Cookie Corrected Audience Data

On June 23rd, 2008, Quantcast launched a breakthrough in direct-measurement audience modeling: cookie corrected audience data. This development enables directly measured cookie data to be modeled into people-based estimates. Advertisers and marketers have a deep desire to evaluate people-based measures, and this enhancement in our service provides the industry with numerous advancements, including:

• First-ever audience measures for properties or arbitrary collections of content which to date, have been too small to be projected by syndicated panel services
• Improved accuracy in rankings for co-mingled directly-measured (quantified) and estimated (non-quantified) properties in our Quantcast Planner tool
• Visibility into property-specific cookie to people translation ratios

 

Many industry participants have maintained this breakthrough would never be accomplished. We have always disagreed; confident that with access to massive data sets and sophisticated statistical modeling the complexities of cookie to people translations could be overcome.

Quantcast’s audience measurement service helps all industry participants address the complexities of today’s changing media market. Unlike panel measurement services which report a standardized traffic metric for all sites by extrapolating from their sample, Quantcast directly measures traffic for quantified publishers and provides rough estimates for the rest of the web (non-quantified publishers).

Our directly measured and reference data sources complement each other. We combine them through sophisticated statistical modeling techniques, which let us triangulate better estimates than either source generates on its own. However, until now our traffic estimates have been based on two different data sets: cookies for quantified publishers, and people for non-quantified publishers.

There are many components to our model, which you can review in detail by downloading our Cookie Corrected Audience Data White Paper. Key topics covered by this white paper include detail on how we address cookie deletion and non-acceptance, and model for multiple machine use by individuals, and multiple person use of specific machines. Additionally, we have provided a broad FAQ below.

You will notice that generally, quantified publisher traffic counts — now people based — decline versus historically reported cookie counts in our service. This results in increases in other key measures (visits and pages per user, for example). This change should be embraced, as Quantcast now provides the key metrics that advertisers care most about — insights at the person, not cookie level.

We continue to report cookie counts for quantified publishers on the traffic tab. Quantcast is the only measurement service to provide this level of cookie and people based visibility — an important step in helping publishers and marketers better understand the differences between publisher log-file traffic counts (cookie-based) and syndicated audience estimates (people-based).

Non-quantified publishers continue to have people estimates reported in the Quantcast service. We use an ad-hoc statistical approach to build these estimates. Publishers who want more accurate numbers should join the Quantified Publisher program.

We look forward to hearing your feedback to this innovation. We welcome your comments at publisher@quantcast.com.

Cookie Corrected Audience Data FAQ:

Adjusting for Cookie Deletion and Machine Usage

What is a cookie?

A cookie is a small text file that your browser uses to identify you to individual sites. It might be used to automatically log you in to your web-based email, or store personal preferences on your favorite news site. Cookies are also used to aggregate statistics on website activity levels. Quantcast uses cookies as part of its audience measurement process.

Why might the number of cookies and people be different? As text files, cookies may be deleted from time to time. If the cookie for a particular website is deleted between visits, a browser is issued a new cookie upon its return. In such cases, the website will have recorded two cookies for the same individual browser and the resulting cookie count is actually larger than the number of browsers that have visited the site.

Adding complexity to this problem, while many people may use the same browser to access a website, others may use multiple browsers or machines (at home and work, for example) to access the same website. This creates a more complicated environment and is why Quantcast has developed a sophisticated model to deal with cookie deletion and multiple machine usage to produce accurate estimates of the number of people that visit web destinations.

How big an impact do cookie deletion and multiple machine use have?

The impact of cookie deletion and multiple machine use/people use per machine is not uniform and differs significantly between sites. In some cases, though, it is not unusual to see an average of two (or more) cookies for every browser over the course of a month. A variety of factors must be taken into to account to properly adjust property level translations from cookies, to people. Quantcast has issued a Cookie Corrected Audience White Paper which discussed many of the considerations we factor into our model. At a high level, these include time period of evaluation, frequency of visit of user, cookie deletion rate and machine use behavior of a specific properties audience.

How does Quantcast deal with cookie deletion, and translation to people?

At Quantcast we have access to multiple complementary data sources. We have several clickstream data sources which allow us to calibrate our cookie correction models. Our Quantified Publisher program captures over 75 billion media consumption events every month, generated by more than 1.4 billion cookies (data as of June, 2008). What’s more, many of our Quantified Publisher partners share anonymous identifiers with us that are independent of cookies. This mass of data – with different collection processes, biases and issues — allows us to effectively “triangulate” and develop dynamic, and accurate projections. For additional insights into how our model works see our Cookie Corrected Audience White Paper.

Why has Quantcast implemented cookie correction to its Quantified Publisher traffic data? While cookies provide valuable learning related to traffic flow on a site, they do not provide the people-based insights that advertisers demand. We believe facilitating a view of both cookie and people-based intelligence is a critical advancement that will enable addressable advertising.

What data has changed on the site as a result of cookie correction implementation? Prior to our cookie corrected model launch (June 23rd, 2008), quantified publishers had unique cookie data reported as their default traffic count. This data was included on site profiles, and in our recently released Quantcast Planner tool. All quantified publishers now have people counts reported as their default traffic count, across all components of our service.

I rely on Quantcast’s cookie data for business and programming analysis, is it still available?

Absolutely. You can continue to access cookie data on the detailed Traffic Tab. Just select the “cookie” tab on the traffic chart to switch from people to cookie counts.

My position in the Quantcast Planner tool has changed – reported audience counts have decreased and resulted in a lower ranking. Why is this?

Prior to June 23rd, 2008, quantified publisher traffic counts in our Quantcast Planner tool were cookie-based.Non-quantified publishers had, and continue to have, people-based traffic estimates reported. In order to provide the marketplace with the best possible evaluation tool for understanding audiences, we have migrated all quantified publisher traffic counts to a people-based default. As our Cookie Corrected Audience White Paper explains, cookie counts typically overestimate the actual number of people visiting a web property. Primarily for these reasons, your traffic counts and ranks have changed. It is important to note that our new estimates are more comprehensive, and representative of actual people-based data. This should help you in your ad sales, and revenue generation activities.

Why do non-quantified sites only provide people-based data counts, and no option to view cookie counts?

Cookies are directly measured. Quantified publishers, by nature of their active participation in Quantcast’s service, have this data collected and reported. We do not track census level cookie data for non-quantified publishers, and as such, do not provide this data. If you are a publisher, and would like improved audience data to be reported for your property, you can register for our Quantified Publisher program for free here.