So, it’s been about a year and a half since Google rolled out real-time analytics. A year and a half where analysts, developers, marketers, and webmasters (remember those people? Whatever happened to them?) spent hours glued to their monitors watching little bars move across the screen. And then maybe three of them did something productive with what they saw.
The truth is, there are a very few specific instances where real-time analytics are useful, and even fewer cases where they have a clear advantage over standard time-delayed analytics.
We’re going to talk about those edge cases today, and help you decide whether you need to ever bother clicking on that real-time button. Read on, brave number cruncher, read on!
What’s the Difference Between Real-Time Analytics and Standard Analytics?
Well, obviously, the data is coming in in real time. Duh. That’s not the only advantage, though, and to understand where the other strength of real-time analytics (RTA henceforth!), you have to understand how most major analytics suites work to give you deliciously actionable data.
First, you have to know that 75% of what you see in your analytics report isn’t technically true. Not in the objective, philosophical “This is the way the world is” sense of the word, anyway. A large chunk of data given to you by your analytics software is “statistical” true. This is because most analytics software doesn’t actually look at every single one of your individual visits when calculating a large number of metrics, at least not at the top level. Instead, they take a sample of visits, and then extrapolate from there for things like time on site, bounce rate, etc. They’ve gotten much better at it over the years, to the point where the metrics are starting to be replaced with actual factual aggregates of every individual visitor, but there’s still a lot of calculating going on behind the scenes.
So not only does RTA give you immediate numbers, it also gives you more accurate numbers. This is an important advantage in some cases.
Second, you have to know what the time delay on your standard analytics views is. Keep in mind that this varies from number to number. For example, Google Analytics is pretty good about giving you up-to-date information on things like number of visitors, source, etc. The delay there is roughly 2 hours, give or take. Other things, like time on site, conversions, events, etc. can take longer, as much as a day. The SEO section is delayed by two days, always, as Google takes that long to confirm, verify, and extrapolate query impressions and the like. So while you can see how many visitors you’ve had today, it might be a while before you can figure out exactly how many of them converted.
When Is RTA Better?
So, with those things in mind, when IS RTA better? Mostly when you desperately need to know exactly what’s going on RIGHT NOW with usability metrics. Usibility metrics, like bounce rates, time on site, and conversion/goal/event completions are usually the most delayed and the most statistically manipulated numbers found in standard tracking packages. When you need to know if your users are ABLE to use your site, and HOW WELL they can use it, RIGHT NOW, is when RTA comes in to play. So let’s say you’re rolling out a new conversion form: you want to see people as they get to it and go through it immediately, because if any technical or usability problems arise, you need to know right away to fix them. Same with new page, redesign, and other such initial deployment scenarios.
Whenever you’re rolling out a new product, the ability to get immediate feedback can save you hundreds or thousands (and maybe even hundreds of thousands, depending on how big you are and how bad the bugs are). Especially if you don’t have a long beta period. Without a long a diverse beta, being able to sit an engineer in front of an RTA display and have them monitor for any suspicious activity for the first couple of days can mean the difference between a great new site improvement or a post-launch nightmare.
The other great case when RTA makes sense is when you suspect that there might be a technical or usability issue with your site, but you can’t quite pin it down or replicate it. Now that Google lets you use standard and custom filters and segments in their RTA implementation, you can (for example) isolate your traffic from visitors using Internet Explorer and figure out why they’re converting at such lower rates. If you combine that with a pushable live chat solution, you can follow customers who are experiencing issues and intercept them right at the problem area, giving you immediate feedback on what went wrong where. This is pretty major, and can help you troubleshoot problems that have been killing your Quality Assurance, Development, and Marketing teams for months.
The last thing I would suggest using RTA for is monitoring for weird traffic surges and one-off events. Things like getting retweeted by a major influencer in your field, mentioned in the New York Times, lambasted in Congress, etc. But this isn’t something that requires steady monitoring, and is best handled by alerts, anyway. One thing to keep in mind, though, is if you do have a rogue traffic phenomenon, RTA can be great to identify the path that this rogue traffic takes and see if you can interrupt it to make it go to a place that’s more likely to convert. So if you have a blog, and you see one post getting hundreds of visitors from Inc all of a sudden, it might behove you to edit it slightly to get your visitors into your funnel. Or throw a quick landing page up somewhere. Or a little message bar to welcome the new traffic boon (check out our friends at Spinnakr, they make this easy). Find someway to disrupt their natural pathway and throw your messaging in there.
When SHOULDN’T You Use RTA?
Pretty much every other time. The thing with analytics is that they mean almost nothing outside of the larger context. Your bounce rate on any given day is absolutely meaningless. Nine times out of ten, you aren’t looking at metrics: you’re looking at CHANGES in metrics over TIME. The one time out of ten that the isolated metric is critical is either when you’ve just changed something (see above), or when you’re contemplating changing something. And even in the latter scenario, it’s far better to look at trends and averages over time and over large numbers of visitors. Staring at your RTA and trying to make good decisions based on a small sample over a small time frame is the surest way to over-react to statistical noise and make poor business decisions. Don’t do it.
So while real-time analytics are important, and should be familiar to every analyst and marketer out there, I think we’ve spent way too much time over the last year and a half going crazy about it. Good things come to those who wait, haste makes waste, patience is a virtue. All of those things are doubly true when you’re trying to make guesses and inferences based on incomplete data (which is all any kind of analysis is – it’s very incomplete data, even at the best of times). Making that data even LESS complete is a critical mistake, and one that we shouldn’t be making. So turn off the monitor, step away from your desk, and go get some fresh air. I promise your data will still be there when you come back, and it will be even better than real time.
And since I’m sure I’ve left off or forgotten tons of valuable use cases, I’d love to hear all of the awesome (or disastrous) ways you’ve added RTA to your well-balanced online marketing program!