A Web marketer commented to me recently that he thought his firm should focus first on basic “blocking and tackling” before moving on to more sophisticated efforts. When I asked what he considered basic, he said analyses like understanding customer value. He considered the advanced projects to be things like site optimization and multi-variate testing.
My initial reaction was that he had it backwards. To me, it seems harder to assemble accurate customer data across multiple channels and systems than to implement off-the-shelf testing software.
But maybe he had it right. Even though assembling comprehensive customer data may be much more challenging, optimization without a proper measurement capability can lead to bad decisions. So even though it might be quicker and easier to set up the testing systems, it’s might be best to concentrate on assembling the customer data.
As a devotee of lifetime value, I find this a comforting notion. But the hard-nosed businessperson in me says testing will yield immediate benefits even if it uses short-term measures like response rate. You need to look at the back end too, and as quickly as possible. But building a comprehensive measurement system may take a long time. Not testing until that's done means deferring the benefits that testing can provide. Unless testing is likely to lead to harmful choices, it's worth moving ahead even with imperfect measurements.
Wednesday, April 25, 2007
Subscribe to:
Post Comments (Atom)
3 comments:
The realist in me says there are a lot of moving parts, and you [often] can't simply wait to perfect one before moving on to another.
Any firm that tries to "perfect" CLV measurement before moving on to site optimization and MVT will ultimately find itself well behind the curve.
At the recent DMA Financial Services conference, Martha Rogers told the audience that, in her firm's experience, it takes 8 quarters to get Return on Customer measurements accurate.
So...readers of this blog...how many of you want to wait two years before building site optimization and multivariate testing capabilities?
So perhaps the more relevant question is, how bad do your existing metrics have to be before you can't use them? It's easy to make really harmful errors if you, say, look at only gross rather than paid response. So there must be some minimum level of metric quality required before you can start testing. (Perhaps this has no general answer - but there might still be some criteria to judge by.)
David / Ron,
Stick with the "hard-nosed business approach" - the processes and metrics for overall web site optimization are pretty much standardized at this point. There would be a lot of opportunity cost involved in "waiting for LTV".
A second concern is without site optimization first, you are probably going to be working on developing a suboptimal version of LTV, based on the "wrong" customer.
For example, we all know how customer source and offer can be predictive of LTV. If you're not doing the basics well in terms of campaigns first, you really could be wasting a lot of time divining LTV.
And I, like David, am a very strong supporter of the LTV concept!
Post a Comment