Hi, All,

On another message board, someone who's running a small ecommerce site
asked for general advice to get started with WA. Below is my
off-the-cuff response. Of course, free advice is worth what you pay
for it ... but what other things would you include for a first-pass,
where the idea is to have someone jump in and get started, saving the
worries about theory, multivariate optimization, cross-channel
analysis, cookie / browser / privacy tracking limitations, and all the
rest of these things for later?

Here was my response -- what one or two main things did I leave out?
And what did I toss in there that really shouldn't be worried about
until later (for some undefined value of "later")? And of course,
apologies to my dear colleagues at the various competitive
benchmarking sites for reasons you'll see later -- I'll buy you a
drink when next we meet, or maybe help you close a sale some time.
;-)

WDave

--------------------------------------

Well, if you're doing it on a budget, get the free Google Analytics
for the new site, and track it from day 1. GA gives you about 80% of
what you need, for free(!), which is pretty darn good. You can focus
on improving quite a bit before you run out of GA capability.

Hmm -- just glanced at your sites, and noticed that the site map on
the old site is more like a catalog index, but isn't what most folks
expect from a site map. (Which is a list of the sections of your
site, like About, Wholesale, Contact, etc.)

One of the things you'll want to do is put things where your customers
*expect* to see them, even if you don't think that's the way things
belong. You could get some user interface and user experience folks
to help with that, maybe on a one-time consulting basis. They can run
things like card sort tests (give a test subject some cards, one for
each category, and ask them to arrange them in a way that makes sense
for them. I.e., to come up with menu trees that flow naturally.
(E.g., Do you put ink cartridges with printers, or with
accessories?)), and give other feedback, like position, size, and
color of button placement, or promo coupon placement and wording, etc.

As for WA, you'd use that to measure a site and all the clicks and
revenue generation of each button or link on every page, then make
whatever changes are suggested, then measure again to see if it's
better or worse. That will give you plenty of iterative testing to
do. (Measure, change, measure again, change, measure yet again, etc.)

Your first basic metrics are 3 measurements and 3 ratios, all along
the customer conversion funnel. 1. Unique daily visitors to your
site. 2. Number of orders placed. 3. Revenue. Then the ratios are
4. Orders divided by visitors = "conversion". 5. Average Order
Value. 6. Revenue per unique visitor.

Establish your baselines (averages) for each of those six, by day of
the week, week of the year, etc., and over a year or five, get a
handle on seasonality (holiday buying peaks, spring fashions, etc.).
Measure your "lift" over each baseline caused by each marketing thing
you do -- affiliate traffic, search engine traffic (both for natural /
organic SEO and also paid SEM), email campaigns, print and TV
campaigns, etc. Calculate your lift divided by your expense, by each
marketing channel, and figure out which ones are most profitable
(return on ad spend = ROAS) for your specific customer base. It may
not be the same for you as it is for, say, drugstore.com, or dell.com,
as your customers may be different, and/or what they expect from your
site and the way they use it may be different than how they use those
other sites.

Ignore any industry benchmarks -- they will only lead you astray.
Measure, test, measure, etc., is the way to go here.

So track a bunch of stuff, see if it makes sense, intuitively, and
when it doesn't (because it probably won't!), come ask in the WAF.
But try it yourself first -- you'll probably surprise yourself with
how much you can figure out just by looking at the numbers and
thinking about them carefully.

WDave