What do you do with your stats?

How do you use them to improve your site and business?

[meg8]

Well, here at my little e-commerce site the rush for christmas widgets is over and it's reporting time where I take a breath and work out what the plan might be for 2007.

I've been looking at my stats, and now I'm overwhelmed by information.

My site is sort of split in two, with lots of juicy information pages for both my customers and my search engine rankings and then lots of shop pages, a lot of which also rank pretty high.

What I've found is that the common search terms which bring a few hundred people to the site never convert into money, while the terms that convert are always completely individual, and a bit difficult to anticipate. And some of the individual terms that I'd imagine would convert, just don't.

This leads me to think that rather than react to the stats I should just concentrate on making the site better. (but what's the point in having this information if I don't use it?)

What do other people do with their stats?

Cheers.

[jatar_k]

one thing to do is take those terms and try to find similarities so that you may be able to better forecast what type of content stimulates these converting words

are they always precise product titles?
do they always contain a certain terminology?

maybe, as you said, you don't worry about them at all.

take a look and see if they are all expansions on those terms you thought would convert. It is possible that increasing your percentages on those broader terms will increase the frequency of unique converting searches.

[chewy]

Can you detect any patterns? Thinks that make sense, things that don't make sense yet?

Do "converting terms" come into different pages than the usual suspects?

Can you see a lot of "typein" conversions?

Having looked at logs for dozens of clients, these are they type of patterns I look for.

Have someone who's been around look at the logs and reports. Often patterns emerge over time.

Also, for lack of better terms, there is what I refer to as the small number problem. If you have 10,000 real visitors per month with a 3% conversion rate, that means 300 customers. Many of those will are likely to be typeins from visits before so they are mysteries.

So if that is half, that leaves 150 - this will be split over 10 - 20 terms most likely. Subsequently 7 - 15 conversions per term can start to look significant. Sites w fewer visitors or different metrics show thin enough numbers so you can't really see the patterns, or be able to see causality.

Plus, there are seasonal and other complicating factors that can throw things off.

Do you use PPC conversion tools? This can often provide instant insight, but again, you've got to spread things out a lot and let things run for a while before you get enough data.

This would be a good thread to dig into ROTs (Rules of Thumb) that people use to determine causality.

What are your ROTs?