Web Analytics Demystified

kaushik.net/avinash/web-analytics-demystified/

Avinash Kaushik December 18,

2007

New to web analytics? Confused about web Analytics? Think it is

too hard? Scared of tools and consultants?

This post is for you, its goal: Web Analytics Demystified!

Yeah!

Web Analytics is complex. That is what it is. Complex.

Get the nuance? Complex. Mysterious. Inviting. Come in. Sit down. See what's there. No free
rides. You'll do your part, your efforts will have a rich payback.

Complex holds the promise that you'll get it. Nay, you can get it. Come in, welcome.

Start with this post. [It builds on my post from last week: Web Metrics Demystified.]

Five simple things you can do with absolutely no knowledge of Web Analytics (or Web
Metrics as Europeans tend to say!). No matter what your site does you can use the ideas
below to instantly benefit from your website data.

On a scale of zero to a hundred (with a hundred being the highest analytical awesomeness
peak) this post will move you from zero to thirty five.

That's it.

You are new to data. I hope to move you to forward part of the way. And I promise it won't
hurt your brain and the rewards will be many.

[If have have achieved the peak of awesomeness then congratulations! The post below
collects some online marketing nuggets from my experience that you might find
interesting.]

Set aside a couple of hours to do the work, 20 mins to read this post, 1 min to print it.

Demystifying Web Analytics Summary:

#1 Get the Primitive Basics Out of the Way.

#2 Understand Traffic Sources.

1/14
#3 Fix Stuff / Save Money!

#4 Two Words: Site Overlay!

#5 Focus on Outcomes (To get rich is glorious! – Deng Xiaoping).

Details, Details, Details:

Get ready to swim in numbers, and survive and thrive (!)……

[Important : What Do You Do Next in each section is bonus material, it will make you super
smart but not required for this Web Analytics Primer.]

#1 Get the Primitive Basics Out of the Way.

I am sure as soon as you logged into your web analytics tool you saw something like this:

It is ok to look at this, though even in your humble evolution you are going to get very far
beyond these basic metrics.

What is it?

Visits represents the number of sessions on your website, number of times someone
interacted with your site. Bounce is the number of those who left instantly!

The Page Views number is how many pages were requested in those visits. Oh and how
many in each visit, Pages / Visit.

Avg Time on Site, how long did people stay on your site.

% New Visits shows how many sessions, interactions, were from people (yes people!) who
visited your site for the first time.

[Note: Blogs are unique "I'll only read the latest post and be on my way" animals. In as much
both Bounce Rate and Time on Site are poor metrics to judge quality.]

What is it telling you?

2/14
First bask in the glory of how good you are (or be sad at how low your traffic is!), for this first
look.

Then here is how I analyze the numbers you see above…

For a non-profit (above data) that is a huge number of visits in a month. But I'll probably
want to trend it over time, as much data as you have, to see if it is going up or down.

Next I am deeply impressed at the combo of Pages/Visit and Avg. Time on Site. Most
websites will have two or three or five (at most) Pages / Visit. 12 is magnificent, as is almost 7
minutes on the site.

The bounce rate at 35% is very good in context of the number of visits and page views. It is
hard to get so many qualified visitors.

I will now do a small dance on the table.

What do you do next?

Look at trends.

See how things are over the last few months.

See how things compare between this month (above) and last month.

No matter what web analytics tool you use this will require pressing two or three buttons at
most.

The Bottomline for Web Analytics Demystification #1 :

With a tiny amount of effort (15 mins or so) you have learned your core metrics and you
know how you are doing on the surface. That wasn't hard right?

This was 5 out of the 35 points of progress. Feels good?

3/14
#2 Understand Traffic Sources.

I know people are on my site. Now I am most curious where are people coming from?

There is so much exciting here! In this humble little graph.

What is it?

Direct traffic are all those people showing up to your website by typing in the URL of your
website or from a bookmark. Some people also call this "default traffic" or "ambient traffic".

Sometimes you'll see "non direct" traffic in the direct bucket. From badly coded redirects
(remember to make all redirects Permanent) or vanity urls, or even improperly coded
campaigns. Now you know why that is important.

[Update: For a deeper understanding of direct traffic, it's causes and how to ensure that
your direct traffic is really direct and not from other sources, please see this post: Excellent
Analytics Tip #18: Make Love To Your Direct Traffic.]

Referring URL's are other websites sending traffic to you. These could be as a result of your
banner ad's or campaigns. These could be all those blogs or affiliates who link to you (after
you send the blog authors uninvited and usually irritating presumptuous spam!!).

Search engines, well that's you know who. Google, Yahoo, MSN, Ask, others. This bucket
will include both your Organic as well as you Paid (PPC / SEM) traffic, so be aware of that.

Finally there is Other. These include campaigns you have run (and then have configured
your tool correctly). Email, direct marketing, etc.

What is it telling you?

I look at the Direct Traffic first because I want to know how much traffic are you getting
from people who know you deep enough to know your url or have you bookmarked.

I also like lots of direct traffic because it is "free" traffic, and traffic who knows you so will
4/14
usually put up with some of your crap.

Finally on ecommerce websites these are "non campaign" traffic which means that they
don't have promotions and discounts etc, so they can be a bit more
profitable (though be nice and give 'em something off!!).

I am going to look at referring url's to identify sources that I don't

know who are sending me traffic. I might visit the referring pages and
see why. For some solid ones I might want to establish a marketing
relationship with them.

I'll also look for traffic sources that should be sending me traffic and
if they really are doing that. The referring url world also shares a hint
of customer intent, why might they be there. Small gold.

Search of course is God, atleast temporarily. It is how we look for

stuff, it is how we find relevant content.

The number, 29.33% tells me how well I am doing. 80% of the web
starts surfing at a search engine. I want a lot of that! Look at the search traffic and the
keywords to understand which engine is working for you and why.

What do you do next?

I am going to look for trends and see where my growth is coming from in the last six
months.

Is the growth from free traffic? Paid traffic? Have my efforts to get people through
other channels panned out?

I also recommend drilling down to specific websites that send you traffic, and of
course keywords and key phrases that are sending you traffic. Both of those help you
understand that critical customer intent. That rocks!!

Oh in both cases, look for surprises.

Bottomline for Web Analytics Demystification #2 :

5/14
You probably spent 30 mins looking at the above picture and drilling down to the next level
report. You know where people come from and their intent. You are now 10 points into
your 35 points of progress! Hurray!!!

#3 Fix Stuff / Save Money!

Enough reporting already, let's see how web analytics can help you save money (after all
you have been at it for 45 mins already!!!). :)

We'll do two things here, we'll look at a report in your web analytics tool that shows the
Bounce Rate for the Top Entry Pages.

The second report from IndexTools shows Bounce Rate for the search key words sending
traffic to your site..

6/14
What is it?

As you perhaps already realize you have lost control of what the home page of your website
is. Search engines decide what the home page of your website is. People search for you (or
click through from a link on another site) and go directly deep into your site.

The top entry pages report, first one up there, shows you how many people are entering on
each page of your website, and hence they show the top "home pages" of your website.
Super!

By adding the bounce rate (in case you don't know what it is please read: Standard Metrics
Revisited: Bounce Rate) you have a indicator of how engaging :) each of your "home pages"
are.

Ditto for the keyword report. Search is where all the traffic is, for now. I did a quick custom
report in IndexTools and it shows Visits and Bounce Rate.

What is it telling you?

Simple: Stinkiness. Your site's.

You are looking, first, for pages with the highest bounce rate. Remember bounce rate
measures this from a costumer's perspective: I came, I puked, I left . Or technically speaking:
single page view sessions.

Pages with a high bounce rate are not delivering on the promise that is driving customers to
your site. The ones in the top ten Entry Pages report need your attention.
7/14
You fix 'em and you are increasing the likelihood that people will go deeper into your site,
and maybe convert.

The keywords report is even more interesting. Here you have Intent. The customer is telling
you why they might be coming and keywords with high bounce rates are where you are not
meeting that intent.

It could be that you are ranked for the wrong keywords. It could be that the pages these
folks are landing on don't have the right calls to action. Fix it.

[I have mentioned this before: Blogs are a unique beast, people come mostly only to read
your latest post, your bounce rates will be high because of how that metric is defined, it is
ok. On any site that exists to have any customer interaction – for profit, non-profit – it is still
a awesome metric.]

What do you do next?

No matter how sophisticated you are you can use the Google Website Optimizer to do A/B
or Multivariate tests. It take six minutes to set up a GWO A/B test. And the tool is free. Pick
pages you want to fix, create a couple versions of the pages, put them into a test. You'll
improve the pages based on customer feedback.

I look at landing pages of the top keywords (one click report in most web analytics tools)
and see how I can improve the copy / content / images / calls to action.

I'll take time to segment the paid and organic traffic for all the search engines and stop
spending money on the paid keywords where I have high bounce rate, until I figure out why
the bounce rate is so high (maybe wrong keywords, maybe you are driving traffic to generic

8/14
pages and need custom landing pages etc).

Bottomline for Web Analytics Demystification #3 :

It takes you about two minutes to get to each report and another minute to look at the
numbers and press a few buttons. At the end of the half hour you have created for yourself
a specific list of focus areas. You know exactly where you need to start when it comes to
fixing pages on your site and potentially improving your paid campaigns.

You are up to 20 of the 35 points. And you know what you are doing!! Congratulations!!!

#4 Two Words: Site Overlay!

I love this report. There I said it. I love it.

You have improved the top entry pages and key traffic driving campaigns (key phrases).
Now figure out why pages that you want to win on your site are not winning, why pages with
key calls to action are not delivering, obvious things you are doing wrong….

9/14
There are a couple of reasons for the fondness I have for this report:

1) For many people numbers and metrics and spreadsheets are still overwhelming. Site
Overly demystifies all that. You see the data visually represented.

2) Even seasoned analysts are not as good at analysis as they should be because they rarely
actually use the website they are analyzing. A critical flaw. Using the site overlay report is a
great way to walk in the shoes of the customers.

You need to do that as well. And it is so easy with any analytics tool!

What is it?

Click density. The report, in almost every tool, shows the number of clicks on each link that
you have on the page.

Each tool will show it slightly differently. Google Analytics, above, shows "colored bars" and
if you hover the mouse over the link it also shows goal conversion metrics. ClickTracks
shows the clicks as a % (very helpful). Others show a "heat map" and others still draw boxes
around elements on the page.

Perhaps my favorite of them all is ClickTracks because it also shows context, key metrics,
about the page right in the site overlay report. Sooooo helpful…..

[Click on the above image for a higher

resolution version.]

It is showing % Page Views, Time on Page,

Time to the Page (why does everyone not
have this awesome metric?), % Exits and
Keywords that brought people to the
page. Nice.

What is it telling you?

You are looking for clusters of heavy

clicks. I look for the top two or three most
clicked links and I'll try to reconcile that
against links that I want visitors to click
on. I'll see what people are clicking on
below "the fold", any surprises there?

I also look at links that ultimately drive

high conversions (you can have
conversions on a ecommerce website, or as in the GA screenshot above, a non-ecommerce
10/14
website).

I am looking for things that connect with people. Do more

people convert on my site if they click on Product
Comparison on the home page or go directly to a Product
Page?

I'll try to follow the couple heavy clicks and see what people
do next. Walk in their shoes. Experience my own website.

Check out referrers to each page, that could explain bounce

and exit rates.

What do you do next?

Identify improvements to your pages.

Consider merchandizing and cross sell and up sell opportunities now that you know what
people like (for example no one is clicking on your blinking promotion in the middle of the
page because it looks like an ad).

If your tool allows you the ability to do this, segment the clicks. Where do people who
convert click vs everyone else, or vs everyone who comes from a search engine or Canada
or on a DM / Email campaign? Segmentation rocks !

Bottomline for Web Analytics Demystification #4 :

You can take up as much time you have available on this. Initially you'll probably spend 30 to
60 mins exploring your top pages. At the end of that you'll be up to 25 out of 35!

This is a very visual easy to understand way you know exactly how people browse your
website, what is working and what it not. No tables, not numbers, no graphs, even your
HiPPO will get this!

If you segment your data, you can do that later when you are a bit sophisticated, you not
only understand data in aggregate, all visitors, but now you are starting to understand
different types of people on your site. And now you can treat them differently!

#5 Focus on Outcomes (To get rich is glorious ! – Deng Xiaoping).

Most web analytics efforts fail to catch on. Few companies are truly data driven.

One reason.

We all focus on the 14,285 reports that come out of our web analytics tool. We go crazy
about visits and visitors and parameters and every sub-nuance of everything. Except
11/14
outcomes.

When you start, don't do that. Don't not focus on Outcomes . :)

Answer the question why your website exists. Hard core answer. Then go through the four
Web Metrics Demystified principles to identify the two or three key metrics that help
measure those outcomes.

For a blog….

Or a non-profit website……

Or a e-commerce website…..

12/14
You'll have your own sweet things you want to track. It is a crime against humanity for you
not to know what you are solving for (yes, yes I am being a melodramatic "13 year old teenage
girl" about this!).

What is it?

For the blog it is the number of people who visit the speaking engagements page, and go
attend one of the engagements, and since all my engagements are now paid – hopefully
attend the event generating revenue for the organizer.

For a non-profit it is also looking at number of people using the core search functionality (to
look for stuff like jobs) and for the few who pay comparing number of people who pay that
small amount.

For a ecommerce website it is the bottomline numbers. Revenue, conversions, average

order value, products sold etc etc.

What is it telling you?

Visitors are coming to the website, but is it having any impact on you.

If there is a impact on your bottom line. Are you converting enough? Not as much?

What's selling and what is not? Why is it selling? How much of it?

All web analytics tools have a capability to help you understand outcomes. Omniture,
WebTrends, CoreMetrics, VisualSciences, Google Analytics or whatever else you have. Get
the above stuff.

If you don't got it, you ain't going anywhere.

What do you do next?

Consider understanding of these beauties: Days To Purchase and Visits To Purchase (More
here: Excellent Analytics Tip#6: Measure Days & Visits to Purchase )…..

13/14
They help you understand user behavior on your website, and in turn tell you what kinds of
actions you should be taking.

I have mentioned how segmentation rocks, do that. Segment your site traffic (paid vs
organic, campaign vs direct, new vs returning, from different Geo's and so on and so forth).

If you have a non-ecommerce website do this: I Got No Ecommerce. How Do I Measure

Success?

Bottomline for Web Analytics Demystification #5:

You probably spent 30 mins looking through your initial set of reports and since this last
module was worth 10 and you are now up to 35 points!

Congratulations you have graduated this delightful Web Analytics Primer!!

Give yourself a pat on the back, or have someone do it for you. :)

Doing the first set of effort outlined above will be approximately three hours. You needed to
bring absolutely no knowledge of web analytics to the table (hopefully this post fills that
gap). And look how much you have accomplished. 35/100 points!! Not bad for three hours
of work. [Ok, ok, ok and an hour to read this post! ]

You are not God's gift to Web Analytics, yet. You'll get there soon enough.

For now you understand your customers, you understand what action to take on your
website, you are in line for a bonus! I am happy.

Ok now its your turn.

Please share your own lessons, perspectives, critique, bouquets and brickbats via
comments. What works for you? What does not? Add your voice. Thank you.

[Like this post? For more posts like this please click here, if it might be of interest please check out
my book: Web Analytics: An Hour A Day.]

14/14