness of that specifi c ad. Recall scores are very dependent on the time elapsed between

Figure 9.7 Limitations of post-testing

Every time a visitor clicks on a website link or interacts with a website, the web server automatically keeps track of

some of the data in so-called log files. Most servers generate four types of log files. The access log file keeps track of

the file names that are downloaded, the IP address of the client server that requested the file and an indication

of time. The agent log file registers which browser is being used by which visitor’s server for every visit to the site. The

referrer log file registers the pages that were visited before the visitor landed on the web page. The error log file keeps

track of errors that occurred during the process. The log files give information such as the number of ‘hits’ (downloaded

files), the point in time when files were downloaded, the number of bytes transferred, the domains from which the visitor

arrived (Internet service providers, proxy servers, firewalls) and the browsers used. A number of software instruments,

such as Nedstat and WebTrends, have been developed to efficiently analyse and summarise this enormous data

stream. Also data-mining software is used to find meaningful patterns in the large amount of data on website visit-

ing behaviour. Log files have an important limitation. They cannot identify individual users. Therefore, log file data are

often supplemented with cookies , a piece of software that is planted in the system of the website visitor and allows

web servers to register and keep track of the revisit behaviour of visitors. The visiting of ads on websites (banners,

buttons) can be assessed by means of click-through rates. Most click-throughs result in a transfer of the visitor to

the advertiser’s website. A click-through is the virtual counterpart of a response of an individual to a direct mailing.^14

RESEARCH INSIGHT

Measuring website visiting behaviour

M09_PELS3221_05_SE_C09.indd 303M09_PELS3221_05_SE_C09.indd 303 6/6/13 9:21 AM6/6/13 9:21 AM