Sensors and network administering


The Dirty Little Secret Nobody Tells You About Website Split Testing...

I've got a confession to make. This mayprice tests in which we tested $17 vs.
come as a surprise to you (after all,$97 and $97 converted better! (Not just
I'm a testing fanatic)... But, testinghigher visitor value, but actually
doesn't improve website conversion everyhigher conversion!)
time.So it really is impossible to know in
Nearly all "online marketing gurus"advance what version or variable is
promote the benefits of conducting splitgoing to win. Experience and prior
tests and multi-variable tests on yourtesting data can help you make an
website. But, few (if any) tell you theeducated guess, but you still have to
WHOLE truth about how to test, what totest it and see.
test and what results to realisticallyI view selecting test variables, kind of
expect from your testing.like picking stocks. You can do all of
The consequences? Many online markersyour homework, use a variable that has
become confused by and disappointed withboosted response on every other site you
their first fledgling attempts attested and still not have a winner.
conducting testing campaigns on theirAnd the opposite can happen. You can
websites.test a variable that you are almost
Because their results often don't livepositive will be a dog, (or even make an
up to the "hype", they become discourageerror when setting up your variables)
and simply assume that "testing doesn'tand it can boost response. I've often
work" or "it's not worth the effort".seen headlines that I hated or graphic
I hear this from my clients all theimages that were down right UGLY convert
time. They will tell me, "oh... I triedbetter than the more appealing versions.
testing on my site and it didn't help."So just like picking a stock, past
Well... in an effort to clear up a fewperformance doesn't guarantee future
of the misconceptions about testing andresults! Every single variable you test
help folks overcome some of thewill not be a winner. In fact when it
unrealistic expectations that simply setcomes to the stock market, if you can
them up for disappointment, I'm about topick a winning trade just 51% of the
tell you the "dirty little secret" thattime, you'll become rich.
the gurus usually fail to mention.The same principle applies to test
Every variable you test will NOTvariable selection. If you're able to
automatically improve sales. In factconsistently pick winning variables 50%
MOST of the variables you test willor the time or more, you're doing an
either reduce response or make nooutstanding job! (Or you're simply not
measurable difference!running enough tests...)
In fact, after conducting over 6,000Now, am I telling you all of this to
split and multivariable tests over thediscourage you from testing? ABSOLUTELY
past six years I've generated someNOT!
pretty good data on what you can expectLike I said at the beginning of the
to achieve in a typical testingarticle, I'm a testing fanatic.
campaign.I've seen small businesses transformed
In most multivariable tests only 1/4 tointo "big" businesses almost overnight
1/3 of your variables will boostvia proper testing.
response. Another 1/4 to 1/3 of yourI've seen a laid off, single father go
test variables will actually reducefrom struggling to pay the bills selling
response. And 1/3 to 1/2 of youran ebook to making a solid six figure
variables won't make a measurableincome by only tweaking and testing 4
difference.items on his sales letter.
(Well... that's not entirely true. IfDone right (and with proper
you run any test long enough a winnerexpectations) testing can indeed change
will usually emerge, but it is not worthyour life and transform your business.
the trouble to let a campaign run forI just want to make sure that you go
months on end just for a 1/10th of ainto the process with the right, long
percent improvement in conversion.)term mindset and that you don't become
Of course, the more experience you gaindiscouraged and quit if your first
in testing, the better your odds ofattempts at testing fail to yield earth
finding a variable that will be ashattering improvements in response.
winner... but you still have noThe good news for testing is, because
guarantees.you can flush the variables that don't
Often variables that you are sure willboost response and only keep the
boost sales, wind up having no effect or"winners", even if you only pick a
worse, decrease response.winner 10% of the time you'll still make
I've seen tests where a 1 year guaranteehuge gains in conversion and response
was tested against NO guarantee and theover the long haul.
no guarantee version won. I've conductedHappy testing!



1 A B C 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95