Paul Holstein and the Devil in the Details

[The last post I’m going to do on eMetrics speakers. I’m returning to Internal Search next and then I’m going to start a series on SEM Analytics. Full disclosure on this post, Paul's site, CableOrganizer, is a client of Semphonic.]

One of the sessions I attended at eMetrics had the promising title of ‘10 Actionable Things to Measure Tomorrow’. To be honest, I might not have taken the title seriously except I know the presenter, Paul Holstein. And when Paul says something is "Actionable," I’ll pretty much lay money on it.

I’ve always felt that Conferences (even the best) are long on high-level presentations about what great things can be done in the abstract and mighty short on detailed presentations about how to do those great things. I think Paul must feel the same way, because his presentation was focused on providing those one or two implementable ideas that might actually help make your business better.

The suggestions in the Presentation were very specific. Things like:

• Insure your pages load as fast as possible – use compression on pages and images
• If using Broad Match on PPC report on exact terms in web analytics to find negative keywords
• Check your web logs for 403 errors – places where a directory doesn’t have a default page
• Email yourself a warning whenever a broken link error page is served

As I went through the list (there were actually 12 not 10), it struck me that I knew about almost every one of them intellectually, but that most of them were things I’d never taken the time and trouble to do on our site.

I don’t think that’s unusual. And I’d be willing to be that most of us who walked out of Paul’s presentation still haven’t yet, and probably won’t ever, take most of his suggestions.

Why is that? Part of the explanation lies in my basic theory about human nature as expressed in the business world. Namely, that we are all lazy and we never do anything well unless we have to. Semphonic’s web site is much, much less important to us than Paul’s website is to CableOrganizer. For CableOrganizer, it is the business.

But I’d be willing to bet that for even for company’s where the web site is the business, many of Paul’s suggestions will never be implemented.

Take one of the simplest: whenever CableOrganizer generates an error page from a broken link, the site sends an eMail to Paul letting him know. Sure, they could catch these problems by reviewing web analytic reports. But by then the problem might have existed for days, weeks or months. Do you want these problems persisting for days?

Same story with checking for 403 errors – most commonly caused by places where users have tried to access a directory without a page name appended and the web server has no default page. This is easy to fix. It’s a small problem. On most sites it probably accounts for less than 1/10 of a percent of volume. Does that mean it shouldn’t be fixed? Especially when the fix takes about a minute? I’d actually recommend not even checking logs if you don’t use a weblog solution – just make sure that every directory on your site has a default page specified.

I was talking with someone at eMetrics about the presentation and she was surprised that he uses web logs and noted that he seemed awfully hard on Omniture. I explained that CableOrganizer is actually an Omniture customer. And if Paul’s hard on them, it isn’t because he doesn’t like the product. If he had HBX, he’d be hard on that. If he had Webtrends he’d be hard on that. When you are really using a web analytics package, you’re going to have complaints about it. Doesn’t mean it’s not a good package, it just means it isn’t perfect.

And how many of us will take the trouble to use weblogs to find the few things we can’t get from our enterprise-class tag-based solutions? Not me usually!

I mention all this because I think it drives to two very important points about web analytics in general.

First, web measurement isn’t always about finding new problems. Sometimes it’s about reminding us of problems we should and do know how to solve but have never taken the time to take care of. This is a perfectly respectable use. One we shouldn’t discount just because it isn’t sexy. We often talk about web reporting as if it’s only function is to help us find big new problems. But think about those newspaper reports that detail how long a City government has let some infrastructure problem fester. After they come out, the problem usually gets fixed. You can laugh, but it works the same way in business.

The second point is that most valuable ideas aren’t big ideas. They are small and simple. Most of the actionable items on Paul’s list dealt with items likely to make only a small difference in overall site performance. But if you keep knocking away the small stuff, eventually they are going to add up to pretty substantial difference.

When you are on the web, you’re competing with lots of people like Paul who are sweating all the details. If you want to compete, you have to be willing to find and deal with lots of small, annoying very un-glamorous problems. Chances are that solving those problems won’t get you a guru-slot at eMetrics. But it will get you a significantly better site.