Press advertising - Does colour make a difference?

Drawing on his important ColourWorks study for the Newspaper society, Jim Chisholm looks at the effect of colour in newspaper advertising. Colour can greatly improve ad effectiveness, but the effect is not guaranteed. Other factors, especially the size of the ad, can be just as important

Jim Chisholm

DESPITE the rapid growth in the use of colour in newspaper advertising, we know very little about the effect of colour in advertising. How much, for instance, does colour add to the effectiveness of an ad? Why do some ads work better than others? And what elements within an ad make it more effective, and the colour work harder?

To find answers to some of these questions, the Newspaper Society asked The Chisholm Business to look into the relative strengths of colour advertisements. The aim was to go beyond simple measurement of relative benefit to find out what makes some advertisements work better than others.

Two locations, Northampton and Bristol, were chosen to provide different mixes of population, readership, product and colour experience.

In each location 24 advertisements were chosen by the local publishers. The advertisements were invariably the most easily available and represented a good cross-section of local colour advertising. After printing it was felt that some ads were unusable, because they contained mistakes or were not printed correctly, so that, of the original 48 advertisements, 36 were used in the research.

Each publisher then produced six, 16-page dummy newspapers, with an advertisement on every page. The content of these newspapers was taken from two recent issues of the parent paper. The resulting 96 pages were sufficient to ensure that each advertisement could be permutated around early and late, content, left and right, colour and mono to produce random results.

The research had two stages. The first was an eye-movement study, conducted by Dundee University. The second stage was a series of hall tests, undertaken by NOP.

The aim of the eye-movement study was to see how people read the dummy newspapers. Seventy respondents were recorded on location in Bristol, noting the order in which they read things, and how long they spent on different elements of the page or the advertisement. The research examined how people read the pages of the newspapers, to find out how advertisements are read in the context of the total newspaper.

A second eye-movement study looked specifically at the reading of advertisements themselves. This was completed at the Micro Centre in Dundee and provided data on the specific reading behavioural differences between advertisements. Twenty-four respondents were recorded. The results from this test show the intensity of reading in terms of reading duration (the time spent looking at each element) and fixations (the number of times the person looked at that element), as well as measurements of the order of reading.

For the second stage of the research, NOP carried out 600 hall tests in each location. Each respondent was asked to read one of the dummy newspapers. They were instructed to 'read the newspaper in any way you want, as you would usually do. We would like you to read both the stories and the advertising'. They were then told: 'At the end of this time we will ask you some questions about what you have read or noticed. Do not try to memorise anything. We are simply interested in what your views are about what you have read.'

The respondents read their dummy newspapers for an average of 8.1 minutes in Bristol and 6.4 minutes in Northampton. While this is considerably less time than they would be expected to spend reading a 'real' copy of the paper, the dummies were obviously smaller, and average reading time per page was comparable.

Respondents were then asked a series of questions. First, without prompting:

  • 'What did you read or notice in the newspaper? What else?'
  • 'Can I ask you which advertisements you noticed?'
  • 'Who were the advertisements for? Who was the advertiser?'
Second, they were asked the following prompted questions:
  • 'Here is a list of things that have been advertised. Which of these can you recall seeing?' (Shown list of items that were advertised in all 24 advertisements).
  • 'Now which of these ads do you recall seeing?' (Shown scrapbook containing all 24 advertisements).
From this five measurements of recall were established:
  • Unprompted content.
  • Unprompted advertiser.
  • Unprompted combined (merging the unprompted results).
  • Prompted content.
  • Prompted advertiser.
These results showed levels of recall. By including the eight unseen advertisements it was possible to measure 'background effect' - that is the level of claimed recall of advertisements that respondents had not seen but thought they had.


Typical examples of page reading behaviour are illustrated in Exhibit 1, with the percentage of time spent on each element shown. In the page on the left, 2.7 per cent of time spent reading the page is devoted to the advertisement (the darker box). On the page on the right it is none.


In Bristol five dummy newspapers were subjected to the eye-movement test. The overall results suggest that readers are capable of absorbing vast amounts of advertising stimulus, even though they spend little time on the advertisement - see Exhibit 2.


Dummy version% of content that is adsReading time per page (sec)% of fixation on ads% time spent on the ads

On average respondents spent 38 seconds reading each page of the dummy (this equates fairly well with a typical 64-page newspaper that is read for an average of around 34 minutes). Only 2.2 per cent of fixations dwell on advertisements, and readers spend less than one per cent of their time reading the advertisements. However, this is long enough for them to register the advertisement and to absorb its message. Readers of course spend longer on the advertisement if it proves to be interesting.

The eye-movement study revealed that many respondents could recall advertisements even though there was no evidence of them actually fixating it. This suggests that some advertising information is picked up by the eye's peripheral vision and that this is enough to achieve recall, or alternatively convince the reader that the material is not worth spending time on.

It is clear that readers initially receive an impression of the entire page using both primary and peripheral vision. Having done this they identify four types of object as a map: headlines, body text, pictures, and advertisements - including obvious logos or content.

The reader forms this map in an instant and uses it to select items of interest, ignoring those that are easily dismissed as irrelevant. The role of the advertisement therefore is to ensure that the reader is attracted to it. From there the advertisement must convert that initial interest into detailed interest. Many of the advertisements in the study failed at this initial interest stage.


The eye-movement study revealed that three elements of the advertisement - the picture, the headline and the copy - account for 86 per cent of the reading traffic of the advertisement.

The most commonly looked at part of the advertisement is the main picture which accounts for 27 per cent of the reading time. The headline and main piece of information or copy each accounted for 17 per cent of reading time. A further 22 per cent was spent on any second and/or third piece of information or copy. Logos accounted for five per cent and addresses or maps two per cent.

It is clear, therefore, that the attention of the reader is only attracted to the advertisement if the message is compelling and easily absorbed at an initial glance. The presentation of the advertisement must be capable of attracting attention even though the reader may not specifically look at it. Having attracted the reader the advertising must hold attention through good use of illustration and copy.


Exhibit 3 shows how levels of recall rise as prompting increases. One can see how the gap between colour and mono increases through the levels. It also shows the level of background effect that occurs with respondents claiming, under prompting, to recall advertisements that they had not actually seen. Background effect is higher at the content levels since readers expect to see key categories such as motors or property or food retailing.



Two of these levels were chosen as key indicators:

  • Combined unprompted recall: The ability of the advertisement to be recalled in some form or another without prompting. This was felt to be a tough, but fair, measure of effectiveness.
  • Prompted advertiser recall: The ability of the advertisement to encourage recognition at a subsequent event. This was felt to be the easiest level of recall and should therefore generate the highest result. Where unstated, the results refer to the combined unprompted measurement.
These results show the average levels of recall in each location. However, this method of calculation disproportionately favours larger advertisements with higher levels of recall. Having determined the different results achieved at each level of questioning, it is best to present the benefits of colour as an index against the mono levels. As Exhibit 4 shows, colour increases unprompted recall of an advertisement by an average 31 per cent and prompted recall by an average of 39 per cent. It has the most dramatic effect on unprompted recall of an advertiser's name where it improves levels by an average of 53 per cent.


INDEX (B/W=100)


For every advertisement that respondents claimed to have seen they were then asked: 'I am going to read out some statements that people have made about advertising. I would like you to tell me whether you think they apply or not to each of the advertisements that you mentioned seeing'.

We called these results imagability, referring to the ability of the advertisement to convey a range of different images about the product or company being advertised.

Respondents were also asked about their purchasing intentions for a range of goods and services and some general questions about their attitudes to advertising.

The results are shown in Exhibit 5. Colour is shown to have had the strongest effect on improving the design image of the advertisement content. Agreement with the statement 'I like the design of this advertisement' increased on average by 16 per cent. The second strongest effect was on the statement 'It makes me want to buy the product', where agreement with the statement increased by ten per cent.


index (b/w=100)
'I like the design of this advertisement'116
'It makes me want to buy the product'110
'I always read advertisements like this'109
'This looks like a popular product or service'108
'Appeals to people like me'107
'I always read advertisements like this'104
'This is a good quality product'103


It is possible to create a model from the research results showing average response levels for colour and mono advertisements of different sizes. Exhibit 6 shows the level of recall for both colour and mono advertisements according to size. The line, set against the right-hand axis, shows the percentage increase in recall gained from colour.

Various conclusions can be drawn from this:

  • Colour adds little to the effect of very small advertisements.
  • In advertisements under 100 cm the increase in recall is not proportionate to the increase in advertisement size. After this point, doubling in advertisement size more or less results in a doubling of recall.
  • The percentage benefit peaks at around 100 cm where colour improves recall by 43 per cent. However, after this point, the level of benefit of colour is constant rather than proportionate. In other words, after a certain size there is no additional benefit to be gained.
This has two major implications for publishers and advertisers. Advertisers may be advised to take more frequent smaller advertisements. Publishers may wish to revise their rating policy from percentage premiums to fixed prices since the benefit is more uniformly enjoyed, particularly across the middle ad sizes. Publishers may also consider introducing premiums for solus colour or discounts for shared colour. This also recognises that the effort and resources required to print in colour are the same regardless of ad size.

Findings suggested that whilst average results in Northampton were generally higher than those in Bristol, colour was of less average benefit. This can be simply explained by the difference in the average ad size in the two locations: in Bristol the average was 74 cm and in Northampton 183 cm.

The model described can be used to predict the expected results for the two locations given the average advertisement size. If the results in Bristol and Northampton can be proved to be similar - despite the differences in their markets, products and colour experience - the results can reasonably be applied to other locations.

Exhibit 7 compares average recall for the two centres using both actual and modelled data. These show how the results vary between the two locations but also that this is reasonably predictable when the model is applied to the average ad size in each location.


Colour %B/W %Colour %B/W %
O=Observed; M=Modelled; R=0.924


When it comes to the ability of an advertisement to improve image, there is some relationship between size and the level of improved image. However, the results when comparing colour and mono proved inconclusive. It is clear that other factors, such as advertisement content, design, message and size were far more critical than colour.


Having now established the ground rules for measurement and demonstrated the average results for newspaper advertising, we can develop some conclusions from individual advertisements. The analysis of each advertisement is covered in the full report.

Exhibit 8 compares levels of unprompted recall and the benefit of colour on that level for each advertisement in the study. The horizontal axis shows the level of recall, with C&A enjoying the highest results in the trial. The vertical axis shows the percentage by which colour has increased or decreased that result. The Jersey, Venus and Reproduction ads, though showing lower levels of recall, benefited most from colour, while levels of recall for Apple Country, Greener County and Helix, decline in colour.

The chart suggests five groups of advertisement:

  • High recall, colour no advantage. Ads which show very high levels of recall, where colour would appear to make little difference on the levels of recall. These ads are, without exception, household names.
  • High recall, high colour benefit. Ads which show high levels of recall and benefit a great deal from colour. These advertisements were characterised by strong images, with good use of illustration, simple design, and clear benefits.
  • High recall, low colour benefit. Ads which show high levels of recall but do not particularly benefit from colour. These tend to be utility advertising such as traditional motoring or other information ads.
  • Low recall, high colour benefit. Ads which show low levels of recall but benefit from colour. These ads again benefited from simplicity with the colour enhancing the message.
  • Low recall, low colour benefit. Ads with low recall where colour offers little improvement. These tended to suffer from overcrowding with too much material, where the colour detracted from the message.
These five groups are summarised in Exhibit 9.


High recall no benefitHigh recall high benefitHigh recall low benefitLow recall high benefitLow recall low benefit
Ad size (cms)259.0127.8142.832.866.2
Unprompted content20.
Unprompted advertiser24.
Combined unprompted25.
Prompted content21.827.830.67.813.0
Prompted advertiser50.742.635.114.819.2
Want to buy37.022.235.433.323.7
Like design70.470.770.764.761.6
Looks popular80.762.678.557.663.0
Good Quality71.762.068.563.460.8
Welcome ads69.660.166.259.860.1
Always read51.341.247.148.436.7
TOTAL IMAGE63.461.759.857.850.8
Activity (% change)
Readin time+5.5+9.0+1.1+15.6-8.5


Inevitably this research created more questions than it answered. Advertising effectiveness is determined by the strength of the medium and the strength of the advertisement. This study shows that while colour can greatly improve the effectiveness of an advertisement, it is not guaranteed.

Many of the advertisements in the research, including some from major national companies, did not achieve either the levels of recall or the benefit from colour that the advertisers should have anticipated. Several advertisements failed because they did not follow the most basic design rules.

Traditional recall research and reading behaviour studies in themselves do not necessarily offer a reliable measurement of an advertisement's potential. However, the combination of techniques could be used, not only to measure the potential effectiveness of an advertisement, but also to measure the differences in effectiveness across different media.

Colour Works is available from the Newspaper Society, tel: 0171 636 7014.


Jim Chisholm

Jim Chisholm

Jim Chisholm is a director of The Chisholm Business Ltd., which offers a specialist strategic planning and marketing service to the newspaper industry. He specialises in business development as well as strategic and marketing issues, and in addition to 17 years experience in the uk national and regional newspaper industry has conducted a number of projects in Europe and Africa.
Press advertising - Does colour make a difference?

Exhibib 1. Examples of reading behaviour

Press advertising - Does colour make a difference?

Exhibib 6. The effect of colour and advertisement size

Press advertising - Does colour make a difference?

Exhibib 8. Strength of recall and impact of colour on recall