Our media strategies editor tries to answer a perennial online media question -- how much is enough creative for an online campaign?

There are always questions that clients ask their agencies when they are setting out on a campaign for the first time.

What are the best media to use? What are the levels of each media needed to achieve the stated communication delivery goal? How long do I have before I have to make a decision on a schedule? What are my cancellation options if we decide to end a campaign before the originally planned end-date?

One question that I've heard over and over again over the years, second only in frequency to "how much should I be spending?" is this:

"How many different creative units should I run over the course of a campaign?"

More often than not, this question lands in front of online media planners rather than the creative team. So much of online advertising is still driven by media considerations that for clients it is natural for them to pose it to the planner on the account. The number of different sizes and their placements are often a function of the media plan and not the creative strategy. This is changing: as the numbers of people viewing online creative are doing so through broadband connections, this has led to a more robust palette and more adventurous creative strategies, but a majority of online ad campaigns are still determined by the media.

The question isn't really just about the number of different kinds of units. This is not the kind of "need" the client is referring to. It is the number of creative executions that are required for an online media schedule.

The reason this question frequently ends up before the media group is because there is a sense that the number of executions are dependent upon the media weight. The question is borne of a consideration of the number of people the marketer wants to speak to (reach) and how often the marketer is likely to speak to them (frequency).

If audiences are confronted with the same message too often, there is a risk of fomenting annoyance or resentment in the audience. If the audience doesn't see the message enough, there is a risk of not having an impact at all.

So, the most common answer to this question -- and the one that you, gentle reader, have probably answered to yourself under your breath -- is that it depends.

It is a good question to ask. You don't want to have too many different pieces of creative and dilute the impact that any one of them might have. At the same time, you don't want to limit your chances of attracting different potential consumers with a variety of messages and images that might appeal to different segments of your audience.

Why Does it Matter?

In traditional media there is some conventional wisdom that holds the "wear out" threshold for a single execution is somewhere at a 20-to-25 time frequency for your first audience quintile. That's traditional media speak for "the heaviest one-fifth."

Quintiles of media usage break down as follows: heaviest, heavy-medium, medium, medium-light and light. I realize that this reads like a list of milk choices at an organic market, but this is how it's done.

The reason so many brand advertisers run several different creative executions is to keep the message fresh, even if the brand the message promotes is the same. With multiple creative executions, an advertiser can push the limits of achieving maximum reach and frequency without annoying a majority of audience with the same ad over and over again.

Online, until you can break down a potential frequency distribution against a unique audience, the pursuit becomes how many creative should be rotated for a given impression guarantee. For data-driven campaigns, where the media plan is subject to regular rounds of optimization based on the achieving particular numerical goals, the rule of thumb I've always used was that 10,000 incidents of a unique piece of creative in a unique placement (e.g. blue.gif in the home office radiology section of a site run for 10,000 impressions) will give you enough statistical evidence to determine whether or not your creative is performing.

This, of course, is tricky, because it doesn't take into account the unique audience. Until we can buy against audience rather than blind gross impressions, this will remain tricky. However, if the online campaign has response objectives, it can be argued that the frequency distribution is not of great concern, only the rate of response.

Depending on the objective, particularly if there is some response metric being used to measure success, you should have at minimum four pieces of creative ready regardless of the impression level. That way you can set up a benefit versus offer message test and have one piece of creative for back-up either to rotate in to replace the loser of the first test or to serve as test against the control creative for whatever message category wins out  (i.e., the winner of your first test).

It goes like this. I've got a banner that is red and says, "Live longer!!" I've got another piece of creative that is a blue banner that says, "Live Happier!"

I've got two formats (red versus blue) and two messages ("Live longer!" versus "Live Happier!"), but I should also have at the ready a blue banner that reads "Live longer!" and another red banner that says, "Live Happier!" You can test not only which format might be doing better for your online campaign, but also which message in combination with that format will pull better response.

Experience and research can help to predict what works and what doesn't, but you never really know what is going to work with a particular campaign until you try it.

Another thing to think about when this question does come up is just why the client is asking. The client could be asking this question because they are thinking about being spare with a production budget. Or maybe it is a brand effort, and the client simply doesn't want to burn out their audience.

For campaigns run against the principles of direct response, the appropriate number of creative executions to be used can eventually be determined through testing and optimization, but I've found that the rule of thumb above works well for the purposes of both predictive and post-performance analysis. For a branding campaign, it really depends on not only the level of media weight, but also the level of tolerance the prospective audience has and the appeal the message will have to it.

Jim Meskauskas is media strategies editor for iMedia Connection.