THE SHELF SPACE MODEL OF ADVERTISING
Advertising Needs to be Present, Because Not Being There With A Message Is Like Being “Off the Shelf.”
By Erwin Ephron
Henny Youngman said it. "I've got enough money to last me rest of my life, provided I die at three o'clock." That joke is the agency's nightmare.
There's never enough money for a full year of advertising at effective weight-levels, so we make do. We use more 15's, buy cheaper TV and run fewer and fewer weeks, driven by our belief in "effective frequency," the idea that advertising has to be concentrated for it to work.
Yet despite billions of advertising dollars and years of brand experience, the industry has failed to prove the value of effective frequency compared to other scheduling strategies.
Now there is strong evidence we have been riding the wrong horse. John Philip Jones at the Newhouse School of Syracuse University and Walter Reichel of A to S Link, both using Nielsen single-source panel data, have found one exposure to a brand message in the week preceding purchase has a far greater effect on brand share than additional exposures.
More disturbing, a recent ARF paper by Colin McDonald, whose research in the late 1960's was basic to the development of effective frequency theory, disputes the conclusion that his work ever supported the greater value of short-term frequency.
Perhaps effective frequency was never right. Perhaps it is just wrong for today's consumer markets. Either way, we had better start rethinking how to spend the client's money.
Effective frequency theory says advertising works through repetition, because it takes repeating for consumers to learn and retain messages. It is largely based on the work of Ebinghaus (1890), who wrote about the effects of repetition on learning, Zilske (1978), whose research covered the effects of message frequency on advertising awareness and retention and Krugman (1970), who theorized a three-stage process to explain the way consumers process advertising information.
More recent work by Jones (1995) and Reichel (1994) used purchase data to measure the effects of frequency and found repeating ad messages a waste, probably because advertising works best when people are ready to purchase. Their data say buy reach and ignore effective frequency.
The issue is bigger than media planning. It's the question "what makes consumers respond?" A frequency strategy says advertising works like stuffing turkeys. A reach strategy says it's more like waiting for hungry fish to bite.
My own view is "stuffing" is how TV used to work, when many products were so new, consumers had to learn to need them, and "fishing" is how it works today, when you don't have to teach consumers about consumables any more. They know them and buy when the box is empty (or in the case of durables, when gets old or it breaks).
The primary role of advertising is to influence the brands consumers select. To get them to buy your brand, you need to reach them when they are ready to buy a brand.
The Real Media
Target is Purchases
Today, the real media target isn't consumers, it's their purchases. And since these occur 52-weeks a year, the sensible way to intercept them is with continuous advertising. This is the idea of "renting-the-shelf" so a brand message will be waiting. (Likewise, the trouble with flighting is your advertising is "off the shelf" when the flight is over.)
The "Rent-the-Shelf" model of advertising requires a new kind of media planning. Planners and buyers have to forget about frequency and learn to go for weekly reach.
Weekly Reach Planning
The case for weekly reach planning is built on three assumptions.
The usual function of TV advertising is to remind the consumer of the brand.
Consumers know most major brands in the product categories they use. That familiarity is called "brand equity." Unless it's a new brand, a new benefit or a new use, there is not much learning involved. Most TV campaigns are "sustaining" as opposed to "introducing," and most TV messages function to remind more than to inform.
A message is most effective
close to the purchase.
Movie companies pay premiums for Thursday evening to fill the theaters Friday night, Saturday and Sunday. Political candidates spend close to the vote. It's common sense. Close to the purchase the message is more relevant and it is relevance that gets people to notice, remember and act. Relevance isn't mysterious. For Kellogg’s Corn Flakes it may be as simple as an empty cereal box at breakfast.
Close to purchase one
exposure can be effective.
The Jones' analysis of Nielsen panel data shows this clearly. In the week before purchase, the first message does most of the work. For the 142 brands Jones studied, one exposure showed an average share increase of 11%. All subsequent exposures added only 3%. It is not that frequency has no benefit. It's just not a cost-effective way to spend the money.
Using The week before purchase as a scheduling goal appears to be a wild card, since we don't know which consumers are likely to purchase which weeks, but we know something just as useful. Each week some consumers buy the product, so if we rent the shelf and advertise each week, our message will be there when those consumers are ready to buy.
And since the first exposure is most cost-effective, if we can reach a large number of consumers without piling-up second and third exposures, we can use the frequency-money to advertise for more weeks, to reach more consumers who are ready to buy in those weeks.
Recency drives advertising effectiveness and weekly reach is how to execute it. That's why it offers planners remarkable leverage. A closer look at the Jones data shows reach accounts for close to 80% of advertising's short-term effect, frequency only 20. 
Half our messages are
made less effective
because of timing.
How important is this? Since the weekly frequency for most brand schedules is close to two, half of the messages are far less effective, only because of timing. We can turn weekly frequency (20% impressions), into weekly reach, (80% impressions), simply by reducing weekly GRP’s and adding weeks. If we can transform 25% of weekly frequency into reach, we can increase the value of the schedule by 15% (25% times 60%, which is the difference in response value between a first and a second exposure.)
The caveat is we need to do this while maintaining a weekly reach of 35% to 40%, a monthly reach of 65% to 70% and a quarterly reach of 80% to 85%, so we talk to enough people to have a measurable effect on brand sales.
That is the logic of weekly reach planning. It works for products regardless of purchase interval, because the key is shopping behavior not shopping frequency. Even though consumers buy automobiles and computers less frequently than paper towels and detergent, these products are also purchased continuously across the year.
Learning Theory Is Irrelevant
Seldom Teaches .
The idea that one message can produce a response is hard for advertising people to accept. It goes against everything they've been taught about the importance of repetition to learning. But learning is usually the wrong model for advertising.
Learning theory is largely irrelevant because brands are competing for purchases, not teaching messages and it is dangerous because it pushes us into flighting advertising to buy the frequency needed to teach short-term brand awareness--which it assumes, will protect the brand when it is out of advertising.
That is wishful thinking, especially when there are other brands clamoring to be purchased. Reichel's data and the earlier analysis by Colin McDonald, show the sales effects of messages carry across time, but get progressively weaker and are soon no match for current competing messages.
Flighting is a competitive trap. It buys a brand frequency, instead of reach. If the competition has figured this out and advertises continuously while you're flighting, and has as much money to spend, they'll clean your clock.
New Media Planning Goals.
The media goals for weekly reach planning are specific. The first is the highest 52-week total of weekly target reach points. Using the 52-week total penalizes the plan for weeks when there is no advertising and that is as it should be.
The second goal is a target reach of 35 for the average week, 65 for the month and 80 for the quarter. This is to insure the brand message will reach most purchasers in the course of the year.
Flighting and targeting, our usual planning devices, won't achieve these goals. Reach is bought by dispersing messages across programs and dayparts, while targeting concentrates them. And flighting's higher GRP levels spend far too much of the money on frequency. The cost-effective way to buy the highest 52-week total of weekly target reach points, is to limit target rating points to between 60 and 80 a week, disperse the schedules and advertise for more weeks.
Here is an example of how weekly reach planning goal forces lower GRP levels and more weeks of advertising.
The two schedules shown in Table 1, spend the same $15 million. The flighting schedule, which is typical of the way we plan, runs 100 target points a week in six flights covering 26 weeks. It has a weekly reach of 50% when the advertising is running.
Table 1. Total Weekly Reach Points (Equal dollars)
The continuity schedule uses 67 target points a week for 39 weeks. It has a weekly reach of 40% when the advertising is running. Because the continuity schedule with 67% of the weekly TRP's achieves 80% of the weekly reach, (that's a function of the shape of the reach curve), and has 50% more weeks, it delivers 20% more weekly reach points over the year 1,560 (40 reach points x 39 weeks) compared to 1,300 (50 reach points x 26 weeks). That is a big difference in performance.
If we embrace the idea that our planning goal is to schedule messages to intercept weekly purchases, the media planning process gains a crystal-clarity. Let's begin by assuming the product is purchased uniformly across the year.
To cover all weekly purchases with advertising, (i.e. reach all potential purchasers with a brand message within one week of each purchase), we need a weekly reach of 100 for each of 52-weeks, or a total of 5,200 weekly reach points (Table 2.). It is the unattainable ceiling for a weekly reach plan.
The media planning assignment is to generate the highest total of weekly reaches with the budget, expressed as a percent of the 5,200 reach point ceiling. The $15 million flighted schedule we looked at earlier reaches 25% of potential 52-week purchases (1,300/5,200 = .25). The continuity schedule covers 30% (1,560/ 5,200 = .30).
We can do better by using 15-second commercials. The mix of 30's and 15's is not a media decision, but the shelf-space model encourages 15's, because it assumes most brand messages are reminders. Using 50% 15's allows us to add weeks. Now with 52-weeks of advertising at 67 target points a week, the brand can cover 40% of the entire year's weekly purchases (40 reach x 52 weeks = 2,080 / 5,200 = 40%).
TABLE 2. PLANNING FOR WEEKLY REACH.
* % purchasers receiving a message within a week of purchase.
Next, we move to cheaper TV, shifting money from Prime broadcast to Day, Syndication, Cable and Unwired Networks. This provides more TRP’s and greater dispersion, which increase our weekly reach to 46% and our 52-week coverage to 46%.
Now the weekly reach goal pushes the planner to consider other media options. Fifteen percent of the dollars shifted to BDI spot will increase annual coverage of weekly purchases to close to 48%, because spot uses new day parts and focuses on high per-capita sales areas. 
Another 10% of the dollars moved to other media -- monthly magazines and network radio, for example -- will increase our coverage of weekly purchases to close to 50%. Print and radio CPM’s are often lower, so we can buy more target points for the money and because a television schedules duplicateother media vehicles less than it duplicates itself.
A strong seasonal skew in sales might let us add another few percent coverage of sales, but 50% to 55% appears to be as far as a packaged goods planner can get for $15 million today.
The plan progresses towards the theoretical goal of 5,200 weekly reach points (as shown in Table 2). From flighted, all 30's, high-cost TV (a), to continuous, 50% 15s, lower cost TV and other media, adjusted for market-by-market and seasonal variation in product purchases (g).
The shelf space model of advertising is a very different world for the media planner. It emphasizes reach and ignores frequency. It uses dispersion, not targeting, moderate GRP-levels and continuity, not GRP-concentration and flighting and it thinks about incremental Cost-Per-Target-Reach-Point, not just target CPM.
Thirty years ago, Herb Krugman, the father of effective frequency planning, said:
"Advertising needs frequency, because, like a product sitting on the shelf, you never know when the consumer is going to be looking for you, so you have to rent the shelf-space all the time."
We should have been better listeners.
(This was the first comprehensive written discussion of recency planning. It was published in the Journal of Advertising Research in February 1995.)
 The relationship of +11% and +3% to the average share increase of +14%.
 Spot TV should be used to add weeks of advertising, not GRP weight
 In private conversation Herb admitted a better word would have been "presence."
- February 1, 1995 -