What Analytics Can Tell You… And What It Can’t

In a bricks-and-mortar store you can learn a lot just by watching the flow of customer traffic through the doors and around your premises.

Website Analytics tools aim to provide the same kind of visibility for webmasters and business managers, watching what's going on on their websites.

Flat illustration of expert with control panel. Analytics and management - vector illustration

Web analytics tools will give you a lot of stats and charts. But it's vital to approach analytics with a clear idea of what they can and cannot tell you.

The ubiquitous free tool is Google Analytics, often shortened to GA, but there are other solutions out there, some of them replicating what GA does except more free and open-source e.g. Piwik, and some of them cloud-based tracking tools focused on more detailed behaviour measurement such as Clicktale.

Regardless of the tools, in this article we take a look at what analytics can and cannot tell you about your website and website visitors. It's important to have the right expectations, come up with good analytics questions to ask, and know where the limitations are.

1. Analytics will tell you how much traffic you have.

Often, this means you are seeing how you don't have enough traffic!

In analytics you can track where traffic is coming from, and verify that indeed magic is not a channel.

Notice that what analytics can't tell you is how to get more traffic.

With the right expertise, analytics will throw up lots of ideas for things to improve within your site, but ultimately it's not a magic source of marketing insights about how to become more popular with your target customers.

2. Analytics tell you what content is keeping people on your site.

Often this means the reverse: what content is correlated with people not sticking around on the site, i.e. bounce rate and exit pages.

Notice that analytics can't tell you why certain pages or content are working or not.

3. Analytics tell you what people do on your site.

For example, which pages they visit, in what common sequences, and also what links and buttons they click on.

Notice that analytics can't tell you who the visitors are or what they want.

When it comes to these people's motivations and intentions, you can only guess. (Unless you start creating conversations, e.g. via live chat. But you'll never get truly objective 'data' about these subjective things you want to know about customers' thoughts and behaviour.)

4. Analytics can show you where people have usability trouble on your site.

With a bit of detective work you can usually identify major usability stumbling blocks. These manifest themselves as users repeatedly refreshing a page, failing to visit a link you thought was obvious, or giving up and exiting at a crucial stage such as a checkout final screen.

Notice that analytics can't tell you much about how to fix it: you can perhaps identify problems, but analytics doesn't magically offer up a specific design solution to fix it.

This issue applies to mobile usability especially: you will see a lot of issues such as lower average time on site and higher bounce rates, but it's a non-trivial question how to design your interfaces so they scale down successfully for small mobile screens with touch navigation.

The more interaction you require, such as ecommerce or form filling, the more you will see obvious problems but no readymade solutions. As you use websites yourself on your mobile devices, keep track of interfaces that you think provide good inspiration for your own mobile usability/design.

5. Analytics can show you which advertising campaign is doing better than another.

With the right tagging set up, you can identify statistics about sources of traffic from different marketing campaigns, and see how they stack up.

The first requirements are to ensure you have good campaign naming consistently done by everyone creating marketing campaign links (just use UTM tagging under a company naming convention), and decide on an attribution model.

However, notice that analytics out of the box don't tell you how much your campaigns cost.

Measuring campaign performance is ultimately going to be about cost-efficiency, so your analytics stats need to be in combination with good cost tracking.

Analytic information, info graphic and development website statistic.

When tracking marketing campaign performance in analytics, ensure you have a consistent company-wide policy of link tagging.

6. Analytics shows you what devices people are using to look at your site

...But crucially analytics will not reliably tell you whether these visitors are using different devices in combination.

And even if you work on ways to track unique users across devices, analytics can't tell you how to take advantage of this knowledge!

The same issue goes for repeat visits: analytics provide you with interesting-looking statistics, but commonly there isn't much actionable to come out of that knowledge. See points 2 and 3 above, you might correlate certain content with higher returning visitor rate, but it doesn't directly tell you why or what those users want.

7. Analytics measures your direct traffic

...which is great but what actually is "Direct"? In essence it just means source is untracked.

So analytics can't tell you where some of your most valuable traffic comes from and why.

This is similar to tracking organic search volume in analytics: you look at changes in traffic quantity but analytics can't tell you why. It is a common and entirely fruitless discussion as everyone sits looking at a report wondering "seasonal?" "market trend?" "competitor?" ...and for sure your analytics tell you none of that.

8. Analytics measures your traffic from organic search

...which is great but unfortunately you get next to no data any more about which search terms (keywords) brought people to your site.

To track organic search term performance you will need to invest in some different tools, and in analytics your best bet is to segment by Organic Search originating sessions and then look at the landing pages. This at least tells you which pages / topics pull people in.

You should also connect your Google Analytics with your Google Webmaster Tools (aka Search Console) data as there are some data in there about which keyword searches your pages are showing up in, and some average ranks. Again, it doesn't tell you why, or what to do about it, but it might help prioritize which themes to work on for your content marketing.

9. Analytics may indicate which pages load slowly

Some analytics tools will help you identify slow-loading pages. Assuming it is just some isolated pages and not your whole site performance being slow, this is almost always simply because of multiple large un-optimized images.

But in general analytics can't tell you the true measure of performance, i.e. the speed/responsiveness for actual human users in their different browsers in different locations on different ISPs.

So this is one area you don't want to wait for analytics to signal problems: get proactive about making everything as fast as possible. After you've tackled the easy stuff then you can assign a techie to pull their hair out trying to get a good score in Google PageSpeed Insights.

As for your users, their internet connection will always be slower than you expect, and their time to lose patience, faster!

10. Analytics gives you nice aggregated graphs, percentages, and trends

But what analytics can't do is the opposite: drill down to a single individual user.

In Google Analytics and others, the anonymity is by design. So despite the fact that you can engage and identify your users through very routine interactions like live chat, support tickets, logins, and transactions... there is no ordinary way to connect that back to analytics type data.

If you are asking those questions, such as "which marketing campaign did Mr Smith come from?" then you will need to post analytics type data into your CRM. Before you go too far into such awesomeness though, consider what (if any) actionable improvements you would be able to make from that knowledge.

As with any metric or report, if you are actually going to care about the data and make improvements based on it... then go ahead. But don't fall into the trap of measuring and reporting for the sake of it - something Rory Sutherland of Ogilvy memorably called "measurebation"...


 

Bonus list 1:
5 things Google Analytics doesn't do out of the box but you should consider doing the extra work to set them up:

a) tracking click events such as accordions, javascript navigation, forms, and downloads (begin by implementing Enhanced Link Attribution)

b) time on page (a curious one that, but GA only measures time differences between different pageviews... meaning for example that a single pageview, i.e. a bounce, gives you no measurement on whether the user bounced after 1 second or 1 minute)

c) segmenting engaged visitors separately from general traffic (create your own definition of 'engaged' and add a custom segment: this is vital for comparing different traffic channels)

d) validation problems in forms (if you have to squeeze users through multi-input forms, or a checkout, set up events to tell you when validation errors arise)

e) people who did X also did Y (correlate behaviours using custom segments)

Bonus list 2:
5 things Google Analytics can't do at all and you should consider other solutions:

i. cross-device and cross-browser testing

ii. speed testing and performance monitoring

iii. attribution data passed from your site actions into your CRM - particularly if you want to make a serious stab at putting sales value on your conversions

iv. comprehensive phone call tracking (e.g. track which ad campaigns led to sales phone calls)

v. click tracking (i.e. exactly where users' mouse movements went and coordinates of where they clicked)

 


 

Thank you for reading this Analytics article from KnowledgePower

Contact us today to get help with advertising campaigns and tracking your KPIs using Google Analytics.

We also provide white label analytics reporting for other agencies.