Pay-Per-Click Strategies
for Search Engine Marketers Part Three
We continue
with our PPC special, talking
to three very experienced pay-per-click
operatives - Ammon
Johns , Andrew
Goodman from Page Zero Media and
Jim Banks from Web Diversity who
have all kindly agreed to share
their knowledge. (Continued
from Part One and Part
Two).
We pick up
this week on the topic of keyword
selection....
PD:
PPC and SEO share the same
end:
visitors. Both need to ensure
the right mix of keywords is
chosen if the aim is to attract
and convert qualified visitors
to actions/customers. I've
always found that the PPC approach
allows for a lot of experimentation
and refinement in this area
given the immediacy of the
environment. Can you elaborate
on how you find the right keywords
to target? How specifically
do you use PPC to decide which
keyword terms are worthwhile?
AG:
Keyword selection, I suppose,
comes
in two flavors: basic and advanced.
Basic can get you pretty far.
Everything SEO's generally
say about keyword research
is correct. Use Wordtracker,
use Altavista's Prisma, etc.
etc. Let's face it, folks -
someday, software is going
to do most of this job for
us, and do it in much more
sophisticated fashion. I don't
think it will be Wordtracker.
The
fact that Overture has Match
Driver and that Google
has rolled out Content Targeting
is another harbinger of things
to come. These companies have
in-house technologies which
marry semantic/linguistic analysis
with the tasks facing the keyword
advertiser. It's not too difficult
to look ahead to the day when
more of that functionality
is available within the interface
of major PPC's. Search123 recently
acquired Simpli.com, which,
along with Applied Semantics
(formerly Oingo, which I just
have to add because I like
to say "Oingo") is
a clever startup doing research
in this field of meaning-based
search.
The client's own specialized
info may be even more useful
than the stock keyword research
effort, however. Search frequency
information can always come
out after the fact - there
is no harm in just entering
phrases to see what sticks.
Clients have product lists,
information on competitors,
industry buzzwords, etc. -
that may not come to light
as well as they should in Wordtracker.
Remember, you can get super-granular
here. The name of a single
character in a sitcom, for
example, may only get 82 searches
a day, but if you dumped a
list of 500 such characters
into your keyword list, now
you'd be getting 40,000 impressions
a day adding up to let's say
300 clicks per day (no doubt
bargain-priced). Yep, keyword
research is important!!
OK, that's basic keyword selection.
Then
there is advanced class.
You need to look a bit beyond
keywords, for starters. "One
big dump" of key phrases
all tied to a single ad is
not nearly as good as organizing
these around a wide number
of carefully written ads that
correspond more closely to
a small group of key phrases.
And then there are several
advanced tips and tricks which
shall remain proprietary, copyrighted
info as long as I'm sitting
here :)
Better keyword selection can
even cut down on fraud and
malicious clicks.
As for determining which are
worth the most to your business:
as Jim has mentioned, you need
to be using software which
tracks sales conversions at
least by ad group, and preferably
by keyword.
For the majority of advertisers
who either have no tracking,
very rudimentary tracking,
or some tracking in place but
don't act on the info: I'd
argue that perfectionism will
only hurt them. Instead of
trying to get the ultimate
stats package set up, get something
a bit better than you have
now... and act more systematically
on what you learn.
If you have a larger company
replete with CIO, you of course
have my permission to spend
the wad on metrics reporting,
and to assign staff to brief
execs on the meaning of these
numbers.
It's simple, really. Looking
at Google's or Overture's interface
will show you how many clicks
you received for a given search
phrase over a given period
of time. It will even show
you the CTR. But that info
is useless without knowing
how many of those kinds of
clicks (on that keyword, or
ad, or group of keywords -
however you choose to track)
converted to revenues, and
how much revenue. If none of
the clicks converted at all,
then those are effectively
useless clicks, and you should
stop wasting money on that
keyword!
I should
add the caveat that "desired
actions" may not just
be sales, but newsletter signups,
downloads, and other on-site
actions. Whatever metric you
are comfortable with.
Jim Novo of Drilling
Down and ex of the Home
Shopping Network takes us
a bit deeper - showing that
it is possible to predict
repeat buying behavior. I
think there is a continuum
with this stuff. Jim is probably
to the "hyper-rationalist" side...
if you can analyze the data
as well as he does, you may
find that certain types of
buyers turn out to be your
worst (non-repeat) buyers
when you thought they were
your best. Great if you can
do it, but not for the faint
of heart.
Most companies I work with
just need to install something
that tells them if they're
egregiously wasting their money
on PPC or on aspects of their
campaigns. Most aren't ready
for the hyper-rationalist mode.
One vexing thing is that it
is still difficult to track
some of the PPC traffic back
to the correct source - or
to sort amongst, say, Google.com
Adwords clickers and those
clicking in from AOL, Ask Jeeves,
Earthlink, and Sympatico. It
isn't impossible, but it is
difficult. The same goes for
Overture, Looksmart, and most
of the smaller PPC's as well.
They use many different partners.
As I understand it, you'd need
some pretty sophisticated log
file analysis to sort it all
out.
All of this hemming, hawing,
and prevaricating probably
indicates that we've come a
long way from the days of buying
eyeballs and hoping for the
best!
JB: Boy, you can tell
Andrew has written a book.
I'm still on my first novel
and when I finish it I might
read another.
I think Andrew is right on
the money. There are no good
or bad keywords really. You
could choose a lousy bunch
of keywords that still convert,
or you could find some kick
ass keywords and write a lousy
creative and get a poor response
rate.
Overture
recently introduced some
technology, which, for
a lot of advertisers will be
a godsend. But I see it differently,
what it will do is it will
drive the cost up. Some of
the skill has been removed,
and the "tool" has
been provided to enable advertisers
to have more keywords in their
account and therefore to spend
more money.
In the gold digging days people
would pan for gold in the same
river, and two people could
get different results within
yards of each other, by adopting
different techniques.
As Andrew said, the whole
world can find the huge words,
but the real value will come
from having a few hundred keywords
that have no competition and
a decent number of searches,
so you can dominate the market,
or it will come from being
in the right place, at the
right time for the right keywords,
but withdrawing at the right
time also.
What is interesting now is
that Adwords has been around
for a little over a year. So,
we have historical data from
the same period last year to
benchmark against with some
clients. Quite often a good
campaign can be picked from
the cupboard, blow the dust
off and put your loyal servant
back into the game. I think
this is where companies like
ours will score heavily, because
we have a lot of data, for
a lot of industries, in addition
to the sexy techniques that
are closely guarded secrets
that not even a Lambourghini
would pry from me.
On that last point, I think
Overture and others should
take note. Many historical
data is only available for
95 days. Sure you can print
off a report for posterity,
or download the data as a spreadsheet,
but it should be simpler.
When we take over some client
accounts where they have been
doing their own thing it's
often quite funny to see how
they have set things up. On
Google they might have 3 campaigns
(aptly named Campaign #1, Campaign
#2 etc.. for ease of reference),
they never use phrase or exact
match, and although they get
traffic in decent quantities,
I bet if they were directing
the clicks through their logs
that the majority would not
be relevant to what they were
selling.
If
you think about it most manufacturers
have more than
one product range, so bidding
on the keyword "ticket" would
get you a whole bunch of unwanted
stuff if all you sold was rail
tickets, "sony" would
get you a lot of electrical
goods over and above the video
cameras you sell.
Jim Novo is a great source
of information, but for the
average advertiser it is probably
a step beyond their capability.
If the client doesn't have
the right metrics measurement
tools in place, we look to
sell them one. Of course we
make a small amount of money
on the sale, but it's more
about visibility. The data
gives us the chance to extract
more value out of the information
than most.
We measure the number of single
visit clicks that come in.
If you are paying 20 cents
a click and get 100 a day of
those 1 second visits it adds
up to $600 a month of waste.
Sometimes you need to be ruthless
in culling keywords that don't
work. Sure, try other things
like different landing pages,
different titles and descriptions,
use the price in your ad if
you are selling something that
is price sensitive.
A good metrics package can
be a bit like a personal shopper.
We follow someone round the
site from the time they arrived
(which PPC provider/search
engine/keyword they came in
on), to the time they leave.
it's amazing how little things
make all the difference. Things
like a favourites button make
a big difference to repeat
orders/visits, it's a simple
thing to do, but so many sites
don't do it.
The
other interesting thing is
to note the average buying
pattern. Many people probably
give up on keywords because
they don't get instant gratification.
If you use your metrics you
can often find that a buying "cycle" might
be 8 days, 3 months, and by
using cookie tracking this
is relatively easy to establish.
I've gone off at a tangent,
what was the question again??
PD:
Tangents are most welcome.
Encouraged,
in fact. However, we'll leave
it there for this week. Part
four..
|
