ellipsis flag icon-blogicon-check icon-comments icon-email icon-error icon-facebook icon-follow-comment icon-googleicon-hamburger icon-imedia-blog icon-imediaicon-instagramicon-left-arrow icon-linked-in icon-linked icon-linkedin icon-multi-page-view icon-person icon-print icon-right-arrow icon-save icon-searchicon-share-arrow icon-single-page-view icon-tag icon-twitter icon-unfollow icon-upload icon-valid icon-video-play icon-views icon-website icon-youtubelogo-imedia-white logo-imedia logo-mediaWhite review-star thumbs_down thumbs_up

What is Optimal Frequency?


Reach and frequency are the watchwords of media planning. Achieving the right reach and frequency is obviously a critical goal of media planning. But what is right? What is optimal?

Although marketers in all media are interested in the concept of optimal frequency, few know how to actually measure it. Common sense tells us that too few impressions won’t generate significant impact, while too many impressions delivered to a given user results in over-saturation and a waste of media dollars. So optimal frequency is simply finding an average frequency that maximizes ROI, right? Not exactly.

Over the past two years, we’ve analyzed the reach and frequency of countless campaigns and have studied the impact frequency has on conversion rates. Three insightful trends have emerged. 

1. A large proportion of your impressions are going to users who have already seen too many.

If you’re buying online media on a CPM basis, the cost of impressions to users who have long passed the optimal frequency level can be significant. For example, a typical online marketing campaign with an average frequency of five delivers one-third of its impressions to users who have already received 10 ads. Here, the average frequency doesn’t capture the fact that most users only get one or two impressions, while a small group of high frequency users pull the overall average up. 

2. Conversion rates typically diminish around four to six impressions.

We analyzed campaign data from 38 different advertisers to understand the impact frequency has on conversion rates. For each advertiser, we calculated click-based conversion rates at each frequency level and aggregated the results. The overall results showed the conversion rate on the first impression was the highest, and the first three impressions all had at least double the average conversion rate.

These findings may come as a surprise to many of you, but they follow one of the basic laws of direct response advertising. If a user hasn’t responded to your offer, it’s an indication that the offer isn’t relevant to him right now. Thus, rather than making another offer to the same person, you’re better off delivering that next impression to someone who hasn’t seen it yet. This is the reason maximizing reach and lowering frequency almost always increases conversion rates.

3. The most profitable frequency levels are often much higher than the most efficient frequency level.

Maximizing your conversion rates is always a critical goal. However, the frequency level with the highest conversion rate may not necessarily be the same frequency level to maximize your profits. There will always be a trade-off marketers have to manage between efficiency and volume. Restricting frequency to only one ad per user might achieve the lowest possible cost per conversion, but you may end up with a very low total number of conversions. The question then becomes, “at what level of frequency am I still meeting profitability targets on my conversions?”

This is an answerable question and depends on three variables: the conversion rate at each frequency level, the CPM of the media, and the cost-per-conversion goal of the advertiser. By applying the CPM to the conversion rates, you can calculate cost per conversions for each frequency level. As long as the conversions garnered at the higher frequency levels hit their profitability targets, marketers should be willing to pay for those impressions. 

What does this combination of data, methodology and analysis buy us? In a word, results. It is important to recognize and react to the amount of money being wasted on excessive high frequency users. The culprits are not the users who consume four, five or six impressions, but rather the thousands of users who receive hundreds of impressions without any response. Basic frequency caps can translate into double digit gains from one campaign to the next.

Frequency distribution reporting from third-party ad-serving systems is one of the most valuable, yet underutilized, tools to estimate and manage the amount of waste that might be occurring. The amount of waste will vary dramatically across sites and will increase over the life of campaigns. If the amount of gross waste is significant, knowing how various caps will impact campaign performance and monitoring whether negotiated caps are truly in effect will help in planning and buying media more intelligently.

Consider performing your own statistically valid optimal frequency analysis. This will enable you to precisely identify the frequency levels that maximize total conversion yields, while still meeting your cost-per-conversion goals. It’s important to note that even within the same industry category, optimal frequency levels can vary greatly. Diverse media strategies, different demographic targets and even seasonal variables all affect the results. Only by analyzing your data in the context of your campaign strategies can you determine your optimal frequency.

Achieving optimal frequency is one of the most direct paths to reaching and exceeding profit and conversion objectives. But it can be a path that is hard to follow without the right data, the right methodology and the right insights.

Karl Siebrecht is senior vice president and general manager of Atlas, a provider of digital marketing technologies and expertise, and an operating unit of aQuantive, Inc. Find more information about the research discussed within this article here.

Well, that's a good question. And the answer is everyone's favorite: it depends.

One of the problems with targeting based on narrow data fed through assumptions about action is that your results are only as good as the initial assumptions. And the problem with knowing how good those assumptions are is that you can never really know how good they are. You can only assume they are or are not based on actions that may only appear to have anything to do with the assumptions made in the first place. The truth of the matter is that our results satisfy two, possibly more, logical fallacies: "appeal to belief" (the notion that because enough people believe something to be so, it is so) and "questionable cause" (to conclude that one thing causes another simply because the two are associated on a regular basis).

Another problem is transparency. A senior media executive at an agency best known for its powerful branding for major companies says, "My main concern with [behavioral targeting] is the lack of auditing. There needs to be a third-party auditing system that rates efficacy of claimed behavioral targets so that if I have three nets on my media, each one can have a score as to how well they delivered on those behaviors."

The point is: how do I know a certain profile is a certain profile?

And finally, there is the possibility of overtargeting versus under targeting.

The specificity of the "levers one can throw," as Eric Porres of Underscore puts it, or the lack thereof, can create problems.

"The problem is the age-old 'embarrassment of niches' problem," he says. "Get too granular and you give up reach. Get too broad and you give up specificity."

There is no doubt that BT as a tactic is going to be a major part of our online media planning and buying future. As the media executive from the brand-focused agency says, "Behavioral targeting will probably be very important to our media plans, and the major stake holders in internet advertising certainly think [it is here to stay]," as is evidenced by this year's consolidations.

But will it ever remove the "X" from the "X-factor" of advertising? Probably not.

While it's typical to rely on numeric measures to articulate success or failure of an advertising campaign and the tactics deployed for it, sometimes those aren't always meaningful when determining behaviorally targeted media success or failure.

Behaviorally targeted media is itself a result of decision engines that must rely on an imperfect array of assumptions about what a machine-readable action, a single datum, signifies about the agent of action. That is, does a click by a guy on an automotive ad that is shown to him on a travel site -- shown to him because in the past seven days that guy visited five different pages relating to travel or that had travel-like content on them -- mean anything at all? Maybe it's just a click?

"As an industry, we are way too focused on clickthrough rates, and that type of thinking does not necessarily apply to behavioral targeting," notes Regina Sebring, vice president, channel management for Revenue Science.

Though at the same time, it is possible that when a person clicks on a behaviorally targeted ad, the advertiser can deduce that the clicker is already interested in whatever is being advertised. Behavioral targeting ostensibly eliminates the random clickers who might end up aggregated into a standard CTR. But a simple numeric focus doesn't allow for the kind of read of data one needs to commit in order to properly assess if a behavioral targeting tactic works or not. The approach ignores the same method of correlative interpretations that went into constructing a behavioral profile in the first place.

Sebring continues, "Just looking at CTRs is the wrong metric. When advertisers use behavioral targeting, they need to evaluate the campaign results from a holistic standpoint and not just their average CTR."

So, then, what else should you look at when determining the efficacy of behavioral targeting tactics?

When BT works, it works very well.

Eric Porres, COO of online media strategy agency Underscore Marketing, said that for client Bombardier Flexjet, a very high-end service providing private business air travel, BT was very successful, using both Tacoda and aQuantive's DrivePM (aQuantive's purchase by Microsoft this year for some $6 billion is another testament to the marketplace's belief that technologically driven advertising is where things are headed).

"When you're selling a multi-million [dollar] product, you need to weed out the aspirational traffic from the 'real' traffic, and BT profiling -- if properly executed -- can add that prequalification layer," says Porres. Doing something like that requires customizing a behavioral profile build, one of the advantages of behavioral targeting's technological foundation.

A leading Northwest, high-end auto franchise that wanted to drive traffic to its site to increase online sales leads looked to Revenue Science, another provider of behaviorally targeted services and media. To help achieve its objectives, Revenue Science created a customized targeting solution that included geo-targeting, site re-targeting and something they call "Auto Behaviorally Targeted Segments."
With the targeting scheme implemented by Revenue Science, according to the company, the franchise found a 60 percent increase in unique users who visited its website, a 54 percent increase in site visits overall and a 73 percent increase in page views. Looking at other metrics, none of which were articulated, the franchise was able to attribute a 29 percent increase in email leads. Its biggest life was found in its offline metric. The company realized a 98 percent increase in web-driven phone calls based on behaviorally targeted campaign.

The challenge, however, is defining "success" of behavioral targeting in a way that is appropriate to the tactic.

Karl Siebrecht is senior vice president and general manager, Atlas Enterprise Solutions. He has been with aQuantive since 1999. His career there has included roles across analytics, product management, business strategy and operations. Prior to...

View full biography


to leave comments.