Site Server Express - The Report Writer

Once you've imported log files into the database, you're ready to produce analysis reports about the activity on your Internet sites.

Usage Import reconstructs actual requests and visits from users and organizations interacting with your Internet sites. The Report Writer enables you to analyze and cross-reference several properties of those requests, visits, users, and organizations. Its graphical interface provides you with a powerful and flexible tool for understanding activity on your sites.

The Report Writer Catalog

When you start the Report Writer, you are presented with a sequence of windows that facilitate the creation of your reports. The first window gives you the option of choosing an analysis from a catalog of standard reports, creating your own from scratch, or selecting one you have previously created.

 

The file window displays the last four analysis reports used. You can access others by double-clicking More files and browsing your file hierarchy. Initially you'll want to choose one of the standard reports; as you gain proficiency with the Report Writer, you'll be able to devise your own quite rapidly.

The Report Writer Catalog contains more than 25 standard reports. They are organized in a tree hierarchy, according to the contents of the Catalog subdirectory of the folder where the Report Writer is installed. Once you've created analysis reports of your own, if you save them to the Catalog subdirectory, they'll also be available from the catalog window. The standard reports are grouped as Summary reports, Detail reports and Audit reports.

Highlight a report to view a brief description of its characteristics, as summarized in the table below.

Analysis file

Description

Executive summary

This report shows top level information to give you an at-a-glance look at the activity on your web site. For detailed information on a particular section of this report, refer to the corresponding detail report. This report requires that IP resolution, Title lookups, and Whois queries be completed before analysis.

Bandwidth report

This report shows byte transfers on an hourly, daily, and weekly basis. It identifies trends over time as well as averages per day of week and hour of day. Use this information to plan maintenance or upgrades, or alert you to the need for additional capacity.

Browser and operating system report

This report shows browser market share, trends in Microsoft and Netscape versions and security support, and user operating systems. Use this information to design your site for proper user experience. This report requires user agent data within your server log files.

Content report

This report shows the number and distribution of media type categories (documents, applications, and scripts) and media types (HTML, pdf, and CGI scripts) throughout the directory structure of your web site. Use this information to manage the content on your sites.

Geography report

This report shows the top cities, states/provinces, and regions in the U.S. and Canada as well as the international countries visiting your site. Use this information to target your efforts towards the origin of your visitors, and to determine where to mirror your sites. This report requires that IP resolution and Whois queries be completed before analysis.

Hits report

This report shows server hits on an hourly, daily, and weekly basis. Learn about trends over time as well as averages per day of week and hour of day. Use this information to plan maintenance or upgrades, or alert you to the need for additional capacity.

Organization report

This report shows the U.S., Canadian, and international organizations that visit your web site. Use this information to monitor your target audience, identify new target customers, and generate leads from frequently visiting organizations. This report requires that IP resolution and Whois queries be completed before analysis.

Path report

This report shows the sequence of requests that users make when visiting your site. Use this information to assure proper location of key content and to optimize the navigation within your site. This report requires that Title lookups be completed before analysis.

Referrer report

This report shows the top external organization names and URLs that users linked from to reach your site. Use this information to evaluate the effectiveness of online advertising or promotions, or to identify synergistic locations for such programs. This report requires referrer data within your server log files, and IP resolution must be completed before analysis.

Request report

This report shows the most and least requested documents over time and by directory. Learn which elements in your site attract the most attention from users. This report requires that Title lookups be completed before analysis.

User report

This report shows trends in first time and total user visits, user visit frequency, and in unregistered and registered users. Frequent user visits signal that you have created a compelling site. This report requires registration on your site or cookie data within your server log files.

Visits detail report

This report shows when your users visit, on an hourly, daily, and weekly basis. Learn about trends over time as well as averages by day of the week and hour of the day. Use this information to understand the behavior of your customers.

Comprehensive analysis

This report shows detailed information to give you an in-depth look at the activity on your web site. This report requires that IP resolution, Title lookups, and Whois queries be completed before analysis. It includes, organization, geography, request, visit, referrer, browser and bandwidth summaries, as well as an overview.

Audit report (I/Pro format)

The format and content of this report reflects that of an I/Pro I/Audit report. It only contains request information on HTML pages. If your site generates HTML pages from scripts or similar means, you should amend the root filter expression to capture this page activity. You should also perform IP resolution from the Import module for best results.

Audit report (ABVS compliant)

This report is compliant with ABVS web site activity reporting standards. It only contains request information on HTML pages (that is, the root filter contains the expression media=html). Requests for HTML pages are defined as Page impressions. You should perform IP resolution from Usage Import for best results.

Audit report (BPA compliant)

This report is compliant with BPA web site activity reporting standards. It only contains request information on HTML pages (that is, the rootfilter contains the expression media=html). Requests for HTML pages are defined as page requests. You should perform IP resolution from Usage Import for best results.

Creating a Standard Report

To choose a particular report, highlight it and double-click, or highlight and click Next on the Analysis catalog window.

Note For the present purposes of introduction, you only need to select an analysis report. The three screens thereafter can be left without selections.

If you have configured multiple Internet sites, the next window determines which sites are to be analyzed. If you've configured only one Internet site, then you won't see this window. The selection box will list all sites for the current database configured in the Server manager of Usage Import (default is all of your Internet sites).

Selecting an aggregate of all your Internet sites treats all data you've imported as if it came from one site. Within the selection box at the bottom of the window, you can select any combination of your Internet sites. All data from sites you select are "aggregated" together and treated as one data set.

The window that follows allows you to specify a date range for your analysis (default is every request).

If you haven't imported any log files for the time periods specified in the middle of this window, those options will be grayed out.

The final window of the opening sequence prompts you to enter a filter to delimit the data you want to analyze. You can filter your analysis using any of the database properties. (See "Analysis report properties" in Chapter 5.) If you specified sites or dates in the previous windows, they will appear as Boolean filter expressions here.

When you have completed the filter window, click Finish to bring up the analysis you chose from the Report Writer catalog.

Note Clicking Finish at any point before the end of the sequence of four windows will bring up the analysis you have selected with the settings you have specified to this point, and with default settings for any windows which you have skipped or left unchanged.

Analysis Content

Once you have completed the sequence of windows for a catalog analysis, you will arrive at the analysis content window (see figure below). The analysis used here as an illustration is the Request Pattern Report.

 

In the Report Content window, the various features of the analysis are represented by a graphical tree, which you navigate by clicking branches and folders to open or close sub-elements. At the highest level is a filter corresponding to the specific analysis report (represented by a blue filter-cone icon); beneath that are individual calculations (a calculator icon) and their component dimensions (a cube icon), measures (a sphere), and presentations (either a graph or a table). For more elaborate reports, there can be a number of sub-filters, which correspond to subsections of the analysis report. At this point, all of the various report properties, including those set in the four analysis setup windows, can be edited by right-clicking the individual icons. Details of the report elements and their properties are discussed below.

Creating a Report

To create a report, you don't need to do any editing; just click the green Start button in the toolbar, which will bring up the document managing window. Use this window to select the type of report file you want and decide where you want to save it.

The file formats supported for reports are HTML, MS Word, MS Excel, and fixed-width text ready for email distribution.

Note If you do not have Word or Excel installed, you won't be able to save reports in those formats, since Report Writer depends on having the applications running in order to create the actual report. i

Analysis Statistics

Statistics for the completed analysis are displayed in the Analysis Statistics window. These statistics are valuable in determining how individual calculations work and how long they take.

If certain calculations take too long, consider using a down-sampling factor (see "Filters: Down sampling" in Chapter 6) to reduce complexity, or deleting them from future analyses.

 

Note Report Writer uses query results caching between calculations; therefore, the number of queries required to complete a calculation may vary for a single report.

The completed report will be opened and displayed by the appropriate program for the type of file you selected. All of the reports can be edited and saved in the usual way once the Report Writer has created them. The figure below illustrates the output for the Microsoft Word Request pattern report.

Properties

The actual calculations made by the Report Writer in producing your reports depend upon a detailed structure of data properties maintained in your database. The standard report composed above made use of them, but they are also available to you directly, so that you can tailor your analyses specifically to your sites and interests.

Understanding Your Data

You manipulate the details of your analysis by means of a series of filters that operate on data properties. Typical reports might sort all activity on the site which originates from non-academic hosts, for example, or the average number of requests by the second user from each organization versus the average number of requests by the first user from each organization. The following tables summarize all of the data properties which can be filtered and calculated.

Request Properties

Each request to your Internet sites has the properties shown in the table below.

Property name

Property description

Example values

Date

The date of the request.

October 4, 1958, 8/29/95

Timestamp

The time and date of the request.

10/4/58 1:22 PM

Hour

The hour of the request.

0-23

Weekday

The day of the week.

0-6, Sunday, Monday

Week

The week of the year.

1-52

Month

The month of the year.

1-12, Jan, January

Quarter

The quarter of the year.

1-4

Year

The year.

1995

Filename
("File")

The requested file (URI of the URL).

/index.html, /order*, *.html

Title

The <TITLE> of the requested file, if media = html, otherwise title is blank.

"Home page"

Level 1 directory

The first directory from root ("/") that the request file is in. The root directory is also considered a level 1 directory.

/a

Level 2 directory

The second directory from root ("/") that the request file is in.

/a/b/

Level 3 directory

The third directory from root ("/") that the request file is in.

/a/b/c/

Level 4 directory

The fourth directory from root ("/") that the request file is in.

/a/b/c/d/

Level 5 directory

The fifth directory from root ("/") that the request file is in.

/a/b/c/d/e/

Level 6 directory

The sixth directory from root ("/") that the request file is in.

/a/b/c/d/e/f/

Media type
("Media")

The media type of the request file.

html, gif, jpeg, au, wav

Media type category
("MediaCat")

The category of the media type.

Archive, audio, document, image

Visit Properties

Each visit to your Internet sites has the properties shown in the table below.

Property name

Property description

Example criteria values

Site title
("Site")

The title of the Internet site as specified in the Server manager of the Import module.

Usage Analyst marketing web site

Site ID
("SiteId")

A unique number Usage Analyst assigns to each of your Internet sites.

1,2,3, etc.

Browser product
("Browser")

The browser used to view the site.

Microsoft Internet Explorer, Netscape Navigator

Browser product version
("Version")

The version of the browser used to view the site.

1.1, 1.1N, 2.0

Browser product security
("Security")

The level of encryption security that the browser supports.

Exportable SSL (40 bit RC4 encryption)

Browser operating system
("Os")

Operating system of the visitor's computer.

Windows (32 bit),
UNIX - Sun

Referring URL
("RefUrl")

The referrer URL of the first request of the visit.

https://www.yahoo.com/awesome_sites.html

Referring organization type
("RefOrgtype")

The type of the organization whose site the referring URL is from. See Orgtype.

See orgtype

Referring organization name
("RefOrgname")

The name of the organization whose site the referring URL is from. See Orgname.

Your Corporation

Referring organization domain
("RefDomain")

The domain of the organization whose site the referring URL is from. See Domain.

yourcompany.com

Referring organization continent
("RefContinent")

The continent of the organization whose site the referring URL is from.

See continent

Referring organization country/region
("RefCountry")

The country/region of the organization whose site the referring URL is from.

See country/region

Referring organization region
("RefRegion")

The region of the organization whose site the referring URL is from.

See region

Referring organization state
("RefState")

The state of the organization whose site the referring URL is from.

See state

Referring organization city
("RefCity")

The city of the organization whose site the referring URL is from.

See city

Referring organization zip code
("RefZip")

The zip code of the organization whose site the referring URL is from.

See zip

User Properties

Each user of your Internet sites has the properties shown in the table below.

Property name

Property description

Example criteria values

Cookie

The persistent client-state HTTP cookie used to identify the user.

INTERSÉ_!aQFG@4gd

Username

The user name used by this user to register at the site.

Bob

Hostname

The Internet host name address used by this user.

gateway.interse.com

Organization Properties

Each organization with users of your Internet sites has the properties shown in the table below. In order to obtain this information for your database, you must first perform IP resolution and Whois lookup. (See Chapter 5.)

Property name

Property description

Example criteria values

Organization type
("Orgtype")

The type of organization as determined by parsing the Internet host name.

Commercial, academic, non-profit, Internet administrator, government, military, online service

Organization name
("Orgname")

The name of the organization that registered the domain of the host name.

Any Corporation

Organization domain
("Domain")

The part of the Internet host name that uniquely identifies an organization.

Anycompany.com

Continent

The continent where the country/region of the organization is located. The country/region is determined by parsing the Internet host name.

Asia, Europe, Africa, North America, Latin America, South America, Oceania

Country/Region

The country/region where the organization is located. The country/region is determined by parsing the Internet host name.

USA, France, Germany, Canada

Region

The geographical region of the United States of the organization that registered the domain of the host name.

New England, Mid-Atlantic, East North Central, West North Central

State

The geographical state of the United States of the organization that registered the domain of the host name.

Full state name or its postal abbreviation (California or CA)

City

The geographical city of the United States of the organization that registered the domain of the host name.

Sunnyvale, Palo Alto, New York, Los Angeles, and so on

Zip

The geographical Zip code of the United States of the organization, or postal code of the Canadian organization that registered the domain of the host name.

94086, 94087

The Data Hierarchy

Site Server Express Analysis organizes your data within an object-oriented data structure, so you can conduct complex analyses on trends within your Internet sites. With any request, visit, user, or organization property, you can create a calculation Filter or Group. Understanding this data structure is extremely important to making optimal use of the Report Writer.

As mentioned earlier, the data structure has four distinct objects:

  • an organization

  • a user

  • a visit

  • a request

The figure below illustrates the relationship between these objects.

Each request is assigned to a distinct visit; each visit is assigned to a distinct user; and each user is assigned to a distinct organization. Consider the complete picture of the traffic to your site as the world of major league sports. You could say that requests are like a series of games played by a professional athlete. Each game (request) is part of a season or series (visit) and each athlete (user) belongs to a distinct team (organization).

Note that each organization, user, and visit object in the hierarchy has three children: a "first," an "any," and a "last." Specifically,

All visits have three request objects associated with them:

  • The first request of that visit

  • Any requests of that visit

  • The last request of that visit

All users have three visits objects associated with them:

  • The first visit from that user

  • Any visit from that user

  • The last visit of that user

All organizations have three user objects associated with them:

  • The first user to access the site from that organization

  • Any user from that organization

  • The last user to access the site from that organization

Every property has a path from a root object; where root objects are the requests, visits, users, and organizations that can be referred to as part of "any request, by any visit, by any user, from any organization." Properties along this any-any-any-any property path don't need a prefix when being referenced within a Filter. For example, if you're just interested in requests during January, then stating Date < 2/1/95 works.

All other properties must be referenced by a path from a root object. For example, if you want to reference the first visits of all of your users, just prefix the visit properties by "User, first visit, property name goes here."

Although it's possible to refer to a property by its full path in the data hierarchy ("Organization, first user, first visit, first request, date," for example) there are numerous shortcuts, as shown in the table below

Full path

Shortcut

Visit, first request

VFR_

Visit, first request, next request

VFRNR_

Visit, first request, next request, next request

VFRNRNR_

Visit, first request, next request, next request, next request

VFRNRNRNR_

Visit, first request, next request, next request, next request, next request

VFRNRNRNRNR_

Visit, any request

VAR_

Visit, any request, previous request

VARPR_

Visit, any request, previous request, previous request

VARPRPR_

Visit, any request, previous request, previous request, previous request

VARPRPRPR_

Visit, any request, previous request, previous request, previous request, previous request

VARPRPRPRPR_

Visit, any request, next request

VARNR_

Visit, any request, next request, next request

VARNRNR_

Visit, any request, next request, next request, next request

VARNRNRNR_

Visit, any request, next request, next request, next request, next request

VARNRNRNRNR_

Visit, last request

VLR_

Visit, last request, previous request

VLRPR_

Visit, last request, previous request, previous request

VLRPRPR_

Visit, last request, previous request, previous request, previous request

VLRPRPRPR_

Visit, last request, previous request, previous request, previous request, previous request

VLRPRPRPRPR_

User, first visit, first request

UFVFR_

User, first visit, any request

UFVAR_

User, first visit, last request

UFVLR_

User, any visit, first request

UAVFR_

User, any visit, any request

UAVAR_

User, any visit, last request

UAVLR_

User, last visit, first request

ULVFR_

User, last visit, any request

ULVAR_

User, last visit, last request

ULVLR_

Organization, first user, first visit, first request

OFUFVFR_

Organization, first user, first visit, any request

OFUFVAR_

Organization, first user, first visit, last request

OFUFVLR_

Organization, first user, any visit, first request

OFUAVFR_

Organization, first user, any visit, any request

OFUAVAR_

Organization, first user, any visit, last request

OFUAVLR_

Organization, first user, last visit, first request

OFULVFR_

Organization, first user, last visit, any request

OFULVAR_

Organization, first user, last visit, last request

OFULVLR_

Organization, any user, first visit, first request

OAUFVFR_

Organization, any user, first visit, any request

OAUFVAR_

Organization, any user, first visit, last request

OAUFVLR_

Organization, any user, any visit, first request

OAUAVFR_

Organization, any user, any visit, any request

OAUAVAR_

Organization, any user, any visit, last request

OAUAVLR_

Organization, any user, last visit, first request

OAULVFR_

Organization, any user, last visit, any request

OAULVAR_

Organization, any user, last visit, last request

OAULVLR_

Organization, last user, first visit, first request

OLUFVFR_

Organization, last user, first visit, any request

OLUFVAR_

Organization, last user, first visit, last request

OLUFVLR_

Organization, last user, last visit, first request

OLULVFR_

Organization, last user, last visit, any request

OLULVAR_

Organization, last user, last visit, last request

OLULVLR_

Time

Only requests have a specific date and time associated with them. When you want to restrict a calculation to visits, users, or organizations active during a specific time period, you must use a specific request object, with a complete path from the visit, user, or organization object you're analyzing; for example, "UFVAR_Date < 2/1/95", filters out all users except those whose first visit was before February 1, 1995.

Filtering Data

The Report Writer provides for filtering of data at the section level, the calculation level, and the measure level of the analysis hierarchy. This allows you to look at highly specific subsets of data. For example:

  • For only last week or this week

  • For all files except audio

  • For all users in the U.S. using Microsoft Explorer

  • For all users in France using Netscape Navigator

  • For all users in Germany using NCSA Mosaic

Criteria Syntax

To provide the maximum flexibility, specify criteria using Boolean expressions of name/value pairs:

<name><operator><value> [<conjunction> <name><operator><value>][…]

Criteria Names

Criteria names are based on the properties of requests, visits, users, and organizations. Every property listed in the tables at the beginning of the Analysis module section can be used within a filter expression. Unless otherwise specified, the criteria name is the property name. For example,

  • Weekday=Wednesday

  • RefOrg=InfoSeek*

Criteria Values

Different criteria values are appropriate for different properties. In general, a criteria value is either a date, an integer, a string, or a relative criteria expression. The property charts within the previous section provide example values for each property name. In general, values follow these rules:

  • Dates can either be in the format x0:00 AM/PM or 00:00 (for example, 3:30 p.m. or 15:30)

  • Integers can't have commas

  • Strings are case insensitive

  • Strings containing spaces must be enclosed with quotes, otherwise quotes are optional

  • Carriage returns are ignored within the filter expressions

Relative Filter Values

Relative criteria expressions are useful if you want to create reusable files. For criteria names and expressions, see the following table.

Criteria name

Relative criteria values

Year

ThisYear, LastYear

Quarter

ThisQuarter, LastQuarter

Month

ThisMonth, LastMonth

Week

ThisWeek, LastWeek

Day

Today, Yesterday

For example, if you want to run a report every week that analyzes data for just the current week's traffic, rather than constantly editing a filter to include traffic between the date seven days ago and today's date, you can simply add the criterion:

Week = ThisWeek

Any of the relative values can have "[-+]#" appended to indicated a time period previous to the relative time period, measured in increments of that time period. For example, "ThisWeek-1" is the same as "LastWeek", "ThisWeek-2" is the week before last week, "Today-3" is the day before the day before yesterday, and "Month-2" is two months ago.

Wildcard Criteria Values

Criteria values may include wildcard characters as shown in the table below.

Character

Usage

Example

*

Matches any number of characters. It can be used as the first or last character in the character string.

d* matches Debra and David

?

Matches any single alphabetic character.

jo?n matches John and Joan

[]

Matches any single character within the brackets.

[jt]erry matches Jerry and Terry, but not Kerry

!

Matches any character not in the brackets.

[!jt]erry matches Kerry, but not Jerry or Terry

-

Matches any one of a range of characters. You must specify the range in ascending order (A to Z, not Z to A).

[c-e]d matches cd, dd, and Ed

Note When searching for *, ?, [, or - you must enclose the item you're searching for in brackets (cgi[-]bin matches cgi-bin, for example.)

Criteria Operators

Criteria operators define the relationship between the name and value. Valid operators are:

=

when name equals value, i.e., Country=USA

<>

when name does not equal value, i.e., Country<>USA

<

when name is less than value, i.e., Hour < 12

>

when name is greater than value, i.e., Hour > 12

>=

when name is greater than, or equal to, value, i.e., Hour >=12

<=

when name is less than, or equal to, value, i.e., Hour<=12

Criteria Conjunctions

Using criteria conjunctions and parenthesis, multiple name/value pairs may be strung together to create more complex criteria. Valid conjunctions are:

and

when both expressions must be true for the criteria to be true

or

when either expression can be true for the criteria to be true

Example Data Filters

For examples of filters, see the table below.

Filter goal

Filter criteria

Only include files from the product directory

file = /product/*

Exclude files from the product directory

file <> /product/*

Include only the products, but not widget A

(file = /product/*) and (file<> product/widgetA*)

Only analyze requests for HTML files

media = HTML

Only analyze visits using an IE browser

product = Internet Explorer*

Analyze all traffic on December 25, 1995

Date = 12/25/95

Analyze all traffic from cities that start with San in the state of New Mexico

city = san* and state = NM

Analyze traffic for the previous month

month = LastMonth

Sections

The top level of the hierarchy in the analysis tree is the Section, represented by the funnel icon. A section may contain sub-sections as well as calculations. They may also contain a filter expression which is applied to all sections and calculations within that section.

Section Properties: The Filter Tab

 

 

Sections can be filtered on all of the database properties.

Down-Sampling

Down-sampling works by performing analysis operations on a reduced portion of the data from your database. Depending on the factor you select, the analysis module will examine only those visits beginning in the first fraction of an hour. The integrity of the information is preserved by looking at entire visits (as opposed to imposing an arbitrary time cut-off, which would produce large numbers of open visits). In order to compensate for the artificially low numbers generated, the analysis module then multiplies all down-sampled results by the reciprocal of the down-sample factor. This effectively says, "Let the visits beginning in the first 1/x of each hour be considered typical, and multiply the resulting request-data by x, to approximate additional visits beginning in the remainder of the hour."

Note is an operation performed during analysis only, and doesn't otherwise affect the information stored in your database.

 

The Report Writer supports down-sampling of your data, which reduces effective data-size to as little as 1/3600th of the original. This can greatly speed the process of analyzing your data. The down-sampling tab allows you to specify the down-sampling factor for an analysis.

Down-sampling accepts integers from 1 to 3600, with the following results:

  • Factor = 1/ 1 (default)

    Your data is not affected by down-sampling. Results are unmodified. 

  • Factor = 1/ 2 to 1/ 60

    All visits that begin within the first (factor x 60) minutes of the hour are analyzed, and all results are multiplied by (1/factor) to compensate. 

  • Factor = 1/ 61 to 1/ 3600

    All visits that begin within the first (factor x 3600 ) seconds of the hour are analyzed, and all results are multiplied by (1/factor) to compensate. 

Down-sampling reduces your computational time in direct proportion to the down-sampling factor. However, keep in mind the results of down-sampled calculations are approximations.

Section Properties: The Advanced Options Tab

The filter expression within a section can be treated in several different ways, represented by the advanced tab of the section form.

 

Persistent Caching

When the Persistent caching option is checked, the Report Writer creates a set of temporary tables in the database containing the results of the filter entered in the Filter tab. These tables are then used for any calculations that need the filter. This process can greatly speed up reports than contain many calculations using the same filter, especially if the filter eliminates a large portion of the overall data set. For example, if your report is meant to display data for last week, and there is a year's worth of data in your database, using a persistently cached filter of "week = lastweek" in the root section will greatly speed up the report. The temporary tables will continue to exist in your database and will be referenced by any future reports with compatible filters. These tables will be removed once the date defined in the expiration box is reached, or if new data is imported which invalidates them.

The Confine 'first' and 'last' dimensions to this filter checkbox allows you to define how this filter will treat first and last dimensions. If this box is checked, then any first and last dimensions will be limited to the data in the temporary tables. If not checked, the first and last dimensions are absolute over the entire data set within the database. For example, if the box is checked, a "User First Visit" property on a "week = lastweek" filter will refer to the user's first visit that took place last week. However, if you wanted to count the number of new users visiting your site last week, you would want to count the number of "User First Visits" without the checkbox checked. That way, a user's first visit would only be counted if it was the first visit for that user over the entire period contained in your database.

Time Averaging

The time averaging option is by default set to Fast, but Sometimes Approximate. Consider using Accurate, but Sometimes Slow, if the section contains a measure or dimension that is averaged over a time period (average requests per hour, for example). As a rule of thumb, the fast time averaging should be used unless the filter removes chunks of time within the endpoints of the filter time period. For example, a filter that removes weekends or non-work hours should probably use the accurate time averaging.

Section Properties: The Description Tab

The description tab allows you to create a text description that will be contained in the report just under the section header. Use this option to create clear and readable reports.

Note You must select the option to display the description on the Header presentation tab (by default section descriptions are not displayed).

 

Section Properties: The Header Presentation Tab

The Header Presentation tab allows you to customize the look and feel of the section header and description report elements. The Filter name bar in the upper window is selected by default. To change your selection, simply click anywhere on the filter description bar.

Each element has several customization options: text properties, table cell properties, and table gridlines. In the text properties box, you can control various font elements, including style, size, and color. In the table cell properties, you determine if you want to display the filter name or description and the color of the display banner. Clicking the Visible box in the table gridlines box causes gridlines to be displayed. Click OK after you have made your selections or changes in this tab.

 

Calculations

The calculation is the basic building block of the reports. One calculation represents a single slice of the data contained in the database. Each calculation can be viewed as a table or a graph. Calculations are made up of row dimensions, column dimensions, measures, and presentations.

Calculation Properties: The Filter Tab

A calculation can have its own filter expression applied to it, in addition to any filters contained in its parent sections.

 

The parent filter section of the filter tab displays the combined filters of all the sections above this calculation. Under this is the current filter box, in which the filter expression for this calculation can be entered.

Note The advanced options available for section filters are not available for calculation filters.

To add a calculation filter, you have two options. You can type the filter expression into the window of the current filter box. Or, you can drag and drop filter terms and operators from the Reference terms section found at the bottom of the filter tab. Open the drop down list in that section to display the available filter terms. Select the one you want by clicking it, then dragging it into the current filter window. You will also find the supported Boolean operators (and, or, >, etc.) at the bottom of this list. Finally, you'll need to type in the filter argument (ThisWeek, 2/25/97, for example).

 

As in the Section property box, you can add a description that clearly explains the components of the calculation, its filters, what the results should be, and how they should be reviewed.

 

As in the Section property box, you can edit the calculation name and description style and appearance.

Dimensions

Dimensions are the properties that are used to divide up the data into an understandable table or graph format. Both row and column dimensions can be defined. The data that is output for each calculation will be based on the row and column dimensions defined for that calculation. For example, a calculation with a row dimension of country/region and a column dimension of week will create a table with the various countries along the side, and the various weeks along the top. The intersecting cells will contain the data for each country/region during each week.

Note Including a row or column dimension is not required to define a calculation.

Dimension properties: The Basis Tab

All of the Analysis Report properties can be used as dimensions. (See "Analysis report properties.") All of these properties can be found in the Basis tab's tree divided up by category (requests, advertising requests, visits, users, and organizations). Any of the Measure properties can also be used as dimensions.

To select a dimension, simply navigate through the tree structure, then select the dimension you want by double clicking it or single clicking then clicking OK. Repeat this process to add additional dimensions.

 

Dimension Properties: The Details Tab

Each dimension also has several options associated with it that influence how the calculation will be displayed.

 

The sorting options control how many items of this dimension are displayed, and what order they are displayed in. The results can be limited to a certain number (items) or a certain percent of the total (percent). These can be either the top or bottom amount of items based either on the value for the items themselves, or the total value of the first measure for each of the items.

The aggregate option lets you define whether or not you want the measures for this dimension aggregated for all items displayed. The total, minimum, maximum, average, or median for all items of this dimension can be displayed as the final element of the list. This characteristic only applies to tables.

Note Aggregates only apply to certain measures.

Dimension Properties: The Measures as Dimensions Tab

Properties that formerly could only be treated as measures can now be used as dimensions. If you choose one of these measure properties in the tree on the Basis tab, a Measures as Dimensions tab appears with some options that are specific to this case.

 

 

Time-Averaging Option Box

In order to treat a measure as a dimension, the measure must be averaged over a user set. The user set averaging must also be based on a measure, and this averaging measure must be equal or higher on the object hierarchy. For example, a measure of number of visits can be averaged per user or per organization. If you chose number of visits per user as your dimension, the dimension items would be 1,2,3…, referring to the users who have one visit, two visits, or three visits. If your actual measure was the number of users, then your table or graph would contain each discreet number of visits in the dimension value and the number of users who had that many visits in the corresponding measure value.

A measure treated as a dimension can also be averaged over a time period. In this case, the discreet values for the dimension are rounded to the nearest whole number.

Note When you choose to have a measure as a dimension, any other dimensions in the calculation should be properties that are equal to or higher in the object hierarchy than the property used for user set averaging. Choosing dimensions lower on the hierarchy can lead to unpredictable results.

Measures

Measures are the properties that are actually counted in the calculation, such as the number of visits or the total duration of all requests. A calculation must contain at least one measure. Different measures can have their own filters applied to them in addition to any filters applied to the calculation. All of the measure properties can be found in the tree under the Basis tab grouped by the root objects that they apply to.

 

Measure Properties: The Details Tab

The Details tab reveals certain options that allow you to format exactly how the measure will be calculated and displayed.

 

A measure can be averaged over a time period and/or a user set. The time period averaging will take the measure and divide it by the number of the time periods that are being analyzed. Normalization also takes place, so if your data includes two weeks in August it will be averaged as though the entire month was included. User set averaging involves dividing the measure by some other measure property that is equal or higher on the object hierarchy. For example, you could have visit duration averaged per visit, per user, or per organization.

The Presentation options affect how the value for that measure will be displayed. A measure can be displayed normally; as a running total in row or column; or as a percent of row, column, calculation total, or total. At the end of the table, you can also choose to have an aggregate for the measure. This can be the average, the minimum, the maximum, the median, or the total.

Measure Properties: The Filters Tab

The Filter tab enables you to apply a filter specifically to one measure in a calculation. This tab is the same as the one for a calculation, with a window containing the aggregate of all parent filters that will be applied to this measure, and a box for this measure's specific filter. Any filter that could be used at the section or calculation level can be applied to a measure. This option can be useful if, for example, you wish to have a measure that is the number of requests last week alongside another measure with the number of requests all month.

 

Dimensions as Measures

Just as any measure property can be used as a dimension, any dimension property can be used as a measure.

 

Dimension properties can be used as measures in any one of three ways:

  • Unique count This creates a measure which is the count of the number of unique dimension items that exist (grouped by whatever actual dimensions you define). For example, choosing file name as a unique count will give you the number of unique file names.

  • Maximum This will cause the program to insert an actual dimension value in the cell rather than a number amount. This value refers to the dimension item which had the highest value based on the first non-dimension measure. For example, a maximum hour dimension as measure in a calculation with another measure of number of requests will output the hour with the most requests.

  • Minimum This functions the same way as maximum, but outputs the item with the lowest value for the first non-dimension measure. For example, choosing minimum with file name as a dimension as measure in a calculation with another measure of number of requests will give the file name with the fewest number of requests.

Note Unlike measures as dimensions, the minimum and maximum options actually require another measure to be part of the calculation.

 

The Relationship Between Dimension and Filter

If you set a request property as a row dimension and then specify a filter criterion that includes a visit property, then you're counting all requests from visits that met that filter. For example, if requests is the measure, file name is the row dimension, and the filter is Browser=Microsoft*, you get the number of requests made by a Microsoft browser for each file.

However, if a visit property is your row dimension and you then specify a filter criterion that includes a request property, then you're only counting visits which include a file meeting that filter criterion. For example, if you count requests with referring URL as a row dimension and File=/order.html as a filter, you calculate the top referring URL that delivered people to your order page. Of course, counting the number of requests might not be optimal, since the order page might be requested twice in the same visit. So the best way to create this calculation is to set number of visits as the measure.

The relationship of dimension to filter is consistent when looking up or down the object hierarchy. For example, looking down the object hierarchy:

  • A user property for the row dimension, with a request and/or visit property in the Filter criteria, includes only those users who had at least one visit that met the visit property and/or one request that met the request property.

  • An organization property for the row dimension, with a request and/or visit and/or a user property in the Filter criteria, includes only those organizations that had at least one user who met the user property and/or who had at least one visit that met the visit property and/or one request that met the request property.

Looking up the object hierarchy:

  • A request property as a dimension, with a visit, user, or organization property within the filter criteria, counts all requests from visits, users, or organizations that met the filter criteria.

  • A visit property as a dimension, with a user, or organization property within the filter, counts all requests from users or organization that met the filter criteria.

  • A user property as a dimension, with an organization property within the filter criteria, counts all requests from organizations that met the filter criteria.

Calculation Property Examples

For your reference, several example questions are listed below, along with the corresponding measures and dimensions requiremed to answer the question. The standard analysis files included with Site Server Express Analysis are another excellent reference to calculation property combinations. Just open these files, double-click the individual calculation properties, and view settings that achieve the calculations found within the standard reports.

  • What were our top documents by number of requests for that document?

Measure: 

# of requests

Dimension: 

Request, document title

  • What were our top documents by number of visits that included the document?

Measure: 

# of visits

Dimension: 

Request, document title

  • What were our top content directories by number of visits that requested content from that directory?

Measure: 

# of visits

Dimension: 

Request, level 1 directory

  • What were our top content sub-directories, within the entertainment area of the site, by number of requests that requested content in that directory?

Measure: 

# of requests

Dimension: 

Request, level 2 directory

 

Request, level 3 directory

 

Request, level 4 directory

Filter: 

File = /entertainment*

  • Which registered users requested the most documents in those directories?

Measure: 

# of requests

Dimension: 

User, user name

 

Request, level 2 directory

 

Request, level 3 directory

 

Request, level 4 directory

Filter: 

File = /entertainment* and Username <> Unregistered*

  • For the unregistered users, which organizations requested the most documents in those directories?

Measure: 

# of requests

Dimension: 

Organization, name

 

Request, level 2 directory

 

Request, level 3 directory

 

Request, level 4 directory

Filter: 

File = /entertainment* and Username = Unregistered*

  • How does the number of requests per visit to each content directory differ by the browser the user is using?

Measure:

# of requests (with User averaging: per visit)

Dimension:

Visit, browser product

 

Request, level 1 directory

  • What is the average visit duration?

Measure: 

Visit duration (with User averaging: per visit)

  • How many requests per visit were made by visitors that requested the order page?

Measure: 

# of requests (with User averaging: per visit)

Filter: 

VAR_filename = /order.html

  • What path did visitors take through our site?

Measure: 

# of visits

Dimension: 

Visit, first request, file name

 

Visit, first request, next request, file name

 

Visit, first request, next request, next request, file name

  • What path did visitors take to get to our order page?

Measure: 

# of visits

Dimension: 

Visit, any request, file name

 

Visit, any request, previous request, file name

 

Visit, any request, previous request, file name

Filter: 

VAR_filename = /order.html

  • How many requests for the catalog form were made by visits that included that catalog form?

Measure: 

# of requests (with User averaging: per visit)

Filter: 

filename = /order.html

  • How does average visit duration differ by organization type?

Measure: 

Visit duration (with User averaging: per visit)

Dimension: 

Organization, type

Filter: 

File=/documentation*

  • Which users spent the most time reading the user's guide?

Measure: 

Request duration

Dimension: 

User, user name

Filter: 

File=/documentation*

  • How many requests for Java content were made by non-Netscape browsers?

Measure: 

# of requests

Filter: 

Media=Java and Browser<>Netscape*

  • What percent of our users have browsers that support secure transactions for credit card processing?

Measure: 

# of users

Filter: 

Security=*SSL*

  • How many different people from Swampscott, MA clicked on our banner on Bill's site, came through Melinda's web browser, downloaded our restaurant menu, returned to our site, and then ordered lasagna?

Measure: 

# of users

Filter: 

City=Swampscott and State=MA and Browser=Microsoft* and Referrer=InfoSeek* and File=/menu.html and File=/confirm_order/lasagna.html

  • For all users whose first visit came from our advertisement on InfoSeek during the month of January, how many have requested our order page?

Measure:

# of requests

Dimension:

Request, document title

Filter:

UFV_RefOrg=InfoSeek* and UFVAR_Date < 2/1/95

  • What were the top referring URLs for visits that requested our order page?

Measure:

# of visits

Dimension:

Visit, Referring URL

Filter:

File=/order.html

  • What did the first visitor from Intersé Corporation request from our site?

Measure:

# of requests

Dimension:

User, first visit, any request, document title

Filter:

Organization="Intersé Corporation"

  • What did the first user to register at our site from Intersé request on their last visit?

Measure:

# of requests

Dimension:

Organization, first user, last visit, any request, document title

Filter:

Organization="Intersé Corporation"

  • In what month did we have the most new users from Korea?

Measure:

# of visits

Dimension:

Visit, first request, month

Filter:

Country=Korea

  • What is the trend over the last 10 days in requests for a file X ?

Measure:

# of requests

Dimension:

Request, day

Filter:

Date > Today-11 And File=X

Sort:

by dimension

Presentations

Presentations establish the output of your reports. They take the form either of a graph or a table, each with an appropriate icon. They are set in the hierarchy at the level of the calculation, so that you can customize the look and format of your reports for each actual analysis you make. You are also free to add multiple graphs or tables to a calculation, which gives you enormous flexibility in representing your results.

The two types of presentation, tables and graphs, can both be modified for content (the elements they display) and for design (their color, look, and form). Settings for individual tables or graphs are made whenever a presentation is added to an analysis, or the settings can be edited by right-clicking a table or graph icon and selecting Edit.

Note General settings that apply to all graphs and tables are selected through Options on the Tools menu.

Table Properties: The Design Tab

 

The figure above shows the Details tab of the Table properties sheet. Using this tab, you can customize the design of your table by selecting fonts, colors, layout, and so on.

The upper half of the tab lists all table elements in their standard diagrammatic format. The lower half provides a way to alter text properties, table cell properties, and table gridlines. To edit the text properties for a particular table element, select the table element in the upper half of the tab by single clicking the box containing the label for that element. Then, proceed to change one or more of the following font-related options: type (Arial, Times Roman, and so on.), size of font, style of font (bold, italic), or color of font.

You may also choose whether or not to display the table element in the table. In the table cell properties box, if you check Visible, the table element that is selected in the upper half of the tab will be displayed. The color option selects the background color for the table element.

Finally, the Table Gridlines box contains an option that affects the entire table layout. If Visible is checked, the table will be displayed with gridlines. Otherwise, the table will display table element labels and measure values without separating lines.

The figure below represents more precisely the layout of the elements in a typical table found in any of the standard reports.

Table Properties: The Details Tab

 

The Details tab contains several options that impact the contents and layout of your table. The top box on this tab, Document titles, is only relevant to tables that contain "Title" as a row or column dimension. Within this box, selecting the top option indicates that you want to view document titles in your table. Document titles are found by running the HTML title lookup utility from Usage Import. If the document title is not available, then the file name will be displayed. The second option is to display both title (again, if available) and file name.

The middle box allows you to set the way measures are displayed in the table. Measure labels and values can either be shown in rows or columns. The default value is to show them in columns.

The last box on this tab controls the size of the outputted table. By default, a table can have no more than 500 rows and 50 columns.

Table Properties: The Templates Tab

 

The Report Writer's Menus and Tools

The Tools Menu
Options

The Options selection from the Tools menu allows you to set preferences for a number of general Report Writer operations.

The top section of the figure above concerns the inclusion of usage analysis definitions at the bottom of every generated report.

The Graph Images box contains options related to the output format and file naming for images generated in Report Writer's reports. You have the option of printing graphs with solid colors or patterned lines. If you plan to print reports, you should select printing with patterned lines. In the HTML img src box, select the case to use in naming image and source files. If you are moving images onto a UNIX computer, be sure to select the correct case sensitivity option.

 

The Report Presentation tab provides two general options related to report style that will be applied to all generated reports. The first option is the display of the header information. The header contains information about the analysis time period, site analyzed, and report sections. The default is to display this information. You have the further option of altering the font type, size, and color. The bottom box on this tab controls the background color for the report.

The Default Directories tab allows you to set the path information for completed reports, analysis files, and template files.

The Calculation Cache tab allows you to speed up report generation by re-using previous calculation results. The cache folder shown in the directory in the middle of this tab contains files which store the results of recent calculations.

The Domain search window allows you to examine your database for queries which match any of the domain information maintained by Microsoft Site Server Express. This provides a rapid first check to aid in developing analysis filters. For example, you can query the database for visits from a particular city, state, zip, or organization, or you can use wildcards as criteria for searches on the domain names themselves. When your domain search produces significant results, you may want use them to devise a complete analysis report.

The figure below shows a domain search that details requests originating from educational institutions (*.edu).

The organization, domain and city fields in domain searches accept the standard wildcards; for state and zip code, you must submit complete entries of the appropriate length and syntax (i.e. the two letter state postal codes, and the five-digit zip codes, respectively). Once you specify the criteria you're interested in, click the green Start button on the toolbar or the Enter key. The Search results grid displays a maximum of 1,000 domains that match your criteria. The results can be sorted in the manner of the database contents window, by double-clicking the column you wish to use as the sort key.

The File Menu
Save Analysis

The Save options on the File menu of the Report Writer allow you to preserve your Filters and Calculations. Since analyses provide the data out of which the reports are created, they are saved as discrete files with a unique *.mfa file name and extension. Mfa files may be saved anywhere on your computer, but if you place them in the Catalog folder and its subdirectories, your analyses will be available whenever you open the Catalog from the Report Writer's File menu or by means of the Open Analysis Catalog button on the toolbar. (The reports generated are saved with the appropriate *.htm, *.doc, *.xls, and *.txt extensions, for HTML, Word, Excel, and text documents, respectively.)

Renaming Elements

The tree displays the name of each element or node. The analysis reports use these names to label results. If you want to rename a node, simply click its name to highlight and again to bring up the edit box with the flashing insert-cursor.

 

Note No two nodes at the same level can have the same name (two sections in one report, or two calculations in one section, for example). Node names are case insensitive.

Deleting Nodes

If you're not interested in the results of a specific section or calculation, or if a calculation simply takes too long to complete, you may want to delete a node. To do this, simply click the node, and click Delete. The program prompts you to confirm the delete. You can also use the delete key on the toolbar.

Note Once you confirm that you want to delete the node, you cannot recover that node unless you previously saved an analysis file.

Moving Nodes

To change the order in which a section appears in the report, or in which the results of a calculation appear within a section, you can move the node. Simply highlight the node, and then drag and drop it to the desired location. You also have the option of using the up and down arrow icons on the toolbar.