The number of possible Web Metrics is large and increasing. Multiply by the number of Dimensions, and there is nearly an infinite number of things an analyst can look at. Get your basics down in Web Analytics 101 – Web Metrics.
9. Indices: created by normalizing a group
of metric observations (usually to 100)
and calculating an observation’s value
relative to the top normalized value.
13. The number of possible Web Metrics is
large and increasing. Multiply by the
number of Dimensions, and there is nearly
an infinite number of things an analyst can
look at!
14. How to Cope
Managing infinite web metric options.
15. Prioritize
Make sure your business leaders have designated their top metrics (commonly
called KPIs – Key Performance Indicators)
16. Build a Metrics Tree
Show how all of your metrics interrelate. Connect them to revenue and profit.
Focus on drivers and leading indicators.
17. Revenue is a lagging indicator.
Customer satisfaction is a leading indicator.
New customers is a driver.
20. There really is no “average” Internet user.
Small numbers of users can drive a lot of traffic.
Break down your important metrics into deciles to gain new levels of insight.
21.
22. Threshold metrics add a “mix” focus
to business goals, which can have
major impact on growth.
23.
24. The Law of De-Averages
Be suspicious of averages – taking them at face value can sometimes lead to
wrong conclusions.
26. Keep in Mind
Always understand how a metric will be used, and whether an
average (mean, median, or mode) is the best way to present the
metric.
27. Keep in Mind
Always understand how a metric will be used, and whether an
average (mean, median, or mode) is the best way to present the
metric.
Always make an attempt to understand the shape of the distribution
behind the metric’s average, and if that distribution is changing over
time.
28. Keep in Mind
Always understand how a metric will be used, and whether an
average (mean, median, or mode) is the best way to present the
metric.
Always make an attempt to understand the shape of the distribution
behind the metric’s average, and if that distribution is changing over
time.
Where possible, break a metric out into quintiles or deciles, based on
frequency of observations or users. Map the deciles to usage to
create a customized version of the “80/20” rule for that metric.
29. Keep in Mind
Always understand how a metric will be used, and whether an
average (mean, median, or mode) is the best way to present the
metric.
Always make an attempt to understand the shape of the distribution
behind the metric’s average, and if that distribution is changing over
time.
Where possible, break a metric out into quintiles or deciles, based on
frequency of observations or users. Map the deciles to usage to
create a customized version of the “80/20” rule for that metric.
If the story of the de-average is significantly different than the story
of the average, ensure business owners and other action drivers are
sufficiently educated.
30. Keep in Mind
Always understand how a metric will be used, and whether an
average (mean, median, or mode) is the best way to present the
metric.
Always make an attempt to understand the shape of the distribution
behind the metric’s average, and if that distribution is changing over
time.
Where possible, break a metric out into quintiles or deciles, based on
frequency of observations or users. Map the deciles to usage to
create a customized version of the “80/20” rule for that metric.
If the story of the de-average is significantly different than the story
of the average, ensure business owners and other action drivers are
sufficiently educated.
Where needed, set primary or secondary business goals on threshold
metrics and specific segments, to further drive desired performance
improvements.