To whom is this information useful?
This information might be useful to anyone who is responsible for the success of a website within a business, and who wants to know how to measure that success and share it with colleagues or senior management, in order to justify investment or improvements in the overall digital strategy of a business.
The vast majority of websites I come across have Google Analytics (GA) enabled, but I commonly find that not much time is spent in using the tool to look at how the website is performing.
Lack of time, or unclear internal accountability and responsibility are sometimes reasons for this. Most often though I’d say it comes down to the fact that it’s unclear what kind of information should be measured, who needs to know and to what use will that data be put. If this sounds like you and your organisation, then maybe this post will give you some ideas of where to start. Just to be clear, this isn’t a ‘How to use GA’ tutorial — it’s more a ‘why measure what we measure’ piece.
This list of eight points below should help you establish what your Key Performance Indicators (KPIs) are. Once you know these, you can make sure they’re fully understood across your business and use them as a means to show where your site is/isn’t working and what the opportunities are.
In this article I assume that you are not using Google Adwords, so I won’t be addressing measuring performance of Pay per click (PPC). I’ll also make suggestions as to where else you can get key data from.
Finally there’s the importance of being able to benchmark all of this. Your bounce rate is 40% — is that good or bad? It depends on the context: the type of site you have, its content, the sector you are in. GA will let you benchmark against industry sectors, but the context of your business may render this irrelevant or misleading. To start with the best thing to benchmark against is your own business. What do month-on-month (MoM), quarter-on-quarter (QoQ), year-on-year (YoY) comparisons tell you? How did performance of your site change based on changes you made to the site, the business, or how you promote it? (e.g. newsletters)
1. General audience
The primary measure of audience is the amount of absolute sessions generated by an absolute number of users. Of course, in measuring each you need to know how GA, or your tool of choice defines a user and a session.
For each of these I suggest measuring either MoM or QoQ depending on how much time you have to allocate to reporting and analysis, as well as YoY. It’s also useful to try and get an idea of seasonality, patterns of usage across the week and busy times of the day.
- At miggle I primarily report on sessions as opposed to users as a key indicator of how heavily the site has been used
- Pageviews per session in most cases are a key indicator of user engagement — the higher that figure is, the more engaged are users
- Pageviews tells us how much content has been consumed. If you are selling audience, the amount of pageviews you have creates the opportunity by which you can monetise that audience, because adviews (which other systems will report) depend on pageviews.
- The higher your pageviews per session are, the lower you’d expect the bounce rate to be. The two are not directly related, but bounce rate is the percentage of users who disappear having just consumed just one page. In most instances this would be seen as a negative metric the higher the figure. Because some pages might have an acceptably high bounce rate, it’s often useful to look at bounce rate by site section, or even on individual pages, like homepages or hub pages, rather than as an aggregate.
- If time users spend on site really means something to your business — for example, if you have a lot of video or audio content (particularly if that’s monetised pre or post role) — the average session duration is worth measuring. If it doesn’t have a direct impact, don’t measure it. For example, all I care about on the miggle site is that ultimately prospects get in touch with us. There’s too loose a relationship between the aggregate average time spent on site and completed contact forms for me to need to waste time measuring that.
- New sessions. Maybe important to measure if your site is all about user acquisition, because you want more new users coming in who you can convert to engaged or registered users. Otherwise, I don’t think it’s so important to track.
How you measure audience will depend on who your audience are. Other tools beyond GA can help you measure audience — like registrations tracked via your Content Management System (CMS) for example.
It may be worthwhile measuring active registered users as a percentage of all users, defining a period by which you deem them active — i.e. have logged in within the last 6 months.
2. The makeup of the audience
The makeup of your audience may not vary much over time, so this might not be something that you need to measure regularly. Also, you may not be overly concerned with user demographics. That aside, it’s still a good idea to have a profile of the different types of people that make up your website audience. This can be particularly useful if you are trying to establish personas.
Knowing the location of your users can also be beneficial. In terms of the key regions and languages you want to serve, you should measure the extent to which those users are sufficiently engaged compared to the average overall use — for example, bounce rate should be lower and pageviews per session higher.
3. Where the traffic comes from
The key sources of your traffic are:
In GA it’s easy to get misled by this, as direct also takes into account traffic that can’t be attributed to or doesn’t identify its source. In its purest form, direct is people typing an actual URL from your website into their browser and going straight to you — not via Google or anything else. If the unidentified or accountable traffic could be taken out of direct (or you can explain its origin), if consumer recall of your URLs is likely to be low (that’s most sites), or if you don’t tend to do much offline promotion, then you’d expect direct traffic to be low in terms of its overall percentage share of acquisition.
A lot of sites would expect the bulk of their traffic to come from search, which in most instances means from Google. GA specifically will count as search any traffic from Google organic search (i.e. the search results for which you don’t pay directly for presence) as well as Bing and Yahoo. It will also count paid for search — i.e. that generated by Google Adwords.
For a number of security and privacy reasons it’s harder these days to get a breakdown of what specific search terms generated traffic to your site. Tools like Google Search Console, which you can integrate with GA, help — and using this is key to measuring how your site is set up from a search engine optimisation (SEO) perspective.
That said, the prevalence of Google and the extent to which search is baked into the functionality of browsers means that the percentage of traffic that your site gets which is attributable to search will likely be made up of a huge amount of searches made by people who already had some awareness of your brand and may just have typed the name into their browser. It is effectively more like direct traffic than the unattributed traffic I talked about before.
Did people searching for you already know about you?
Here’s a hypothetical example to demonstrate that, but based on a real business called colourmesocial.co.uk.
Colour me Social focus on the optimisation of their clients’ social channels to ensure that their social platforms are perfectly placed to increase awareness of their business. They don’t have anything to do with colour as such beyond the obvious metaphor (which I like, btw!).
Now, let’s say I take care of their SEO strategy for them. I tell them I’m great because 80% of their traffic comes from organic search.
However, 80% of that 80% is for searches like colourmesocial.com, colormesocial.co.uk colour me social uk etc. i.e. near guesses at the URL or misspelling of the word colour (yes American friends, color is a misspelling…). Only 20% of it (so 16% of their overall traffic) is related to searches around businesses offering social media management services. Am I doing my job? No. Whatever is driving their brand awareness should take the credit and Google is doing the rest by working out what the user meant.
For your business you need to work out what you want to achieve in terms of measuring search and the associated SEO implications. Maybe as the marketing manager of colourmesocial.co.uk you want to optimise around the word colour and colourmesocial.co.uk, but before doing so you need to be aware of why. Don’t let your SEO agency attribute their success to the otherwise general strength of your brand.
Measuring the effectiveness of paid search is obviously only relevant if you do it. And to get it working effectively means many factors need to be considered — so it’s not an area I’ll cover here.
Building up your percentage of referrals is probably the hardest thing to do, but in terms of boosting your site’s credibility, referred traffic is probably some of the most valuable traffic you can get — assuming your site has been referred to for positive reasons, obviously.
In terms of the things you directly control, you can go some way to getting links pointing to your website on directories etc., but to do genuine link building takes time. Unless you are setting this as a strategic objective, it’s probably less important to measure this on an ongoing basis. What might be more useful would be to look at some of the sites that referred traffic to you, and try and establish why they did so.
Knowing what social media channels work for you and what your social reach is is pretty key, as it will help you determine how much effort you put into creating, managing, optimising and sign-posting content on the right channels.
If you undertake email marketing you’ll want to measure the extent to which this activity drives traffic and converts on your website. Obviously, in the same way that you can optimise your PPC campaigns or social media engagement, similar tools will exist in the software you use to manage your email marketing.
The same as above — but we are talking about adverts instead of PPC, social or newsletters.
4. The content they consume
It’s useful to get an idea of what particular page content is popular on your site, but given that the popularity of page content will change over time, there might only be specific pages you want to measure on an ongoing basis, like your homepage, or key section pages
What is more useful to measure on a periodic basis is the extent to which certain sections of your site are popular — like everything under ‘About Us’ for example. You will find differences in engagement in terms of pageviews per session and bounce rate on a section by section basis. This can give you some useful insights into the quality and relevance of your content in each section. You can ask yourself questions such as ‘are our users finding the content that they are looking for?’ or ‘once my users have engaged with one piece of page content, are there clear options for them to start another user journey after that?’
Often when I analyse a client’s site traffic I am able to point out how poor engagement metrics in one section compared to another, alongside content examples we pull out from those sections, demonstrate a clear requirement to improve that area of the website.
The most obvious way to measure conversion is to set up goals. A goal is basically an end-point of a user journey, — it might be completing a purchase, or filling in a form. The starting point might be from an initial call to action (CTA) which was on an advert on another website for example, or a click from a listing on a search engine result page (SERP) .
GA also tracks a number of events as well, which can also be used to measure conversion. Typically these might be things like the number of downloaded documents. These can be extended by using Google Tag Manager to track more custom events, such as engagement with video — e.g. seeing who played 25%, 50% or the entirety of a video.
Finally, it’s possible to set up e-commerce tracking as well, and see what user journeys through your site resulted in purchases. E-commerce tracking, as well as placing monetary values on the completion of goals, can be used to help try and establish what user value looks like.
6. The technology they use
It’s useful to be able to look at how usage of your site is split across various devices, namely desktops, tablets and mobile. The relative ratios aside, the key thing you want to get an understanding of is how behaviour on devices vary.
- Bounce rate and page views per session are likely to be different, particularly between mobile and desktop. You will need to determine whether this is a good thing or not.
- Users may favour certain devices depending on whether it’s a weekday or weekend day, or at different times of the day. You should expect to see a different ratios of traffic acquisition per device.
- Does organic search traffic hold up as well on mobile as it does on desktop? If not, what does this mean in terms of your mobile content strategy, your mobile user experience or the extent to which your site is optimised for mobile?
- If email newsletters and social media engagement are a key part of your traffic acquisition strategy, how does this also vary on handheld devices, given the extent to which users are engaged with both their emails and social accounts on their smartphones?
Browser, operating system and screen resolution
It’s also good to have a clear idea of what sort of operating systems, browsers and screen resolutions that your audience are using in the main to access your website. It could be very different from the set up would you and your colleagues base your decisions on.
For example, you may work mainly on a Windows desktop in your office using Chrome on a site which has a predominantly young audience who are in the main using Safari on an iPhone. When you are conducting quality assurance (QA) or user acceptance testing (UAT) this data can be invaluable in creating a browser and device matrix against which you test your site.
When I tend to analyse the sorts of technology used to access a site I don’t usually look back any further than three months. Technical landscapes change quickly and while it might be useful to show that change QoQ to make sure site developments and investment are keeping up with shifts in technology, in the main I’m generally only interested in the technology that people are using to access sites today. The technical landscape will very definitely vary between different types of audiences and industry sectors, so don’t assume a picture for one site can be assumed to be representative of another.
7. How the site performs
Tools like GA are not necessarily the best ones on which to analyse performance of your site, but you can get some useful insights and in terms of page size and speed of download. At miggle where we’ve seen anomalies in this, it often points to areas of where content management functionality can be improved. For example, if it’s possible for a content manager to make a single blog post which is over 100MB, then something probably needs to be put in place to warn against that happening.
8. Things that are wrong
Again GA may not be the best tool which you can use to establish whether there are issues with your site (i.e. hosting tools, link checkers, code validators etc. should be used) but even so you can get some pretty useful insights.
Unfortunately tools like GA are susceptible to being spammed. The most obvious place you’ll often see this in in your referral traffic. There are filters you can put in place to stop this and while it’s worth having a view which collects all data, spam or not, so you can identify problems, it’s also worth having a view which filters unwanted traffic and using this for your reporting, so you can be clear that you are measuring human activity.
Again while it is worth having a view which collects all data, it’s good practice when reporting traffic to filter out your offices and your agencies’ offices, by IP address, IP range or network identifier. Alternatively you may even want to have a view which measures internal usage — like of a staff extranet for example — and then measure this over a period of time.
Broken or inaccessible pages
Once more there are probably better tools than GA to measure pages which generate 403 (access denied) or 404 (page not found) errors, but from a product quality perspective this might be something you want to measure on an on-going basis, with a view to reducing them.
1. It is worth establishing some simple KPIs you can measure, compare and report on regularly.
- Share these with the business and use them as a means to measure success of and justify ongoing investment.
- Don’t bite off more than you can chew. Focus on ensuring you can allocate enough time on an on-going basis to start raising the importance of KPIs in your business, even if that time means the KPIs are limited to start off with. As interest starts to grow, so will the list of things to measure, against which you can pitch to allocate more time.
2. There are a number of other elements you can measure which will help you understand who your audience are — although this may not shift so much over time.
3. There are things you can measure which give you insights as to what technologies people are using to access your site. This will shift as technologies improve, and the extent to which your KPIs improve will depend on how well your site can be accessed on those technologies.
4. There are things you can measure which will help you improve the technical quality of your website.