: Web analytics accuracy - log files vs JavaScript I've just read a claim on a web stats package site that suggests that log files generate more accurate stats than their JavaScript-based counterparts.

@Kristi941

Surely if it is a good-quality stats package, it would deliver accurate results regardless of whether it's JS, image-tag or log file-based?


If you're referring to the section headed "Get more accurate website statistics with log file analytics versus script-based software" then yes, it's a bit disingenuous. While you can say that a given server- or script-based stats package is or isn't accurate, you can't entirely compare across two such products because they're inherently not measuring the same things or even using the same source data. Interestingly, note that the text below that heading doesn't continue the "better/more accurate" argument. Obviously it's perfectly fine to compare two server or JS packages.

As far as trade-offs, it's generally accepted that bots for one don't have Javascript interpreters and so won't show up in JS-based stats. Then again, most people arguably don't care to see them. Anyone with JS turned off in their browser or with JS blockers set up will be similarly invisible to JS packages, but do show up in log analyzers; that population is generally accepted to be small enough so far that it doesn't do any major skewing of the numbers. And as pointed out, any pages that happen not to have the JS code inserted for whatever reason will simply not be recorded, or even give any indication they're not being recorded unless someone goes poking around and notices.

Vote Up Vote Down