Support/MetricsDashboardPRD/Metrics

Here are all the sections. You'll note that some metrics are in more than one section.

Legend

Unless specified, metrics are range specific. This means they are collected weekly, displayed for the last week, and cannot be displayed for other date ranges. For example, the "number of KB contributors" metric cannot be collected daily and accumulated to display the weekly metric, since the same contributors might have been active during several days, making the weekly metric incorrect.

Unless specified, metrics are coming from the TikiWiki db (using SQL queries).

• + : metric is not range specific and should be collected daily even though we will display the metric with the weekly date range by default. Metrics that are not range specific will support a custom date range if/when we add UI for it.
• @ : this range specific metric should be collected monthly and weekly.
• -: this is a range specific metric and is only collected weekly.
• =: this metric is calculated from other metrics
• (Om) : metric is coming from Omniture
• List based: metric is NOT a number but will be presented as a list. How this will work (ideally) is explained in the section below.
• i, %, f: integer, percentage or floating point number.

Date ranges

• Last week -- this is the default range and is supported by all metrics
• Last month -- only supported by + and @ metrics, as explained above
• Custom date range -- only supported by + metrics, as explained above

Metrics

Overview

• i@Number of unique visitors (Om)
• i+Number of new forum threads
• i+Number of new live chats
• i+Number of new KB articles
• i-Number of active contributors
• i-Number of new contributors
• i+Number of new registered accounts
• %@Combined total CSAT score (for KB, forum, chat)

Knowledge base (All tiki queries)

• i@Number of modified articles
• i+Number of new articles
• i-Number of active KB contributors
• i-Number of new KB contributors
• %=ratio of votes that say articles are helpful (calculated from:)
  • i+number of yes votes
  • i+number of no votes
• %=ratio of votes that say articles are easy to understand (calculated from:)
  • i+number of yes votes
  • i+number of no votes
• %@CSAT score This can also be + in that it can be collected daily and calculated for other date ranges using a weighted average rather than a sum.
• i+Number of CSAT votes

Forums (All tiki queries)

• i+New forum threads
• i+Number of replies by contributors (total)
• i-Number of threads in each status category (after 1 week)
  • Request for more info
  • Solved
  • Proposed solution
  • Locked
  • None of the above
• i-Number of threads with no replies for over 1 week
• i-Number of active contributors
• i-Number of new contributors
• %@CSAT score This can also be + in that it can be collected daily and calculated for other date ranges using a weighted average rather than a sum.
• i+Number of CSAT votes

Live Chat Many of these aren't done with queries so there will be some work with zzxc here.

• i+New chat requests
• i-Number of active contributors
• i-Number of new contributors
• i+Number of chats answered
• i+Number of chats with problem solved
• i+Number of chats to follow up on
• %=chats answered % (calculated from the "number of chats" metrics above)
• %=chats solved % (calculated from the "number of chats" metrics above)
• %=chats to follow up on % (calculated from the "number of chats" metrics above)
• f+Number of hours open
• %@CSAT score This can also be + in that it can be collected daily and calculated for other date ranges using a weighted average rather than a sum.
• i+Number of CSAT votes

Traffic and usage

All of these come from Omniture unless otherwise noted

• i+Total pages hits
• i@Number of unique visitors
• i+Start page # entries
• %=Start page bounce rate (calculated from...)
  • i+Start page number of bounces
• i+Inproduct start page # entries
• i+Inproduct start page bounce rate (calculated from...)
  • i+Inproduct start page number of bounces
• i+Number of searches
• i+Number of people who go to Ask a question
  • from a KB article,
  • from a search results page,
  • directly.
• i+Number of people who go to Ask a question and then go to forums, live chat (Om) -- This is a traffic question: Of the people who land on Ask a question how many try to go to forums and how many try to go to live chat, it lets us know (for example) that everyone is trying to do LiveChat and only doing forums if LC is closed.
• -Top search terms (from search plugin ideally) List based

KB article metrics

This will ideally be ALL of the KB articles with metrics for each. Then presentation would be in the form of lists based on ranking based on these metrics. As such, these are all List based

• i+Number of page hits (Om)
• i+Number of search-hits (Om) Number of times people select the page in search results
• i+Number of front page hits/inproduct front page hits (Om)
• f-Time spent on page (should be on Om) [djst: not possible to get other than a distribution average. i think we might give this one a miss]
• %=Percentage of people who said yes to "was this helpful?" calculated from:
  • i+Number of yes votes
  • i+Number of no votes
• %= Was this understandable votes % calculated from:
  • i+Number of yes votes
  • i+Number of no votes
• %@ CSAT score and vote count This can also be + in that it can be collected daily and calculated for other date ranges using a weighted average rather than a sum.
• Calculated score
• Calculated helpfulness

Locale-based metrics -- Gather each of the below metrics for EACH locale. List based

• i+Number of KB edits/translations/approvals to indicate activity
• %-Percent of KB articles in each group:
  • Up to date
  • Needs updating
  • Needs translation
  • Needs approval
• %-Combined completeness score
• i+Number of visitors (Om)

List based metrics

There are three special types of item: search terms, articles, locales. For each, we'll be collecting a variety of metrics for EACH article/search term/locale over time. On the dashboard the presentation is a ranked list of the top 5 for a given metric. For example:

For articles, we'll collect: page hits, hits from search engine and votes daily. Then when we generate the dashboard, we just have a list of the five articles with the most page hits (for example) as a metric.