Real-Time Campaign Evaluation

Arie den Boon

Now that television seems likely to become less effective – and the same holds for almost all other media – what can marketing managers do to get the best out of their marketing budgets? They plunge into the media synergy field and use more media in their campaigns. But what they get is not clear until the campaign is over: a good or bad surprise. Their task is difficult and there is almost no research that can help them to take tough decisions. Media buying is based on audience research and past experience, but real effect data is missing. Therefore the media plan is always suboptimal, even with the best resources.

It is during the campaign itself that real effects are achieved. With real-time campaign evaluation the marketing manager can monitor, and get fast feedback for, the effectiveness of the campaign. But monitoring in itself is not enough. To be able to take action, a real-time analysis of the performance of the campaign, media, creative and media strategy is necessary. Only with real-time analyses can the marketer optimise and manage communication. Monitoring is looking back, analysing is gaining insight, managing is optimising campaign performance.

This article addresses media strategy and frequency.

MIXED MEDIA CAMPAIGN

To illustrate, we look at a mixed-media campaign for a brand of personal care products (Brand Q). The target group consists of males and females, aged 20–49, who are responsible for the shopping in the household. We look at both advertising awareness and brand consideration as objectives (that is, effects we wish the campaign to have).

The campaign is ongoing and we select a period of 13 weeks to analyse the advertising effects. Online research is carried out among 1,415 people. The questionnaires contain both advertising effect questions and questions that estimate the media usage.

We look first at the trend lines of both prompted advertising awareness and brand consideration. The ad is shown to respondents as a series of nine stills and the prompted advertising awareness is the percentage of consumers who say they certainly or probably have seen the ad.

The consideration variable is the percentage of consumers who indicate they know the product, buy it or are willing to buy it. Figure 1 shows the trend lines of both variables. For some 60%–65% the brand belongs to their consideration set, and the prompted ad awareness is 50%–60%.

The bars show the euro costs of the TV and magazine pressure. The magazines used include both weeklies and monthlies.

Although we have taken a three-week rolling average, we have to admit that nothing much happens to the awareness and consideration trend lines. It would be extremely difficult to assess the importance of the two media in the campaign just from this chart. The advertising recognition of print ads and all TV commercials shows strong rising trends during the campaign. But in the competitive environment of this category of products, advertising awareness and brand consideration do not show clear effects at this aggregate level.

Let us look at the first step in communication: do the ads get attention and are they being recognised? In Figure 2, we see the recognition of a TV commercial and three magazine ads. All show effect increases, some more than others. So the first step in communication is achieved: the ads in both media are recognised.

To find out what effects the different media may be having, we need to look at more detail. For that we carry out a Media Pressure Analysis, in essence a detailed contact-frequency analysis.

FREQUENCY OR REACH

We have two media in the campaign, so we will look at the effects of the number of contacts the respondents have with the different media. For each medium, we divide the respondents into four groups. For TV, we put in the 'No contact' group all respondents that cannot have had any contact with the campaign because they have not watched the relevant channels in the dayparts within which the campaign has run. The rest of the respondents are divided into three groups (Low, Medium, High) with equal numbers of respondents in each. We do the same for print, according to what the respondents have told us about their readership of the magazines in which the campaign appears. The recognition of both the print ads and the commercials shows an almost linear increase from No contact, through Low and Medium to High numbers of contacts. An example for magazine ad 1 is given in Figure 3. The budget also increases rapidly in each of the groups. So more contacts and more budget leads to a higher recognition, but the budget in the three contact groups increases faster than the recognition percentages (see Figure 4).

Because the effects of ads usually erode over time unless reinforced, we have decided to apply, in this case, an assumed decay rate of 25% per week (other levels can also be used, as well as no decay at all). Now we will analyse the prompted ad awareness of Brand Q.

We see from Figure 4 that TV shows a much steeper curve than magazines. So, more contacts with TV increase the effect more quickly. We can see other things as well. The increase levels out at the Medium level. More contacts do not show more effect – it stays at 62%. This is remarkable, because the budget doubles in the highest category. The groups with a high magazine frequency even show a decreasing effect. To explain what happens here is not difficult. The curve is likely to be influenced by relevant differences between the groups. For example, there are differences in social class (see Figure 5). The High frequency group contains only 13% of the top social class A, whereas it is up to 25% in the Low frequency group (the position being reversed for the lowest social classes, C and D). Figure 5 shows how high social class differs considerably from the other groups.

There is no difference in frequency response by sex, although men show much lower awareness percentages than women. So, one of the reasons that the high contact frequency does not increase linearly with the number of contacts is that high and low contact groups contain different people.

There can also be other reasons why the effects do not increase with the contact frequency. It is easier to add percentage points to a low starting level than when the percentage is already 60% or more. We will probably never get awareness higher than 80% or 90%. A second reason is wear-out of the ads, although this is not likely for awareness, and more likely for attitudes and behaviour.

So, for many reasons, a high frequency does not always give us more effect. How does this help us here? First, in this example, some of the budget in the high frequency groups could profitably be reallocated to the lower frequency groups. TV is a mass medium, so a high frequency cannot be avoided completely in a campaign with a high TV pressure. But we could gain efficiency by shifting some weight to groups that have more room to grow. Second, target groups are not homogeneous and therefore this particular deployment of the budget reaches some groups more than others. If the higher social class is important, we do not seem to be reaching it very effectively or efficiently. Third, the effect is also caused by the ads themselves. Some of them are clearly better suited to draw attention, are liked more than others and are more likely to stay effective after many contacts. In the next section we will discuss mixed-media campaigns, which nowadays are almost the rule for large advertisers.

MEDIA SYNERGY

Sometimes, not always, media in combination give us a degree of effectiveness that is greater than that delivered by their combined reach alone. This depends very much on the media and their overlap, but most of all on the ads, product, the message and target group.

In the case of Brand Q, we were able to compare those who received different levels of TV and of print. Table 1 shows the numbers in each cell of this matrix and also the effects (in this case, the percentage claiming brand consideration). There is some fluctuation, but we can immediately see that the medium and high TV viewers who are also high readers of the relevant print media show the strongest effects: over 80% compared to the average of around 60%. This group is just over 13% of the sample (185 out of 1,372).

As well as response by combined reach, we can also look at the costs involved. Table 2 shows the budget involved in achieving each level of reach (costs are in '000 euros).

Are the costs being distributed wisely? The High/High group gives a strong result, but the costs are high and they reach only 8% of the target. If we divide these figures into each other we can see both the costs per numbers reached in each cell and the cost per response percentage point (see Tables 3 and 4).

Would it pay to divert a bit of money away from TV and try to increase the Lower TV plus Higher Print groups, which show respectable levels both of reach and effect? This evidence is only part of the argument, but probably yes. Medium TV viewers who are reasonably high print readers look as if they have more room to grow, and it will be less expensive to reach more of them.

BUILDING EFFECTS OR WASTE

To build and maintain a brand over time, we need a campaign that is both effective and efficient. If we see in the early stages of the campaign that the desired effects clearly increase as contact with the campaign grows, we can continue the campaign. But beware – waste could be building up. When the effects rise and the contact frequency builds up over time, the ads will increasingly contact the same people again. The budget that is wasted grows from week to week and the share of waste increases. We have found that the tail of the frequency graph (the High exposure group) often starts high, but then drops after a couple of weeks. As waste builds up, the campaign effects plateau out and remain more or less the same.

WHEN TO TELL IT AGAIN, AND WHEN TO STOP TELLING

What lessons can be learned from frequency build-up? What can we do if waste increases and effects do not? The central theme in the frequency debate used to be in terms of the minimum number of contacts needed: 3+ or 5+ or even 10+. What we have found here and in many other cases is that it would be more effective to limit frequency than to try to increase frequency. The question is: when to tell the story again and when to stop?

There is no general rule to follow; each product and category, each market and campaign, demands a different approach. Therefore it is essential to calculate the efficiency of each campaign separately, and base decisions on how efficiently effects are achieved, instead of on costs per GRP.

We have seen that the efficiency of a campaign may well depend on contact frequency in a medium and (sub) target group. With the help of a fast real-time analysis system we can calculate and judge the most efficient way to increase brand consideration (if that is the key indicator) each week. When the efficiency drops in the High frequency groups or in important subgroups, we may advise a change in media strategy. In this case we would suggest changing to a lower frequency build-up and an increase in reach. Increasing reach means using more media, more channels and more different programme contexts. If only 20% of the budget from the High-frequency group could be re-allocated, this would increase effectiveness and efficiency with almost the same amount of expenditure overall.