Log Files vs. Pixels: What's the Difference?
Over the last year, a number of new tracking and measurement services have been introduced to support this goal. Most offer an opportunity to gain tremendous insight into marketing campaigns, customer value and site activity. However, in considering which service is right for you, it is important to fully understand exactly how your data will be tracked.
To conduct any kind of analysis, measurement services need data. What fundamentally differentiates today's measurement solutions is the way in which they collect the data. Measurement service providers generally can be grouped into two camps: Web log file-based services and pixel-based services.
The log file-based services get data by collecting the Web site log files from your company's Web servers at the end of every day. Pixel-based services use 1x1 pixel images from third-party servers to produce a separate log file on their end. Since many marketers use tracking data to make multimillion dollar optimization decisions, each collection method has ramifications that should be considered.
E-businesses are increasingly under pressure to produce truly accurate or audit-quality data. And measurement service providers are only as good as the quality of the data they track. The reality is that services that use pixels can produce very different numbers from those of Web site log file-based solutions.
Log files contain data about which particular Web pages and images were served by the Web server, while the pixel-based methodology tracks which browsers requested the pixel image embedded in a Web page. The pixel tracking image is just an image; a user's browser requests it only after the HTML page is served. In addition, a pixel-tracking image that is served by a third-party measurement service provider generally loads last after all other images on the page are loaded. Pixel-based tracking fundamentally produces different data than log file-based solutions because it counts different things.
Think of how you use the Internet. Have you ever hit the stop button before the page finished loading? Have you ever clicked on a link in a Web page that is in the process of loading? The pixel tracking system "misses" these event. Hitting the stop button while a page is loading tells the browser to stop contacting servers, including the third-party pixel server. If you move from one page to another before the first page loads completely, the browser does not have the opportunity to request the pixel from the third-party pixel server and produces no tracking data.
Another area in which there can be large discrepancies between the data is when tracking shopping events. For example, have you ever hit the back and forward buttons during a shopping cart check out? Each time you hit the "thank you" page of the check out, the third-party pixel server generally records a transaction, sometimes inflating the number of transactions that actually occurred.
There also is the problem of relying on third-party servers in real time. Have you ever noticed the Netscape "N" or the Internet Explorer "e" still spinning even after the page seems to have loaded? Most often, it is the result of a third-party pixel tracking server being down, and your browser's request for that pixel is actually hanging and waiting for the connection. If the user's browser cannot connect to the server, no data is recorded by the measurement service. Site activity, transactions and registrations go untracked.
Also consider your requirements for integrating multiple data sources. Many companies choose to integrate site usage data with ad servers, transactions, registration databases, legacy systems and other data sources to establish a more complete view of their customers.
With the Web server log file approach, the marketer has the ability to integrate the data contained in the log files with any other source of data that they can get their hands on by matching their users' cookie IDs that are recorded in the log file with the cookie IDs or account IDs contained in their other back-end systems.
In the case of the pixel-based approach, the marketer's Web site must be able to append data from these other sources at the time that the pixel is served so that the data can be passed over the Internet to the third-party pixel server.
This approach requires that all other sources of data to be integrated are available in real time and can instantly process the new data so it can be appended immediately to the pixel for transfer. This log file approach can be difficult for companies, especially if their back-end systems are behind firewalls or require more than a few seconds to process or check the data that users are generating.
The way that third-party measurement service providers approach privacy also is a differentiating factor. The recent uproar over how ad servers track Internet users across different Web sites has caused the government and marketers to look at their privacy policies and the way in which customer data is being used.
Because log file solutions rely on the unique individual cookies contained in marketers' log files, there is no way for these service providers to accumulate data across the Web sites that they serve, ensuring that the marketers themselves are in control of how their data is being used.
Pixel-based solutions act much more like their ad server cousins, setting their own cookie on users' computers when they serve their pixel. This cookie is used to track users on all sites for which the third-party pixel server is working. Most pixel-based services do not currently aggregate user data across sites, but the ability to do so is clearly there, sometimes causing concern for marketers who are protective of their customers' data.
Pixel-based tracking and Web server log file-based tracking both have merits. For marketers looking to start in a matter of days, pixel-based services often can accommodate this. For marketers looking for an accurate and integrated view of their customers and their business performance, log-file tracking can offer a solid foundation for analysis. In either event, there are a number of questions you should be asking before selecting the measurement solution that's right for you.