New frontiers in display advertising planning and measurement
April 12th, 2010 | Published in Google Blog
This is the third post in our series on the future of display advertising. Today, Director of Product Management, Ari Paparo, looks at how better data will help marketers plan and measure their display campaigns in the future - Ed.
Basketball teams in the 1980s looked at fairly simple statistics — points, rebounds, assists and shooting percentages — to measure team and player performance. However, in recent years, there’s been a data renaissance — a recognition of the need to develop more insightful measures, and a resurgence in appreciation for the value of data in sports. Now, professional basketball teams measure all sorts of on-court happenings, as well as more ethereal things like team chemistry and player psychology. As advertisers and agencies try to plan and measure their display ad campaigns, they’re much like basketball teams stuck in the 80's. Today, planning display advertising campaigns is largely based on relationships and habits, and often-primitive measures of website traffic. If asked to quantify the impact of their display ad campaigns, many advertisers could show you the number of clicks on their ads, and then shrug.
The Internet has long held out the promise of being a truly accountable, measurable medium for marketers. In search advertising, a decade of investment in analytics and measurement tools has helped to realize that promise. But the same tools for display advertising have lagged.
In the previous post in this series, Neal Mohan wrote about the creative possibilities that new display advertising technology is enabling. But how do marketers work out where to buy these ads, and quantify their impact? Let’s look at what’s becoming possible as we start to use newer technologies, improved statistical models and aggregated data to improve the planning and measurement of display advertising. Imagine an ad agency tasked with planning and measuring a campaign for a new male cologne (specially endorsed by a famous DJ). The ideal target audience is males aged 18-35 who are interested in dance music, well-groomed and who think they’re hip.
Today, it’s possible (using tools like DoubleClick Ad Planner) to find popular U.S. sites that are read by males aged 18-35 who are interested in dance music or who have previously visited the DJ’s website. Of course, there's no way to tell which sites’ readers are well-groomed or if they’re hip, but media planners can add in terms like "clubs," "nightlife", “sample sale” and "fashion" into Ad Planner’s search term correlator to find sites whose users are more likely to search for those terms, as measured across large quantities of data.
Looking forward, what if the agency could seamlessly click a checkbox to pull in site performance data from that same client’s last ad campaign? The planner could rank the sites in the media plan that produced the best results for the last campaign. And what if the agency could click another checkbox to select recommended high-performing sites in the Google Content Network that offer above the fold placements and that fall within the client’s budget and targeting criteria, then buy them with a click of the button in AdWords?
Just as we’re working to make planning more precise, we’re also focused on evolving display measurement tools. For a long time, display advertisers have used fairly simple measures like clicks, impressions or conversions. These are great metrics for some types of marketing campaigns. But not for all. Not every ad campaign is looking to deliver an immediate sale. Lots of advertising — like the cologne campaign — is designed to influence opinions, spread buzz or build brand associations. For these campaigns, measuring clicks is like trying to judge an entire movie after watching just five minutes.
We’re developing new measurement products designed to gauge the impact of ads on brand awareness or on user interest in the product being advertised. Let’s go back to our cologne example. Today, using our new tool called Campaign Insights, the agency can reliably measure the “brand lift” directly attributable to the display campaign. This measurement tool looks at two large groups of users — one that has seen the ad, and one that hasn’t. It then compares the volume of searches and website visits to measure how awareness of the brand has improved as a result of the display ad campaign.
Think about what other measurement tools may become possible. What if the agency could use an even larger real-time focus group like, say, the entire Internet? It could include social features in the ad, and then, by parsing public reactions — tweets, blogposts, status updates, YouTube comments and more — measure, in real time, how the Internet is responding to the cologne and the ad. This could give them an immediate, quantifiable view into the reactions and views of its potential consumers, and measure the viral effect of the ad over time. And what if the agency could precisely measure the impact of the campaign — not just on increased web traffic, searches or online comment — but (using geographical signals) on the actual purchases of their cologne in local stores? Imagine the possibilities — display ad campaigns could even communicate with the advertiser’s supply chain or inventory system.
These innovations in planning and measurement are all exciting, but what’s most revolutionary is what will happen when they’re combined. In the future, campaign measurement will take place in near real-time, creating an almost immediate feedback loop. Currently, the process is very linear — marketers plan their campaign, then buy ad space, then run their campaign, then measure the results, often with weeks in between. Soon, measurement will become truly dynamic and will feed into the planning process itself. Agencies and advertisers will be able to test multiple creatives and media plans, and immediately tweak them to deliver the best-performing ads and reach the optimal sites and audiences as measurement data starts to come in.
We’re on the cusp of a data renaissance in display ad planning and measurement. It promises to vastly improve online advertising for marketers, while resulting in ads that people find more relevant and effective. And by attracting new advertisers with more valuable ads, it will help online publishers earn more money from their online content.
We think that’s definitely something worth shooting for.
Basketball teams in the 1980s looked at fairly simple statistics — points, rebounds, assists and shooting percentages — to measure team and player performance. However, in recent years, there’s been a data renaissance — a recognition of the need to develop more insightful measures, and a resurgence in appreciation for the value of data in sports. Now, professional basketball teams measure all sorts of on-court happenings, as well as more ethereal things like team chemistry and player psychology. As advertisers and agencies try to plan and measure their display ad campaigns, they’re much like basketball teams stuck in the 80's. Today, planning display advertising campaigns is largely based on relationships and habits, and often-primitive measures of website traffic. If asked to quantify the impact of their display ad campaigns, many advertisers could show you the number of clicks on their ads, and then shrug.
The Internet has long held out the promise of being a truly accountable, measurable medium for marketers. In search advertising, a decade of investment in analytics and measurement tools has helped to realize that promise. But the same tools for display advertising have lagged.
In the previous post in this series, Neal Mohan wrote about the creative possibilities that new display advertising technology is enabling. But how do marketers work out where to buy these ads, and quantify their impact? Let’s look at what’s becoming possible as we start to use newer technologies, improved statistical models and aggregated data to improve the planning and measurement of display advertising. Imagine an ad agency tasked with planning and measuring a campaign for a new male cologne (specially endorsed by a famous DJ). The ideal target audience is males aged 18-35 who are interested in dance music, well-groomed and who think they’re hip.
Today, it’s possible (using tools like DoubleClick Ad Planner) to find popular U.S. sites that are read by males aged 18-35 who are interested in dance music or who have previously visited the DJ’s website. Of course, there's no way to tell which sites’ readers are well-groomed or if they’re hip, but media planners can add in terms like "clubs," "nightlife", “sample sale” and "fashion" into Ad Planner’s search term correlator to find sites whose users are more likely to search for those terms, as measured across large quantities of data.
Looking forward, what if the agency could seamlessly click a checkbox to pull in site performance data from that same client’s last ad campaign? The planner could rank the sites in the media plan that produced the best results for the last campaign. And what if the agency could click another checkbox to select recommended high-performing sites in the Google Content Network that offer above the fold placements and that fall within the client’s budget and targeting criteria, then buy them with a click of the button in AdWords?
Just as we’re working to make planning more precise, we’re also focused on evolving display measurement tools. For a long time, display advertisers have used fairly simple measures like clicks, impressions or conversions. These are great metrics for some types of marketing campaigns. But not for all. Not every ad campaign is looking to deliver an immediate sale. Lots of advertising — like the cologne campaign — is designed to influence opinions, spread buzz or build brand associations. For these campaigns, measuring clicks is like trying to judge an entire movie after watching just five minutes.
We’re developing new measurement products designed to gauge the impact of ads on brand awareness or on user interest in the product being advertised. Let’s go back to our cologne example. Today, using our new tool called Campaign Insights, the agency can reliably measure the “brand lift” directly attributable to the display campaign. This measurement tool looks at two large groups of users — one that has seen the ad, and one that hasn’t. It then compares the volume of searches and website visits to measure how awareness of the brand has improved as a result of the display ad campaign.
Think about what other measurement tools may become possible. What if the agency could use an even larger real-time focus group like, say, the entire Internet? It could include social features in the ad, and then, by parsing public reactions — tweets, blogposts, status updates, YouTube comments and more — measure, in real time, how the Internet is responding to the cologne and the ad. This could give them an immediate, quantifiable view into the reactions and views of its potential consumers, and measure the viral effect of the ad over time. And what if the agency could precisely measure the impact of the campaign — not just on increased web traffic, searches or online comment — but (using geographical signals) on the actual purchases of their cologne in local stores? Imagine the possibilities — display ad campaigns could even communicate with the advertiser’s supply chain or inventory system.
These innovations in planning and measurement are all exciting, but what’s most revolutionary is what will happen when they’re combined. In the future, campaign measurement will take place in near real-time, creating an almost immediate feedback loop. Currently, the process is very linear — marketers plan their campaign, then buy ad space, then run their campaign, then measure the results, often with weeks in between. Soon, measurement will become truly dynamic and will feed into the planning process itself. Agencies and advertisers will be able to test multiple creatives and media plans, and immediately tweak them to deliver the best-performing ads and reach the optimal sites and audiences as measurement data starts to come in.
We’re on the cusp of a data renaissance in display ad planning and measurement. It promises to vastly improve online advertising for marketers, while resulting in ads that people find more relevant and effective. And by attracting new advertisers with more valuable ads, it will help online publishers earn more money from their online content.
We think that’s definitely something worth shooting for.