The science of web analytics is still a young one. Think of it like the medical field circa 1850: surgeons no longer rely on the traditional methods of anesthesia (“Here, drink this stuff and bite down on this cork. You may feel a slight pinch.”), but still a far ways from surgical robotics and organ transplantation.

Similarly, while most digital marketing practitioners have moved beyond the analytics equivalent of the “booze and bite” technique, there are still some commonly reported metrics that, when I see them used, gives me a massive dose of nerd-rage. Some of these metrics persist because they are simply expected by clients or management as being an indispensible part of any web traffic reporting; others are reported because they’re in big important font sizes at the top of the standard reports in Google Analytics; and still more are used because it’s easier to provide an impenetrable wall of data then it is to segment, analyze and draw actionable conclusions from that same data.

From a professional standpoint, this last item irks me the most because it ultimately serves to devalue analytics as a whole. Crappy reporting will teach decision makers to ignore web analytics data.

This really bums me out and makes me a sad panda. L For the first time in the history of marketing, customer data is now in surplus; the deficit of information coming from traditional/offline campaigns has been augmented with an incredible, overwhelming mass of statistics – all within the space of a decade or so. If I may be permitted to generalize a bit, this rapid influx of new metrics, terminology & marketing channels can be a challenge to many executives/key decision-makers. With their teeth cut in the era of faith-based marketing reporting and analysis, the era of Google Analytics, social amplification and multi-channel analytics is a new world.

As such, there is a tendency for these folks to demand data in quantity, rather than in quality. There is a reliance on many sub-standard metrics that are not tied to real-world business objectives. There is the general expectation (reinforced by years of spreadsheets) that any digital channel analysis must be delivered via impenetrable Excel file. There is the delegation of web traffic reporting to the IT department. The word “hits” is used frequently. The role that analytics data factors into strategic decisions is curtailed, as is the role of the analyst within the organization – he or she isn’t able to contribute effectively or prove their value. They may not be looped into non-digital marketing campaigns. They are stuffed into lockers. The jocks from sales steal their lunch money.

Believe it or not, this isn’t entirely their fault. Any digital marketer worth their salt should make education a top priority; if not for my hippy-dippy, “come-on-gang-lets-help-give-analytics-a-good-name” reason, then for the whole “job security” reason. Good analysts prove their value to the organization with awesome, actionable reports that are tied to real world business objectives. When someone requests a report, they don’t ask “what numbers do you need?” but rather “what questions should this report answer?” They’re deeply aware of non-digital channels and offline campaigns, and understand how those affect online behavior. They can speak knowledgably on data that is typically associated with traditional marketing, such as branding, audience segmentation and targeting, and market feedback. As a great digital marketer, they avoid the temptation to be lazy and simply provide massive spreadsheets to the powers that be, in the hopes that the C-suite will look at all the lines of data exported from GA, go cross-eyed, and think, “Those are lots of numbers. Looks like they’re doing a good job.”

As digital marketers, we have an unprecedented amount of consumer intelligence at our disposal – all of which is ultimately useless without some strategic analysis to do it justice. As Spider-Man/Peter Parker’s Uncle Ben says in, “With great data comes great responsibility.” I’m pretty sure that’s what he says.

So to help you exercise that responsibility, here’s a list of some metrics that I believe appear too often, either because they’re entrenched metrics from early web traffic reporting, or because they’re on a lot of standard reports within analytics tools, or because in-depth analysis and measurement is really hard and these are really easy to export from GA! In their purest form, as top-level numbers without any context or segmentation, these metrics provide almost no insight.

Tracked within log files, a “hit” is a request for a specific file made from a server. For example, when this page loaded in your browser, a number of requests were made to the server in order to render the final product.

A holdover from the earliest form of traffic analysis, tracking the number of hits that a server receives isn’t a very accurate measurement of actual visitors to the site. Most pages contain multiple files that need to be retrieved from the server, so one visitor to a web page usually generates multiple hits.

Analyzing log files is still useful for technical reasons (for example, testing server load capabilities). Some traffic that can’t be tracked using on-page tagging tools (for example, search engine robots or visitors using browsers with JavaScript disabled) can be captured in server logs. But with the increased capabilities of many tagging-based solutions, it’s hard to recommend using logfile analysis by itself to make any serious business decisions.

For a given webpage, exit rate is the percentage of visitors that ended their session on a website immediately after viewing that page.

Unlike bounce rate (which is a great metric), exit rate doesn’t just apply to the first page that a user sees when they arrive on your website – it looks at all visitors to that page, regardless of their previous activity on the site. Since you don’t know what users were doing on those other pages, a high exit rate for a particular page doesn’t mean that it’s time for a redesign. For example, what if after a user buys a product, they’re sent back to the homepage? Most likely, they’ll exit via this page – driving up this page’s exit rate, despite the fact that they did what you wanted them to do!

Exit rate can be helpful when analyzing linear pathing through a website, such as checkout processes, surveys, or other situations where users are moving through a defined series of pages. In this case, looking at exit rates for the pages between point A and B can provide insight into roadblocks and drop-off points   

Across all users that came to a page or website, the average duration of their visit.

Without context, there’s no way to measure success using this metric. For example, what is a “good” time on site? If you say, “the longer the better!” that’s not necessarily true – perhaps the reason I spent 5 minutes on a product category page isn’t because I was entranced by the “groundbreaking navigation” or the “artistic product imagery”, but rather because I was confused by the “I-did-drugs-and-then-designed-a-website” aesthetics of the page.

It’s best to use this metric as supporting data; that is, don’t look at a particular value and use it as a measure of success, but rather use it as another data point while doing analysis. This becomes more valuable if you have a website that contains lots of interactive elements on any given page (although event tracking is the best way to capture that behavior). Try to see outliers here; spotting segments of traffic with a much higher or lower time on site/page than the average can lead to further investigations that generate actionable conclusions.

The average number of pages loaded by a user during a session on your website.

Very similar to #3, having a high number of average pageviews doesn’t mean that your website is converting well and making the cash register ring. And don’t listen to people that say, “The more pages people view, the more engaged they are!” Sure, if by “engaged” you mean “livid”, because all those pageviews are me being forced to click through an inscrutable product category structure thanks to a catastrophically broken internal site search. Alternately, maybe I only viewed 2 pages on your website before falling in love, picking up the phone and placing a $250,000 order (in this example, I am very drunk). In a vacuum, average pageviews really doesn’t tell me anything. 

Again, use context and supporting data. Combining average pageviews with direct customer feedback can be very valuable, especially for service-oriented websites. For example, many support websites contain a “Was this article helpful?” widget on each article. By looking at the average pageviews of users that clicked “Yes”, you may notice improvements that can be made to internal site search, or the structure/navigation of your site. Over time, you’d likely want to see this number decrease, indicating that visitors are more easily finding the correct content.