- Describe the similarities and differences between server log, coded page and panel data. Which do you think is the most useful?
Server log – record kept at the server level that records each file request from a website
Coded page – is a technique in which a small image, usually a 1 pixel tag or transparent GIF, is placed on a web page. Used in conjunction with a cookie on the user’s computer, the image returns data about user activity on the Web page.
Coded page > server log data:
More accurate data because the user’s I{ address may change during the session, causing the server to log to, in effect, record different visitors instead of the same visitor. Most browsers are unable to cache web pages, allowing them to be recalled without going back out onto the Internet. This speeds the visitor’s surfing session but it means that multiple requests for the same page in a single session are not recorded in the server logs. The data tags can also be programmed so that they do not recognize automated requests, like shopping bots. It is also possible to program the tags so that they do not return data unless the complete page is viewed.
Data collected by web bugs has smaller storage requirements because it stores only relevant data, not all the site operating data stored by server logs
When data are collected in this manner, it permits real-time processing and viewing of site metrics. Server log data must be transferred to a separate analytic system for batch processing. It can be made available quickly, but not in real time.
Panel data – Process for using panel data to generate site effectiveness: first, recruit a statistically representative panel of Internet users who agree to participate in the data collection
Panel data > server logs
Source of data is unambiguous, which it often is not with server logs
According to server logs, a person using Internet from work and home is two people but planned and maintained panel data can overcome the issue.
The measurement firm can collect demographic and behavioral data from panel households that can be very useful in reporting and analytics. Because server logs identify most users by dynamic IP addresses, it is not possible to use much third-party data to enhance it.
- What are some of the specific metrics that measure Internet traffic, audiences and campaigns? Which ones do you think are most important?
Traffic: hits (number of files requested), impressions (the number of times an ad banner is requested by a browser), page views or deliveries (the number of times a web page is requested), sessions (the amount of activity on a site during a specified period), click through (the number of times visitors come to the site by clicking on an ad)
Audience: visitors (the number of people who visit a site), total includes multiple visits, unique (different people) during a specified time frame, unidentified (anonymous) or identified (registered customer), unduplicated audience (the visitors that are unique to a website), behavior on the site (number of page views, session time, path through site, shopping cart abandonment, entry page)
Campaign: by communication channels like email, mail, online banners, etc, by offer like free shipping vs. 25% off and search effectiveness by keyword
The way to choose the correct measurement metrics out of the plethora is to ask which ones are needed to measure the achievement of the site objectives.
- What kinds of variables are needed to measure the effectiveness of branding efforts on the Internet? Where are these measures obtained?
CPM: cost per thousand advertising impressions, the primary pricing mechanism for many media including Internet
CPA: cost per action; a media pricing mechanism based on the number of visitors who take a specified action, for example, clicking through on an ad or making a purchase
CPC: Cost per click from an ad or referring website
These measures are all obtained via web metrics.