The role of web analytics in seo
In the beginning was natural search, and then ppc came
onto the scene. Sense will eventually prevail one assumes
and the current infatuation with paying for traffic will
fade as the long term cost effectiveness of attracting natural
search traffic will again come back into fashion.
The difference now is that web analytics, traffic monitoring,
emetrics, whichever term you prefer, has moved on and the
software available to analyse what is happening on your
site can be used to tremendous advantage in developing your
natural search traffic. In a nutshell, analysing your traffic
allows you to focus your seo efforts where they will have
most effect, prevents you wasting your time where they will
make little difference, and is invaluable in identifying
where things need to be put right.
The website monitoring basics
Make sure that whichever traffic monitoring software you
use, it is up to the job. There are many on the market from
the free packages that come with most hosting, through mid
range products that offer full revenue tracking to high
end products that are totally customisable for the needs
of different individuals in large organisations. Things
to look out for if you are using a free package are that
they provide
keyword search phrases (not just individual keywords)
ability to separate natural and paid search
robot reporting (for technical reasons this is only ever
available with log file analysis not JavaScript traffic
monitoring solutions)
ability to follow visitors through the site, eg identify
how visitors using different keywords to find the site behave
once they have arrived
Identifying the most valued visitors
One of the most important things we can find out is which
groups of visitors are our most valued. The key point is
that we are always looking for relevant and valuable traffic.
Obviously with free traffic, irrelevant traffic is not quite
so unwelcome as when we have paid for it, but it is still
a waste of time and in extreme cases can clog up band width.
I use three criteria to define relevant traffic:
1 Completed a successful visit, (conversions in sales jargon)
this could be either purchased something from the site,
reached a certain page eg the contact page, downloaded software,
etc
2 Stayed on the site a significant length of time, usually
defined as more than 5 seconds. Most visitors make up their
mind whether they are interested in the site well within
five seconds
3 Have visited the site before - ie returning visitors
I analyse my traffic and see which keywords deliver the
traffic that meets at least one of these three criteria.
I am looking particularly for a high percentage of visitors
from a particular keyword that qualify as valued as much
as for high absolute numbers (ie I am just as interested
in a keyword that delivered 4 visitors but all of them were
valued as one that delivered 4000 but 3900 of them left
the site in under 5 seconds).
I then check the rankings of those keywords that meet the
criteria on Google, MSN and Yahoo! I check both the global
and UK versions of the engines depending on whether the
target market is national or international. Note, I only
use rankings to investigate behaviour I see in the traffic
analysis, I never chase rankings as an end in themselves.
When I find a valuable visitor keyword with a poor natural
ranking then I start the seo process to improve the situation.
The aim of course is to get a page 1 listing for every keyword
that delivers valued visitors. Just as importantly I don't
waste effort on a keyword that might delivers a huge volume
of traffic all of which is valueless. SEO is a time consuming
occupation and identifying where to focus your effort is
key to success.
This approach can often lead to identifying missed opportunities.
I have found on several occasions that I may only be getting
one or two sales from a particular keyword but the conversion
rate is high, even 100%. This has usually indicated that
I have missed a good keyword and it is languishing on page
five, but occasionally a dedicated searcher finds it and
bingo, we have a sale. Get that keyword onto page one and
it is reflected in the bottom line.
Widening this approach with ppc
I use this analytic approach when doing initial keyword
research (usually using Wordtracker). I set up a ppc campaign
to run for a short time, maybe only a week if traffic volume
is high. I then go through exactly the same process and
see which of the keywords are attracting valued visitors.
Obviously this is relevant to managing ppc campaigns effectively
but it is equally relevant to identifying keywords that
merit work to improve their natural rankings.
Identifying when a ranking has dropped
OK so things are going well, we have developed good rankings
on some new keywords and everything is hunky dory. But then
suddenly traffic and/or sales drop. How do we identify the
cause of the problem. Again our traffic analysis comes to
the rescue. This is considerably simplified if your analysis
program includes a "what's changed report" which will immediately
identify a fall in traffic from a particular keyword, otherwise
it is a case of comparing the traffic from high performing
keywords from before the drop and afterwards.
So if it is "pink widgets" that has suddenly stopped attracting
traffic, then either the ranking has dropped or the bottom
has fallen out of the pink widget market. If it's the ranking
that has dropped, this can be addressed with normal seo
techniques although if a keyword has been doing well I am
loathe to dive in and take remedial action as often things
sort themselves out if left alone. A preferable approach
is often to put up a new page targeting the keyword (being
careful of course to avoid any duplicate content problems)
rather than change the first one, especially if it is still
doing well in other engines. Equally, if a drop in rankings
was not the cause of a drop in traffic from a particular
keyword, no effort will be wasted in curing the wrong problem.
Perhaps the market really has dropped out of the pink widget
market, if so you now have some valuable piece of information
if a big offline marketing campaign were about to be launched.
Robot tracking
Looking to see what the robots are looking at on your site
is (or at least should be) a vital part of the seo process.
Quite simply, if a search engine's robot has not found your
page then no amount of optimisation will do any good. Again
this is about not wasting time on pointless tasks that are
not addressing the root cause of a problem.
So what should you be looking for in the robot report?
The most important information is whether the robots have
found a particular page, or if content has changed, whether
they are revisiting regularly and have picked up the new
content.
If the page hasn't been found then why not?
1 Maybe it just needs a bit more time. For any site that
isn't totally dormant, a new page should be found within
a couple of weeks, more probably a matter of days.
2 If the page has been on the site for a while and still
not found then the first thing to check is whether you are
putting a robot friendly link under the robot's nose for
it to follow. A text link is best and the site map is the
ideal place to put this (normally site maps are spidered
frequently but of course your robot report will confirm
this).
3 If it is still not being found you may have to take a
longer look at the whole navigation structure of your site
as it is likely that other pages are not being found either.
Once the new page has been found then it shouldn't take
too long to get into the search engine index and start sending
traffic to your site from the keywords it is targeting (unless
the site is relatively new in which case it can take much
longer). At this point I would check the rankings in the
main engines to check that I have written the page in such
a way as to get the top rankings I am working for. Obviously,
the more competitive the ranking the more difficult a page
one listing will be and it may take anything from a few
days to months to get onto the first page depending on how
much on and off page seo is required for the phrase.
Just as robot visits to new pages should be monitored in
the robot report, so should robot visits to pages that have
new content. Again new content should result in new rankings
for targeted keywords but of course can adversely affect
current listings, which is why I always favour new pages
rather than amending current pages if a current page is
already attracting natural search traffic.
Entry pages and lost opportunities
Looking at your traffic analysis report will tell you
immediately how many doors (entry pages) visitors are using
to come onto your site. If your analysis shows that almost
all are coming in through the home page then a big opportunity
is being missed. Any single page (even the home page) can
only target a very few keyword phrases and these need to
be fairly similar. Hence if everything is coming in through
the front door, then the chances are that there is a very
narrow spread of keywords that are attracting them. Occasionally
this is the nature of the market. Estate agency is an example.
Someone is either looking for a house in a particular area
or they are not, and they are only going to search on a
very limited number of terms to find one. But with most
markets, there will be peripheral and complementary markets
where traffic is available.
Once you start looking at your web stats you will really
begin to understand how your site works - in fact, and I
know it is a cliche, you will wonder how you ever managed
before. Monitoring your traffic will really allow you to
focus your precious seo resources where they will have most
effect, and you will reap the rewards of your efforts not
just today like ppc, but next month and next year…