Case Study

The Pool Lane One: Making a Splash with Online Video

Helen Katz

Starcom Mediavest Group


This paper presents a large-scale study on the effectiveness of different online video advertising formats. During the process of more than a year, six major U.S. advertisers and seven media companies participated, including Allstate, Applebee’s, BlackBerry, Capital One, Purina, AOL, BBE, CBS, Discovery, Hulu, Microsoft, and Yahoo! Data collection and analysis were carried out in three phases. The results indicate the Ad Selector is the most effective format of online video advertising. Implications of this important finding are discussed.

Keywords: online video, consumer choice, online ad effectiveness


In recent years, media have become time-shifted, digitized, shortened, and readily available for consumers to enjoy whenever and wherever they are. In this world, digitization and consumer empowerment reign supreme. We must embrace change, shift with evolving technology, and find advertising models that work.

In 2008, VivaKi-a Publicis Groupe agency umbrella covering Digitas, Starcom Mediavest Group, and ZenithOptimedia-initiated a consortium to find and develop new and improved models of advertising for individual media platforms, supported by consumers, advertisers, and media companies. In bringing these parties together, the collective goal is to pool resources and uncover consumer insights to align the industry behind the engagement models of the future. In that sense, The Pool is designed to help scale and monetize new platforms.

The industry had demonstrated a need for alignment in the online video space, a platform with significant growth in consumer use but in which advertisers generally kept using their traditional, 30-second, made-for-television ads. The many potential ad models from which advertisers could choose were causing confusion in the industry, which meant a lack of the consensus necessary to drive scale. Due to these factors, online video became the first “swim lane.”

Lane One of The Pool launched on November 13, 2008. One year later, The Pool participants met in Chicago for a final meeting. Six major U.S. advertisers and seven media companies participated, including Allstate, Applebee’s, BlackBerry, Capital One, Purina, AOL, BBE, CBS, Discovery, Hulu, Microsoft, and Yahoo!

Over the course of the year, the participants in Lane One considered 29 ad models, tested 43 advertising executions, and spent more than 8,000 hours with industry peers, as well as 230,000 hours with more than 25 million consumers. At the end of the journey, the winning ad model was the Ad Selector, which enables user choice and empowerment in a way that facilitates the consumer’s experience. It embraces not just the unique interactivity of the Web but also the benefits of traditional brand advertising.


Consider the way the Ad Selector works: When a viewer selects online video content, several (i.e., two or three) ad images appear on the screen, along with instructions that tell the viewer to pick one ad to watch within a specified timeframe. Without a consumer choice, a random (default) ad will run. If the viewer chooses one of the images offered, the ad for that brand will run instead. The remainder of this article describes the research process in greater detail.

The Pool Process

The research process for The Pool involves what has been called a “5-2-1 approach.” All participants gather for a kick-off session, at which the media companies each present their ideas for new and improved ad models for the platform of interest (in this case, online video). After the presentation of each model, the group collectively votes to select the top five that will move forward into consumer testing. The qualitative tests of these five models compare consumer reactions to the benchmark (industry standard), which for Lane One was the Pre-Roll, that is, a standard ad that appears before the content, similar to a televised ad. After the qualitative findings are shared with the group, a second vote identifies the top two ad models, which move into quantitative tests, again with a comparison against the benchmark. After sharing the results of that phase with the full group, another vote determines which of the two ad models should move into the third, and final, research phase, the field trial. The benchmark (Pre-Roll) remains as the point of comparison. This graduated, winnowing process ensures that the winning ad model has proceeded through multiple, diverse, and extensive research considerations before it is launched to the industry.

Phase 1 Research: Qualitative

At the first Pool meeting, attendees reviewed 29 different ad models and voted to choose 5 to be tested qualitatively, in addition to Pre-Roll. The testing involved 18 focus groups of persons aged 15 to 54 years, conducted by the research partner Alternate Routes in three U.S. cities (Philadelphia, Chicago, and Seattle) in January 2009. Groups were divided by age (teen, young adult, and adult) and level of online video usage (low or high). Each session lasted about two hours. Respondents were asked to visit specific Web sites to review each of the ad models prior to attending the workshop, then participated in a moderated discussion of their feelings about and opinions of each ad model. The models tested were as follows:

This first round of research produced six overall findings:

  1. Relevance rules. Personal ad relevance is more important than contextual relevance.
  2. Empower the viewer. A good ad model should help viewers engage on their own terms, giving them choice and control.
  3. Respect for time. Viewers desire efficient experiences when viewing video online; delays are frustrating.
  4. Leverage the Web. A good ad model has the potential to make the ad experience unique to the Web environment.
  5. Generational differences. Older viewers are more conditioned to the television ad model.
  6. Two viewing experiences. Short-form content has different orientations, goals, and needs than does long-form.

The focus groups also revealed that consumers like to see relevant ads and like having the ability to pick what they see. Although this phase involved qualitative research, the consumers were asked, at the end of each session, to rank each of the models on a scale of 1 to 5, with 1 indicating the model they most liked. The top two models in this ranking were the Ad Selector (rated 2.2) and The Interactive (rated 2.5). Pre-Roll was ranked last, with a score of 3.9.

Table 1. Most Likeable Ad Model

Ad Model Most Likeable Rating (scale of 1 to 5)
Ad Selector 2.2
Interactive 2.5
Transitional Skin 2.6
Clickable Video 2.9
Embedded Video 3.6
Pre-Roll 3.9

Phase 2 Research: Quantitative

Next, The Pool participants gathered in February 2009 to review the qualitative findings. At that meeting, the attendees voted to determine which two models to test quantitatively. The group determined that Ad Selector and Interactive would move onto the next phase.

These two models, along with the Pre-Roll, were tested with a nationally representative online panel of 2,378 survey respondents. These surveys proved that the Ad Selector has positive effects on both the consumer and the advertiser. A research partner, Knowledge Networks, recruited people from its panel and asked them to visit a beta Web site to view an online video, which included one of the ad models being tested. Respondents had to meet similar demographic and online usage criteria as in Phase 1. After exposure to the content and advertising, they were linked to an online survey, which evaluated their response to the ads. The results were analyzed at the 95% confidence level. Six publishers and six advertisers participated in this phase of the research. Both long- and short-form content were tested, because the qualitative findings indicated key differences in viewers’ experiences between these different lengths.

The survey covered several important topics, such as consumer responses to online video components, ad awareness, and responses to the advertising in terms of their enjoyment, attentiveness, and relevance. They also indicated their receptivity to the ad model they saw. In addition, category and product usage data were collected, though the small sample sizes limited this analysis.

Across several key factors, the quantitative results confirm Ad Selector as the winning ad model. As Tables 2 and 3 reveal, the percentage of respondents who agree that the Ad Selector offered personal relevance reached 19% for long-form content and 20% for short-form. For Pre-Roll, these figures were 12% and 17%, respectively. Approximately three-quarters of respondents could remember an Ad Selector ad unaided; about one in five felt positively toward the ad sponsor, and one in six said they would consider purchasing the featured product. These results were significantly more positive than those provoked by either Interactive or Pre-Roll models. The Interactive had the most positive response from consumers in terms of its ability to leverage the Web, such that 12% agreed it did so for long-form content and one in four (24%) concurred for short-form content. Yet around one-third of respondents considered Interactive “avoidable” or readily “able to ignore,” the highest such scores of the three models tested in this phase.

The quantitative research also reveals superior results when consumers were able to choose from multiple categories rather than different ads from the same advertiser. That is, in contrast with a scenario in which they could choose from three automotive ads from General Motors or three cat food ads from Purina, consumers gave more positive responses when they could select across categories. This finding makes intuitive sense, because it enhances the odds that each viewer finds relevant ads. A viewer who is not in the market for a new car or is allergic to cats has no interest in any of the choices offered in that specific category.

Table 2. Key Consumer Benefits

  Ad Selector Ad Selector Interactive Interactive Pre-Roll Pre-Roll
  Long Form Short Form Long Form Short Form Long Form Short Form
Offers personal relevance (agree) 19% 20% 7% 14% 12% 17%
Puts viewer in control (agree) 15% 22% 30% 20% 13% 14%
Offers choices (selected ad) 39% 20%        
Visually appealing (agree) 49% 37% 23% 32% 40% 35%
Leverages Web 5% 3% 12% 24% 6% 6%
Engaging 37% 23% 16% 28% 34% 29%


Table 3. Key Advertiser Benefits

  Ad Selector Ad Selector Interactive Interactive Pre-Roll Pre-Roll
  Long Form Short Form Long Form Short Form Long Form Short Form
Unaided recall 75% 73% 34% 68% 58% 70%
Attentiveness (watched closely) 12% 20% 7% 14% 12% 17%
Positive toward sponsor (agree) 22% 16% 9% 15% 17% 19%
Consider products (agree) 16% 16% 9% 13% 11% 13%
Able to ignore, if wanted (agree) 17% 23% 31% 26% 21% 21%
Avoidable 16% 17% 35% 22% 22% 15%

Phase 3 Research: Field Trial

The group gathered again at a third meeting in July 2009 to review the quantitative research results and vote to determine the one model to be tested in a live, in-market field trial. The Ad Selector was confirmed as the winning ad model to move into the final research phase. The goal of the field trial was to determine if the Ad Selector was still the consumer-preferred, more effective online video ad model, compared with Pre-Roll, when it appeared in a live, real-world setting. In addition, the field trial provided guidance with regard to the operational needs of the model, before any promotion of industry-wide adoption of the Ad Selector. Four advertisers and five publishers participated in this final phase. The trial ran in July and August 2009.

The field trial focused on seven key effectiveness measures, involving both behavioral and attitudinal research. The behavioral metrics were gathered passively by Vindico Group, a third-party tracking company active in the broadband marketplace. It was able to capture and measure the environmental elements around each ad dynamically. The attitudinal research by comScore involved the creation of test and control groups. Consumers were recruited to take a survey after visiting one of the participating Web sites. If they had been exposed to the Ad Selector or Pre-Roll ad, they completed a survey to assess the attitudinal branding effectiveness of the related ads. The unexposed control group saw a public service announcement or publisher house ad. Each test cell was compared with the control sample. The effectiveness of the campaigns was determined by statistical comparisons of the difference between the exposed and unexposed groups, with results tested at the 95% confidence level with a two-tailed t-test. The data were weighted by age, gender, and income to match the behavioral post-buy reports of which respondents had visited each site overall. Survey analysis focused on the change in each measured metric for the group exposed to the ad model compared with those not exposed. The field trial research confirmed that the Ad Selector model positively influences all metrics, driven by the act of choosing an ad, which greatly increases consumer responses across the board.

The results are summarized in Table 4. For example, the clickthrough rate for the Ad Selector (percentage of those exposed who clicked through to the advertiser’s Web site) was 106% higher than that for the Pre-Roll. When the ad in the Ad Selector had been actively chosen, the clickthrough rate was an astounding 240% higher. The increase in top-of-mind (unaided) awareness was nearly three times greater for the Ad Selector than for Pre-Roll. When the ad in the Ad Selector model had been chosen by consumers, the increase was more than four times.

Table 4. Field Trial Results

  Pre-Roll Ad Selector Overall Ad Selector Chosen Ad Selector Default Percentage Change Compared with Pre-Roll
  Rate Rate Rate Rate Ad Selector Overall Ad Selector Chosen Ad Selector Default
Completion Rate 84% 88% 93% 86% 5% 11% 2%
Clickthrough Rate 0.93% 1.92% 3.16% 1.92% 106% 240% 106%
  Lift vs. Control Lift vs. Control Lift vs. Control Lift vs. Control Ad Selector Overall Ad Selector Chosen Ad Selector Default
Top-of-Mind Awareness 24% 93% 129% 29% 288% 438% 21%
Total Unaided Awareness 14% 45% 68% 10% 221% 386% -29%
Aided Awareness 4% 9% 16% -4% 125% 300% -200%
Online Ad Recall 29% 113% 133% 70% 290% 359% 141%
Purchase Intent 0% 33% 44% 0% Significant lift* Significant lift* No lift

*The Ad Selector induced a 33% lift in purchase intentions for those who saw it and a 44% lift for those who chose an ad. There was no increase for Pre-Roll. It is not possible to calculate a percentage change, due to the zero base.

Behavioral tracking showed how often viewers actually chose an ad. The choice rates varied according to the amount of time viewers had to make the selection. If given 10-15 seconds, they chose an ad approximately 45% of the time. Overall, the field trial clearly showed that the Ad Selector delivers consistently stronger results than the Pre-Roll across all seven measures.


After swimming across the first lane of The Pool, the consortium has determined the best ad model for online video, one that works for consumers, media companies, and advertisers. The Ad Selector model, tested qualitatively, quantitatively, and in-market, provides consumers with an online video ad experience that is significantly better than the standard Pre-Roll. A proposed move by the industry toward the Ad Selector model will provide advertisers and publishers with a superior and successful ad model from which to derive revenues. This development in turn will help to grow the industry.

The Ad Selector showed early promise, even during the kick-off meeting of Lane One, for two simple reasons: First, it empowers viewers by giving them the choice of which ad to watch. Second, advertisers like it because it allows them to use their creative assets already on hand. Research enhanced the Ad Selector’s initial potential. In focus groups, people commented that it “had the most personal value … could actually watch something that was useful to me” and “if you choose it you’re going to pay more attention to it.”

In quantitative research, the Ad Selector exerted positive impacts in terms of both advertiser and consumer factors, such as recall, attentiveness, and personal relevance. Finally, in the field trial, overall increases occurred across all seven metrics tested, namely, completion rate, clickthrough rate, top-of-mind awareness, total unaided awareness, aided awareness, brand recall, and purchase intent.

The Ad Selector is clearly a win-win-win option for the viewer, the advertiser, and the publisher. As the results from Lane One spread across the industry, it should make a great splash.

About the Author

Helen Katz (Ph.D., University of Illinois Urbana-Champaign) is a Senior Vice President, Director of Research at Starcom Mediavest Group (SMG), where she focuses on advanced video research and research contract negotiations. She joined SMG in May 2001 as a Strategic Research Director at GM Planworks, working for General Motors. Prior to that role, Helen had her own media research consultancy, in which she worked with agencies and research suppliers across the advertising and marketing industry. Previously, Helen spent 18 months at Zenith Media in New York as the VP, Director of Strategic Research, working with clients such as Toyota, Bell Atlantic (Verizon), HSBC, M&M Mars, and General Mills. Helen’s media research career began in 1989 at DDB Needham Chicago, where she worked for 10 years for clients such as McDonald’s, Anheuser-Busch, General Mills, State Farm Insurance, Dial, Clorox, Helene Curtis/Unilever, and Discover Card. She lives in Chicago with her husband and three daughters. She has published three textbooks on advertising and media, the most recent of which is The Media Handbook (Routledge). The fourth edition will be released in June 2010. In 2008, Helen received an Advertising Research Foundation “Great Minds” award for research innovation. E-Mail: [email protected].