Dear Tig:
I was in a meeting yesterday with a company whose ad agency suggested that they have recently converted from believing in "effective frequency" as the guiding philosophy for buying media, to a "maximizing reach" approach.
This new approach holds that media plans should focus only on reach—that the most effective exposure is the first one, and that after the first exposure, you have diminishing returns.
When the client protested that this was a radical and wrong-headed approach, the agency countered that this is what the most sophisticated, large advertisers have validated through extensive testing. It is, the agency said, the new standard. This is news to me…and the client!
I have two questions: A) Is it true? B) Is there any landmark research study that represents the new "state of the science" that speaks for or against this perspective?
Thanks, Bill in Virginia
Bill,
Ad agencies are great for this. They love to take running leaps at marketing theories and trends—and to maintain that this maximizing of reach approach is the best for ALL campaigns is most certainly a leap.
It's a fine approach for many types of campaigns, but not for all. Leaning on reach rather than frequency is generally recommended in many—perhaps even most cases—such as:
- Advertising with compelling product categories and compelling ads, creating relatively instant cognition of the message. (The trick here is that no agency is going to admit that their ads aren't compelling for the category.)
- When the product category relies on specific time periods of great activity, like dating services near Valentine's Day.
There are many, many other cases in which reach would be the goal. This company you reference may in fact fit into one of these slots.
On the other hand, there are quite valid reasons to choose a high-frequency strategy:
- Advertising a commodity-like product, where purchasers tend to buy the brand most recently seen.
- A company that wishes to beat its chest (so to speak) and appear to be a very large, legitimate firm to a relatively narrow audience.
Media vendors generally have a tougher time selling deals that involve high frequencies, because they tend to drive fewer sales than those with lower frequencies.
Many studies have been done to determine the best efficiency points for frequency. Most I've seen wind up coming to the conclusion that two to three impressions is a sweet spot.
This works nicely because media planners tasked with putting together a campaign of any media weight at all will find it difficult not to get some overlap in a targeted set of vehicles. In other words, to get a two-frequency, you need to plan as though you were seeking a one-frequency.
Some other studies have suggested useful frequencies as high as eight over a short period of time. A disturbing trend I've noticed is that those studies commissioned by media vendors tend to suggest a higher desired frequency, which doesn't speak well to the neutrality of their methods. Almost all of these studies fail to take into account the more esoteric worth of the branding that goes on during these campaigns, relying almost exclusively on directly attributable sales.
However, the accounting for that branding is generally exaggerated by media vendors. When an advertiser conducts a direct sales campaign, the branding effect is sometimes not overly positive.
It's not all that useful to take a generalized desired frequency number, as several brand-specific factors will come into play.
Taking an online example (the data we get back here provides a good background for this type of analysis), you may find that the average first impression drives a 20 percent return over the investment of media costs, and that the second impression through tenth impression returns a 30 percent loss over the investment.
But, if you buy 10 times as much, you might get 40 percent off the cost of the media, meaning that the larger scale will give you a 100 percent ROI on the first impression and a 10 percent ROI on each subsequent one.
In general, large campaigns with lower frequencies will beat the results of the same-sized spending with high frequencies. But there are devils in those details.
--------------------------------------
Dear Tig,
What's a good click rate for online ads? I'm too embarrassed to disclose what my clients are getting. It's quite a bit lower than what the sites tell me is their average. Is there any way to predict click rates?
Hope you can help, Insecure in Ottawa
Insecure,
I've noticed that funny trend too, whereby sites tell us the average rate is always quite a bit higher than the performance we see in our own campaigns. It turns out to be true of all my friends and their campaigns as well.
I think this may be indicative of what I call “portfolio creep.” If you were to ask the average individual investor what type of return he or she made on stocks, you'd find that people tend to forget the terrible investments and relish the memories of those that did spectacularly well.
This leads to somewhat unintentional exaggeration. (A similar effect exists with memories of Saturday Night Live skits, whereby everyone believes that the show was hilarious 10 years ago, but is drivel today. This roles along so that we'll remember only the funny parts a decade from now.)
Click rates vary wildly by product category, type of creative, type of placement and even such things as the individual brand itself. The only real benchmark to go on is the previous performance of the same product's advertising.
That said, I once created a couple different formulas for predicting click rates. The more accurate model involves a lot of variables:
- You start off with a basic click rate of 0.125 percent—then you factor in the product category influence: Ranging from a multiplier of 0.1 for a boring product—like printer paper—to a factor of 5.0 for something sexy—like lingerie.
- To add in the creative element, you have to factor in the message strategy element, multiplying by a low of about 0.3 for text-intensive treatises to 5.0 for eye-catching imagery. Remember, click rates (not performance) are all about eye catching.
- Multiply by the degree to which the target market is present on the placement site (a factor ranging from about 0.5 to 3.0).
- Take that and multiply it again by the degree to which the non-target audience mistakenly believes they are interested in the ad (from 1.0 to 4.0).
- Take out the frequency effect by dividing the result by the average number of impressions each viewer will see.
- Add an arbitrary 15 percent of the result to the total for each ad server involved in the process, as they tend to throw in garbage clicks of various sorts.
While this will get you reasonably close to the click rate, a far easier solution is just to ask the site's rep what he thinks, then divide by 3.5.
