Introduction

What scientific knowledge do we have about TV? We review how well certain empirical generalisations have held up over the last 40 years and use this knowledge to construct an empirically grounded opinion on the future of TV advertising, encompassing:

• Is there program and/or channel loyalty?
• Do genre specific channels attract
  segmented audiences?
• Why have ratings dropped?
• Why is TV advertising more expensive?
• Will DVRs destroy the advertising
  model?

To start, we look at the three distinct eras of television. The eras are common to all development cycles, but there is no universal timeline as country-by-country infrastructure meant that the medium developed in different periods in different markets.

The ‘Golden Age’

For the first 50 or so years of television, family life in developed market economies often revolved around “the box” in the living room. Everyone watched TV together after, and sometimes during, the evening meal. Children also watched after school and on weekends, at least when Dad wasn’t watching a game. Housewives were the primary consumers of midday soap operas. The evening news was an everyday “must-see” event for all grown-ups in the household.

In the Golden Age, viewers were thought to be highly loyal to their favourite programs. And to a reasonable degree, they organised their lives around their choices. In this somewhat mythical world, advertisers could rely on TV to be vast and fast—it could reach the entire target market within days. For mass-consumer brands, it gobbled up most of their advertising budgets.

Multi-Channel Access

With the advent of cable and satellite signal-sharing, the number of channels started to increase. Genre-specific channels arrived and average program ratings came spiralling down. At the same time, the number of TVs in households continued to increase. The arrival of the VCR enabled viewers to time shift their favourite programs to more convenient viewing times—a technology that was more promising in theory than in practice as few households took advantage of the full capability.

The Digital Revolution

In the last ten years or so the digital revolution has seen TV compete for viewers and advertising dollars against a plethora of “new media’”, while new technologies like digital video recorders (DVRs) have given viewers greater control over their viewing experience.

The introduction of digital terrestrial television (DTT) has further grown the reach of multi-channel access. In the UK the introduction of Freeview (DTT) saw the number of multi-channel households rise to the 80% level that had been achieved long ago in the US with cable penetration. At the same time, screens multiplied into hand-held devices with connectivity that was only dreamt of in the past. Resulting in more screens, larger and smaller, and more content than ever before to fill them.

So the TV advertising world is now a much more complicated place. It would be nice to know if there were any reliable consistencies to guide advertising planning today. It seems extraordinary that law-like patterns concerning TV viewing could survive across the three eras described above. And yet we will show this to be the case.

Decades of Replications

Many of the empirical generalisations we reference come from the work of Goodhardt, Ehrenberg, Collins and Barwise around the 1970s (Barwise et al. 1979; Barwise et al. 1982; Goodhardt et al. 1975; Goodhardt et al. 1987). At the time, their findings challenged established wisdom but won acceptance through the use of empirical generalisation. Detailed and laborious analyses of large complex data sets were summarised to simple patterns. Patterns that held across different years and markets (especially the USA, where television supply was already far more advanced than in the ‘underdeveloped’ UK market of three or four channels) to the point where they became scientific laws of audience behaviour.

In the multi-channel era, Beal replicated this work in collaboration with these pioneers (Beal 2003; Beal et al. 2004; Collins et al. 2003; Collins et al. 2006) finding much the same patterns of viewing. But given the impact of the Digital Revolution over the last eight or so years, we also present further selected analyses for last year (2008). We find that the future of TV looks bright.

TV is Still a Mass Medium – Vast and Fast

Commercial TV remains a broad reach medium. While many news stories suggest that viewership is down significantly, TV’s reach remains robust and impressive. In Australia, the UK and the USA, around a third of the population is watching television at any point during evening prime-time hours. If the sample is extended from just one moment to any point during the entire evening, the reach cumulates to around two-thirds of the population. Indeed, practically everyone in the developed world watches some TV during their average week.

TV viewing levels remain surprisingly impervious to the increases in consumption of new media. The average daily hours spent watching television have remained at almost four hours in Australia, the UK and the US for decades. In 2008, Americans were actually watching more television than ever (Nielsen, 2008).

Not Everyone Watches the Same Amount

Although TV’s reach has always been powerful, many advertisers do not appreciate the fact that there is great heterogeneity in viewing levels between people. The distribution of viewing is very skewed and the “average” viewer is not the typical viewer. Ever since the Golden Age, most viewers have watched less than the average viewer. And a few viewers watch a great deal of TV. This means that advertisers cannot take TV’s reach for granted; it is perfectly possible to buy a TV schedule that delivers low reach per dollar.

This skewed pattern of viewing is a phenomenon that applies across countries and over time as shown by Figures 1 and 2 (housewives in New York in 1974, and Australians in 2006). Some of the variation between individuals is systematic, with age being the biggest single driver. Older adults watch more than younger adults; and, because they go to bed earlier, children watch less television. In recent years low viewing levels for children have been mistakenly interpreted as evidence that younger people are replacing TV with the Internet. In fact they always watched less TV than their elders, before and after the arrival of Facebook and Twitter.

Despite such systematic differences the amount of television an individual consumes is not primarily determined by age, sex, income, or any other demographic classification (Barwise & Ehrenberg, 1988). Within any subgroup, there still exists the pattern of variation (shown in Figures 1 and 2). This further reinforces the true mass-medium status of television: It is too widespread to be closely identified with any particular segments of any developed market population.

This empirical generalisation has been well described (Goodhardt et al, 1975). A key finding for advertisers was that lighter viewers of television are not more “selective” or regular in their viewing choices, but rather are best reached with high-rating programs. In fact, higher-rating programs take a much larger slice of light viewers. Higher-rating programs in the UK have amost double the proportion of light viewers (15 percent of light viewers in their audience compared to only 9 percent for low-rating programs) within their much larger audiences. This finding holds true in 2008 every bit as much as it did in 1971 (see Table 1). This highlights the worth of placing advertising in higher rating programs in order to reach these ‘hard to reach’ lighter viewers.

Fragmentation: The Advertisers’ Enemy

Multi-channel access and the digital revolution have had their effect; television’s reach is still vast but it doesn’t offer the easy reach to advertisers that it once did. The availability of more channels has spread audiences more thinly across the plethora of content. For example, in the 1970s in the UK, viewers could pick from three available channels; the top five programs had an average rating of 34. In October 2008, with more than 200 channels available to UK viewers, the average top-five programs rating had dropped to 17. This means advertisers have to be a bit clever with their TV buying today –they can still reach 34 percent of the population with multiple spots carefully spread over time and across different networks.

Paradoxically, fragmentation is one of the reasons TV advertising has become more expensive. In TV and other media, fragmentation means that the highest-rated destinations (programs or other kinds of content) become rare and therefore more valuable than ever. In parallel, the number of companies that want to advertise has also increased, making the competition for the most highly rated shows especially keen. Industries and professions that never would have considered advertising in the past are now bidding against one another for network time and, by doing so, driving prices higher.

Mass Medium or Targeted?

TV, as a whole, remains the ultimate “mass medium” capable of reaching vast numbers of viewers. But most small channels fall far short of this target. It’s common in marketing to describe small brands as niched and, similarly, many small TV channels are marketed to advertisers as being able to deliver loyal niche audiences. However, Barwise and Ehrenberg (1988) found the same basic double-jeopardy pattern in TV viewing that had been noted in consumer goods markets for many years (see Ehrenberg, Goodhardt and Barwise 1990). Far from being attractively niched, many small channels suffer in two ways: few people watch them and those who do tune in tend not to watch them for very long (See Figure 3). So, advertisers should be very wary of claims of highly targeted, highly loyal TV channels and instead carefully check the audience metrics. Most small channels aren’t niche, they are simply small.

Given the many changes that have occurred in the television industry since the late 1980s, it is startling that replication of this double jeopardy analysis in 2001, 2002 (Collins, Beal et al. 2003) and 2008 yielded the same general pattern. The results are shown graphically in Figure 3, with the historic curve from 1980s data as a benchmark. The curve still broadly reflects the shape of the data to the present day. Across countries and decades, the larger channels still enjoy both high reach and higher hours per viewer with the many smaller channels clearly suffering the effects of Double Jeopardy. Knowledge of this empirical law is particularly important for advertisers to understand which channels they can use to build maximum reach for their campaigns.

The US data shows two particularly marked outliers: Univision and Telemundo are true “niche” offerings with very low reach among the general population but each generates high average-hours-per-viewer numbers. These are Hispanic channels that really are restricted to an audience of Spanish speakers. Of the 150-plus channels we studied, such niche channels are very rare. Even when viewing of a small channel is more intense than average, it still accounts for only a small proportion of its viewers’ total viewing time. This is accentuated by the tendency for viewers of smaller channels to watch slightly more TV (about 10 percent) than the average 25-30 hours. And just as light buyers tend to buy the large, well-known brands in product categories (the Natural Monopoly law), so do lighter viewers tend to watch bigger, less niche channels.

Do Genre-Specific Channels Deliver Special Audiences?

The massive growth in channel numbers has brought both a proliferation of channels with genre-specific programming—children’s, sport, music, movies, news, documentaries—and a number of channels with a claimed focus on a particular population group (especially young adults). As we’ve already discussed, this range does not yield any extreme patterns of loyalty (apart from US Hispanic channels).
If TV channels are rarely or only mildly niched, can they offer targeted audiences? Channels certainly deliver different audiences over the day/week, as do different programs to an extent, and there are channels with specific geographical coverage – but do genre specific channels deliver specific audiences?

Audiences actually vary between channels far less than might be expected. Surprisingly, this finding is true for ultra-small channels as well as mass networks. In the ‘Golden Age’, when there were no genre-specific channels, the audience profiles of the different channels were almost identical (Goodhardt et al, 1975). Years later, the largest channels continue to have almost identical customer profiles (see table 2). The demographic audience profiles (shown in Table 2) provide a solid benchmark to show that some genre-specific channels do attract markedly different audiences, but they are all the obvious culprits:

• Children’s channels: 55 percent children (average: 17 percent); 60 percent women (average: 49 percent).
• Music channels: 70 percent under the age of 35 (average: 39 percent); 5 per-cent aged 55-plus (average: 16 percent).
• Sports channels: 65 percent men (average: 51 percent).

But note how such targeting is far from exclusive, e.g. almost half the viewers of children’s channels are adults. Some channels might appear, at face value, to be very selective in terms of targeting population subgroups, however, they aren’t exclusive and they tend to have very restricted reach even into their targeted subgroup. For instance, in our research, of those who viewed “Men & Motors,” the men’s lifestyle channel in the UK (Tagline; “Fast Cars and Women”), 30% were women, and only 5% of the men who subscribed to the channel actually watched it in a week, and those that did devoted less than an hour of their 30 hours of weekly viewing to it.

It might be expected that viewers of one genre-specific channel are likely to watch other channels of the same genre. They are, but they still spend 75 percent of their viewing time in front of the big general entertainment channels (including, in our UK research, Sky 1, UK Gold, E4 etc.). The significant pull of these mass offerings leads us to conclude that viewers’ tastes are broad and most TV audiences favour general entertainment. We find much the same patterns when we look at viewing of specific programs classified by genre.

Is There Any Channel Loyalty?

At first glance the answer would appear to be no – viewers have favourite programs and any channel loyalty is purely spurious. However, another empirical generalisation, first reported in the famous science journal Nature way back in 1966, reveals there is channel loyalty. The empirical generalisation is that programs share viewers simply in-line with their ratings. That is, any program shares a large proportion of its audience with high-rating programs, and a small proportion of its audience with low-rating programs. The duplicated audience for any two programs is proportionate to the product of the two program ratings (Goodhardt, 1966). Quantitative analysis using this law over the decades has revealed a sub-pattern, another empirical generalisation, exposing a degree of channel loyalty—duplication between programs is always higher within a channel than between programs on different channels. In the 1980s the duplication “constant” needed for within channel was 1.4, compared with 1.0 for between channel. Later, the within/between differential became 1.7 compared with 1.0 and in our very recent analyses it appears more like 2.0 vs 1.0. This rising channel loyalty is associated with increasing audience fragmentation and suggests that, when confronted with a large number of channel choices, viewers restrict their viewing to a small group of learned favourites. Channel loyalty is not necessarily good news for advertisers. It does suggest that spreading advertising across channels is increasingly important if they wish to gain high reach relative to frequency of exposure.

Program Loyalty – Repeat Viewing

One of the most surprising empirical laws concerning TV viewing is how common it is for viewers to miss the programs they like. Early work in the UK and US found an average repeat-viewing rate of around 55 percent (higher for the highest rating programs and lower for the lowest, again in line with the Double Jeopardy law5). By the mid-1980s, after the introduction of a fourth channel, the levels in the UK fell to 40-50 percent (Barwise, 1986; Barwise, Ehrenberg et al., 1982). In the US, the addition of more channels meant lower ratings and lower repeat-rates across the board: 40 percent for most prime-time network programs and 30 percent (or less) for off-peak series (with the exception of daytime soap operas). In early 2000s, levels continued to drop; repeat figures for established channels were around 30 percent and much lower for the new lower rating channels (see Table 3).

Why were these repeat-levels thought to be surprising? Mainly because program-rating levels remained consistent from week-to-week. Many had incorrectly assumed that mostly the same people were watching week in week out. This seems especially reasonable for programs with continuing stories, however, viewers miss most of these episodes too.

Why do viewers miss programs? Largely because they have lives to live. The dominant reason that viewers miss a program they watched the previous week is that they don’t happen to be watching TV at the same time the following week. For lower-rating programs, watching a competing channel becomes a more prominent reason for missing an episode. “Appointment Viewing” is not the norm.  The low repeat-rates explain, in part, why DVDs of TV programs sell so well.It also helps to explain why “re-runs” continue to rate well. For advertisers, it means that programs have a powerful ability to build cumulative reach over episodes.

Finally, this law-like pattern shows that niche programs are rare and seldom particularly strongly niched. Low rating programs almost invariably have viewer-bases that are less, not more, loyal. TV programs in general fail to attract much dedicated loyalty because it’s just TV: It has a place in
people’s lives, but viewers rarely let a show
get in the way of their lives.

Loyalty to TV programs is like consumer loyalty to brands; it’s largely due to habit and inertia, reinforced by weak attitudinal preferences (Ehrenberg, 2004). The search for highly engaging programs looks a bit like an exercise in ‘snark hunting’ (see Carroll, 1876).

Will Ad-Supported Broadcast TV Survive?

The empirical patterns of television-viewing behaviour observed over 40-plus years support the notion that the current leading advertising-supported channels will remain popular, although further fragmentation will reduce their audience share. These channels alone deliver significant reach levels even in short time periods (i.e. a week). They will retain their premium status for advertisers because they will remain the best source of rapid cover build. In terms of cost, higher CPMs may irritate television advertisers. But CPMs are not an arbitrary set of numbers; they are the market’s best estimate of relative value at the time. It is important to recognize, “CPMs don’t ignore media value, they monetise it” (Ephron, 2003); they reflect the value of the large audiences that mainstream television can deliver to advertisers.

Although large channels may have lost share, they remain attractive to major advertisers because large channels achieve higher reach levels than the smaller channels and are watched for much longer by a wider variety of people. That variety includes light as well as heavy viewers. And, in many instances, it’s just those light viewers—young, affluent audiences who have more in their nights than a regular diet of TV—who are most attractive to advertisers. When they watch television, they are more likely to watch the big popular channels. This makes these traditional channels well placed to reach such desired advertising targets. To that end, audience fragmentation actually has been beneficial for the large broadcasters. Even though they’re smaller than they were in the past, they remain the last bastions that reliably can deliver a mass audience.

The television advertising market has been extremely resilient and, according to World Advertising Trends 2008, the medium achieves a 46 percent share of total ad spend ($563bn) (Shown in Figure 4). By comparison, in 1998 TV’s share of total global adspend was 38 percent.

Will Live Ad-Supported TV Survive?

Many of the law-like patterns we have described stem from viewer habits. They have lasted for decades and survived despite enormous changes in TV and competing media. Their continuing presence strongly suggests that television still holds a habitual place in people’s lives.

Will new technologies upset these habits? Undoubtedly they will cause changes but will they be changes that really matter to advertisers? For example, the DVR has been considered a menace to the advertising-supported model that underpins TV throughout much of the world. It has the potential to violate the ‘unwritten contract’ that has provided billions of dollars of content to consumers at a very low price in return for those viewers watching embedded advertising. However, there are good reasons for thinking this threat is minor. Our contention is supported by yet another empirical generalisation: viewers do not watch all the commercials they could. They frequently undertake complete (leaving the room or switching channels) or partial (reading, surfing the Internet, or talking to companions) avoidance. Decades of research, using a wide variety of different methodologies, have converged on a consistent descriptive pattern of audience behaviour during commercial breaks: one-third active viewing; one-third partial viewing; one-third complete avoidance (Paech, Riebe and Sharp, 2003). DVR fast-forwarding and other new technologically-driven tools are then probably more likely to simply replace other avoidance behaviours, leaving the actual viewing of commercials at the same quality level it has maintained for years. This subtle shift in habits would explain why the analytical reports comparing advertising effects in DVR and non-DVR viewing conditions are highly similar.

Back to the Future

In the face of great changes and the emergence of competing advertising media, TV remains the preeminent advertising medium. Although it is more complex and expensive. It is easier than ever for TV advertisers to waste money, and we would argue that advertisers who ignore the empirical laws described in this article are more likely to buy ineffective TV schedules. We have reported a substantial number of empirical generalisations that hold not just across countries, but also persist in spite of the technological and social changes related to TV. They are fundamental knowledge for any advertiser:

• TV has extraordinary reach: Practically everyone watches TV regularly.
• There is a skewed distribution of hours of viewing TV, with few very heavy viewers and many light viewers.
• As people age, on average they watch more TV. But within every age group there is the full variation in viewing levels.
• Declining ratings are due to fragmentation (more channels), not to reduced TV viewing levels. Average ratings halve as the number of channels doubles.
• Bigger channels have more viewers and who watch for more hours – a Double Jeopardy law. Small channels are rarely niche, they usually attract lower, not higher loyalty.
• Duplication of Purchase (customer-sharing law) applies to TV program viewing. Within channel program duplication is higher, and rising, indicating channel loyalty.
• All large popular channels attract very similar audience profiles.
• Genre-specific channels can attract specialist audiences, though very small channels essentially attract small audiences rather than different or unique audiences.
• Program genre loyalty is weak as TV audiences have broad tastes.
• Repeat-viewing rates are low, and related to ratings (another Double Jeopardy pattern) so they have declined as ratings have declined.

These empirical generalisations have held over time and across countries, strongly suggesting that we can expect a number of things to stay essentially the same during and after the Digital Revolution. This should come as a great relief to TV advertisers who need to plan media spend. The results also are reassuring for those who continue to make investments in understanding the effects of TV advertising. Advertising laws concerning TV appear to have considerable longevity – we should discover more of them.