Overlap Matters

Traditional wisdom says that cross-site duplication of online advertising is bad for business. And indeed duplication has traditionally been seen as redundant – a sign of media waste. A report from your third-party ad server would highlight which sites overlapped with which other sites on your buy, and by how much. That way, you could act immediately to pare back the buy and minimize the overlap. But as more digital media research emerges, the industry is learning that overlap is actually a positive thing. As it turns out, customers who see your ads on multiple sites are your best converters.

In other words, your target audience – the people who register on your site, fill out your lead forms and ultimately become your customers – have, for the most part, been reached across multiple sites.

Because overlap boosts the odds of a purchase occurring, we know that a conversion is not simply the result of the last ad clicked. This means that the area of overlap represents a landscape filled with rich resources for marketers. Every engagement a user experiences contributes to the conversion event. And every touch point you share with your consumers contains information that you can use to make better media decisions.


The Atlas Institute analyzed 16 advertisers who tracked their media campaigns with Atlas for the first quarter of 2007. Users were classified as either “exclusive” to indicate they were reached by a single publisher or “overlapped” to indicate they saw ads on two or more publishers. Overlap for users who converted was categorized on the same basis. Primary conversions considered for this study were sales, lead or registration confirmations. The analysis spanned more than 300 million cookies, 5 billion ads served and 1.7 million conversions. As with all research done by the Atlas Institute, the analysis eliminates the bias of cookie deletion by using only stable, long-lived cookies.


On average, 30 percent of users saw ads from multiple publishers, while 53 percent of users saw ads from multiple placements (Figure 1). Within a campaign, overlap among the sites ranged from 8 percent to as high as 60 percent. Placement overlap showed similar variety, ranging from 35 percent to 72 percent overlap per campaign. In addition, increased overlap dramatically drives up frequency. On average, the impressions consumed by site-overlapped cookies were 4.4 times higher than those reached on a single publisher site.


Duplication amongst converters is even more extreme than for impressions. On average, 66.7 percent of users who triggered a primary action tag saw ads from multiple sites (see Figure 2). At the placement level, 86 percent of conversions came from the overlapped group.

Figure 3 illustrates the variability of overlap across the 16 campaigns. Higher-volume campaigns experienced significantly higher conversion overlap than smaller ones. And users reached on multiple publishers accounted for a higher share of conversions – on average representing only a third of the total users reached but two-thirds of conversions. A user reached across multiple publishers was twice as likely to convert as one reached on only a single publisher.

What This Means for Advertisers

It’s critical to understand the drivers of overlap. Not surprisingly, large buys show much greater overlap than small buys. Additionally, buys that advertise heavily on networks or large portals show much higher overlap than those that do not. The degree to which overlap impacts conversions will differ greatly from advertiser to advertiser, since there’s no consistent correlation between reach and conversion overlap. With that in mind, here are our recommendations for using these findings.

Measure the overlap in every campaign. Overlap varies wildly across campaigns. Spend levels, the composition of publishers, placement choice and the usage of ad networks all have a dramatic effect on the amount of user duplication experienced. Overlap should be viewed and weighed within the context of achieving the overall campaign goals.

Maximize overlap for branding. Brand advertisers prefer to surround their target demographic with their messaging and increase brand awareness by maximizing overlap. Identify buys that have high reach on your target demographic and then seek publishers that have high overlap with those buys. The ability to identify the exact amount of duplicated reach and conversions during campaigns provides a powerful negotiating tool with ad networks and traditional publishers.

Manage overlap’s impact on frequency. The Atlas Institute’s Optimal Frequency study proved that increased frequency correlates with diminishing returns for direct response campaigns.[1] Since overlap drives up frequency without the advertiser being aware of it, dropping buys with high overlap and reallocating the dollars to publishers with superior exclusive reach is an easy first step to increasing efficiency. Look for publishers that do a good job exclusively reaching converters.

Keep buys that deliver targeted reach. The ubiquity of overlap among converters highlights the shortcomings of current reporting standards, which attribute 100 percent of the conversion to the last impression or click. Making smart media-planning and creative-design decisions requires an in-depth understanding of a user’s behavior throughout the purchase cycle. Some buys do a great job of reaching your target audience but aren’t credited for conversions because they’re not the last ad seen.

Understanding how your buys are reaching converters may provide important justification for more expensive media buys like rich media, web video and sponsorships.


  1. Chandler-Pepelnjak and Song, “Optimal Frequency: The Impact of User Frequency on Conversion Rates,” The Atlas Institute.

Sponsored Search

Search and navigation are different behaviors, yet search engines are used to doing both. This means a good search engine can handle search and navigation requests. Google’s popularity is in large part due to the fact it is also a good “navigation engine.” Its ability to find the right site when a user can’t recall how to reach the site directly is excellent.
— Danny Sullivan, Noted Search Marketing Speaker and Authority[1]

It turns out that 71 percent of sponsored search clicks are navigational – that is, they come from users who already know what product, service or brand they want; they just don’t know how to get to the right website. So what does that mean to you?

Well, first of all, it indicates the possibility that you’re probably overpaying for branded search. And secondly, it undercuts the theory that the last click before a conversion was the one that drove the sale. It tells you that almost three-quarters of your sponsored search buy is not bringing in new prospects – it’s simply delivering people who are already actively looking for your URL. What’s more, new research by Microsoft’s Atlas Institute shows that more than half of what you pay for sponsored search goes to navigational clicks. Is it worth it?

To understand how navigational search impacts search marketers, we analyzed a large cross-section of paid search traffic, looking specifically for evidence of navigational search activity. Our analysis provides statistics on the pervasiveness of navigational search, as well as insight into navigational behaviors.


For this research, we analyzed click log data for 120,000 unique users occurring between Nov. 1, 2006, and May 1, 2007. The click data we analyzed originated from paid search advertisements purchased through Yahoo Search Marketing, Google AdWords and Microsoft AdCenter. A total of 275,858 paid search clicks were recorded for the set of users during this period. Thirty advertisers were included in the study, each advertiser accounting for 4,000 unique users.

In the discussion that follows, we use the term conversions to describe the primary measurable events advertisers use to gauge the effectiveness of their advertising campaigns. Examples of conversions advertisers might track include order completions, “contact us” form submissions and page views.


In our analysis, we categorized each sponsored search click according to two behavioral dimensions.

Repeat-visit behavior. If a user clicked on multiple ads leading to a given advertiser’s website, we considered each instance after the first to be a repeat visit. We considered repeat visits to be navigational because they imply prior knowledge of the advertiser.

Branded keyword search. A click was assigned to this segment if the key phrase associated with the click included the advertiser’s brand name or explicitly matched the advertiser’s website URL. For example, if a user clicked through to the ACME Corp. website, the following key phrases would be categorized as branded:

  • acme.com
  • www.acme.com
  • acme corporation
  • acme corporation website


The charts that follow summarize our findings. Figure 1 shows the results for each segment as a percentage of total clicks. The percentages given are an average for all advertisers in the study, which means each advertiser had an equal contribution to the overall averages.

Looking at the data, we notice that nearly half (48.3 percent) of clicks were from users who had been to the advertiser’s site before. Well over half (59.6 percent) of clicks also came from branded key phrases. Only 29 percent of clicks were attributed to nonbranded first visits. This relatively small segment is often how search marketing is characterized – as a means of attracting and acquiring new customers.

To quantify the impact of navigational search on overall search budgets, we also looked at the media cost by segment (Figure 2). The 59.6 percent of clicks attributed to branded key phrases represented 34.2 percent of total cost. This shift is to be expected, given the relationship between click-through rate and bid price. Ads with higher click-through rates require lower bids to maintain top positions. Since branded key phrases are directly related to the advertisers’ brand or URL, higher click-through rates and lower cost per click (CPC) rates are expected for these ads.


Navigational search raises some concerns regarding how advertisers measure the performance of their campaigns. Current reporting standards attribute conversions to the last ad click or impression prior to the conversion event. Thus, virtually all campaign reports ignore the fact that consumers are being reached by multiple ads, on multiple sites and across many channels. Since navigational search behavior typically occurs in a user’s immediate path to conversion, navigational search ads often appear to be the source of high volumes of conversions. Put simply, navigational search behavior implies that the user already knows the advertiser. That prior knowledge may have come from a variety of other marketing touch points and interactions, and in many cases, from existing customers who are returning to the site to buy again.

To accurately measure performance, an advertiser should consider all ad exposures prior to a conversion. In support of this point, there is mounting evidence that crediting the last ad or click with the entire conversion is shortsighted. A recent study from the Atlas Institute found that two-thirds of converters were exposed to ads from multiple sites.[2] A separate study found that sponsored search clickers were 22 percent more likely to convert if they were exposed to display ads from the same advertiser.[3] If we take a holistic view of all tracked ad exposures, a conversion should not be attributed to the last ad clicked, but rather shared among all interactions that lead to a conversion. This means that advertisers currently have an inflated view of the value of navigational clicks, and are likely undervaluing non-navigational search clicks and other media types.


Billions of dollars are being spent every year on navigational search clicks, and in all likelihood a significant portion of your search spend counts toward this number. The phenomenon of navigational search isn’t bad for search marketers. It is in fact an indication of the growing reach of search engines. But navigational search behavior should factor in to how you manage your search marketing campaigns. Below are a few tactics to consider when managing paid search.

  • Separate out your branded keywords when looking at clicks, cost and performance. This will help you understand how paid search is contributing to your overall advertising efforts. If branded keywords are driving a lot of sales, it’s likely that other marketing channels are generating interest in your brand. In addition, analyzing repeat-visit behavior will help you optimize your search campaigns. A high incidence of repeat visits is an indicator that a keyword is being used navigationally. This information can be useful in planning natural search optimization and could change how keyword performance is being valued.
  • Test what happens if branded key phrases are bid down or removed altogether. Several of the advertisers in this study were not bidding on any branded key phrases. In most cases, those advertisers ranked at the top of the natural listings for brand-related key phrases, which is often true for advertisers. A well-placed natural listing may be all the navigational searcher needs to click through and convert.
  • Consider all touch points that led to a conversion, not just the last ad. As discussed above, navigational search can mask the impact of other advertising media due to last-click attribution of conversions. To gain an accurate understanding of how different channels contribute to conversions, use one tracking mechanism to measure the performance of all ad types and look beyond the last ad when attributing value.


  1. Danny Sullivan, “Browser-Based Searching,” The ClickZ Network. April 10, 2002. www.clickz.com/1006191
  2. Jed Fowler, Analyst, The Atlas Institute, “The Impact of Overlap on Reach, Frequency &Conversions.” www.atlassolutions.com/uploadedFiles/Atlas/Atlas_ Institute/Published_Content/IODMI.pdf
  3. Esco Strong, Sr. Analyst, The Atlas Institute, “The Combined Impact of Search and Display Advertising. www.atlassolutions.com/uploadedFiles/ Atlas/Atlas_Institute/Published_Content/ crosschanneldmi.pdf