Use Analytics

Pick The Right Medium
July 29, 2019
A Brief Introduction to Search Engine Optimisation
August 21, 2019
Show all

Use Analytics

use analytics

Use Analytics

One of the most useful things about working with digital marketing is that its effects are very transparent. In general, because your website often has to be contacted whenever a page is viewed it can count the number of times that page is viewed, and also because it can serve software that monitors how a user interacts with a page in the same response that it uses to serve the content, a very fine-grained view of how your website performs can be created.

There are lots of different analytics platforms available, but two of the most popular are both provided by Google for free: Google Analytics, which looks at how your website is performing, and GoogleAds which has a built-in analytics package that examines how well your ads are performing.

I’m going to go into analytics in more detail in a separate set of posts about SEO and Pay Per Click marketing, but for the purposes of this series of blogs it is enough to say that  using analytics is crucial to effective digital marketing and so you should find a good analytics package and make use of it.

Why is Analytics Useful?

A key presumption of search engine marketing is that people who are searching for a product are thinking about buying it. There may be people out there who like hoovers so much that they spend their time browsing them, but most of the time I’ll wager that you only go online to view hoovers if you need a new one. But this doesn’t mean that someone looking at hoovers wants to buy one right now, they may be simply trying to decide which one to buy or who to buy it from (ultimately, they want to answer both questions). The first thing to appreciate is that when people buy things they tend to follow a process which involves solving a series of small problems in the quest to solve the larger problem of what they want to buy. This is referred to as the consumer purchase process, and forms a key part of the customer journey.

If the consumer purchase process represents a series of problems that your customer needs to solve before they can make a purchase, then your goal is to present your business as the solution to those problems. How you implement this is specific to the product you offer, but the basic process is always the same: You work out what those problems are and then you work out how to provide solutions for them. You may have to do some research, like speaking to your customers or conducting a survey of visitors to work out exactly what they want from your website. The goal is to expand your customer journey to specifically look at how they use the internet to help them to purchase goods and services that you offer. Ideally, you want to try and create a journey for each customer profile, if only to ensure that there are no differences between them.

At this point you can then consider how well your website currently performs, and what you can do to make it perform better. The online journey is like a schematic for your marketing: For each step, your marketing should have a corresponding element. The online journey is like a schematic for your marketing: For each step, your marketing should have a corresponding element.

For example, let’s say that you know there are two stages that the customer goes through when thinking about a new hoover before he picks one. He looks for different models to consider and he tries to find out what independent experts think of them so that he can decide which ones to focus on.

These actions represent two different problems your customers want to solve before they will buy from you. If he arrives on your site to see how much different kind of hoovers cost, but then leaves to go and look up reviews on another site then you may have lost him – especially if the competition advertises on the other site. So, ideally, you want him to be able to look at hoovers and read expert reviews on your site. 

Such a solution could be a webpage that lists all the hoovers you sell, and for each listing have a link which takes you to an expert review that is hosted on your site, possibly syndicated from a publisher. Not only does this solve the problem of finding a review, but the customer doesn’t have to even look for the review, it’s presented to him right next to the product. Now the customer can review and compare the different kinds of hoover available without leaving your site, which means they remain focussed on you for longer rather than leaving for a competitor.  

How Does Analytics Help?

Analytics tools provide an excellent support for these kinds of decisions. With them you can:

  • See where your customers come from
  • See how they behave on your site
  • Set goals, i.e. a purchase or newsletter signup
  • Define the customer journey as a “goal funnel”

The exact methodology used will differ between individual analytics platforms, but there are a few standard metrics and concepts that are worth a little discussion at this point:


This is the number of times a page has been viewed. More specifically, it is the number of times it has been served to a user. It can be useful in terms of understanding which pages on your sites receive the most attention, but on its own it is not terribly informative. This metric is sometimes referred to as “hits”, but you should be aware that “hits” also refers to a metric produced by your server which tells you how many times a page has been served to a client – which may or may not be human.


In very general terms, an impression occurs when something on a page has been shown to a user. So, if you have a blog post on your website with a link at the bottom to another blog post and a user visits that page then the visit itself is a pageview, and the link has been impressed. It is more specifically used when referring to things like adverts being shown to users, websites appearing on lists of search results served to users, or something appearing in a social media news feed, i.e. the number of impressions your advert gets is the number of times it has been shown to a user.


A session refers to a period of time that a user spends on the site, regardless of how many pages they view. If I spend 10 seconds on your site, that would count as a session, and similarly if I spend 10 minutes on your site that would count as a session too. Some platforms, like Google Analytics, will continue to count a session as “open” for up to half an hour of inactivity meaning that if I go to your website, leave to visit a competitor, and return within half an hour, then that will count as a single session. The duration of a session is an important metric since longer sessions indicate more interest and commitment to the site’s content.  

Unique pageviews

The number of visitors who have viewed a particular page. If a user visits your site and, in the same session, visits the same page three times, this will count as one unique pageview for that page.

Landing Page

The page that a lands on when entering the site

Exit Page

The last page a user looked at before leaving the site.

Time on Page

The amount of time that a user spends on a given page. This metric is useful in helping you to understand how engaging your content is. Users who spend a lot of time on a page are probably interested in its content, so a page which people spend a lot of time on is probably engaging.

Bounce Rate

If someone visits your site, and then leaves without visiting any other pages, then this is referred to as a bounce. The bounce rate is the percentage of users who enter the site and go no further.

Click Through Rate (CTR).

The Click Through Rate tells you the probability that someone who sees a particular link or advert will click on it. It’s usually expressed as a percentage. A high CTR indicates that the advert is attractive and interesting compared to its competitors, and a low one implies the opposite. In advertising terms, Click Through Rates tend to have low absolute values; a click through rate of 1% can be considered high for very competitive keywords.

Goal Funnel

A series of pages which a consumer goes through in order to reach a particular goal. The funnel analogy comes from the idea that some customers will drop off on the first page, some on the second, some on the third and so on until the remaining few complete the goal.

All these different measures can give you an idea of how well your website is performing. However, analytics platforms can generate a gigantic amount of information and it can be difficult to extract meaningful information from the site. It is important to spend time working through the data to check that the headline figures are really telling you what they think you are. A good way to approach this problem is to break the data down into smaller chunks by applying “dimensions” to it. Dimensions are ways of categorising your traffic according to aspects which your site cannot affect, like the geographic location, the types of device they use, or the keywords they typed in to find it.

Take the bounce rate, for example. Your homepage may have a very high bounce rate, and from that you may jump to the conclusion that your homepage isn’t relevant to the people who are visiting the site. On the face of it, this is the correct conclusion, but you still have to be careful about presuming that it is the whole story. It might be the case that your site ranks very well for certain keywords that are irrelevant to its content. This will mean that you attract users who are not looking for your site at all – and they will bounce. On the other hand, those who came to your site using keywords that are relevant may be highly engaged but not of sufficient numbers to make the bounce rate appear successful.

The way to approach analytics is twofold. Firstly, learn how the system you are using works and what all its different reports tell you. Google offers free training for Google Analytics, for example. Then, return to your customer profiles that were discussed in an earlier post and try to consider what the reports tell you with them in mind.

The kind of questions you can ask yourself are: What kind of experience are these people looking for? What keywords will they use to find your content? What kind of devices do they tend to browse with?  What sort of imagery will they respond positively to?

For instance, if you know the sort of keywords your customers are likely to use then you can narrow your report to only consider traffic that arrived on those keywords. If you know that they tend to prefer to use a desktop computer to a mobile phone, then you can narrow the reports to only look at people who use desktop computers and those keywords. Then you have a much better idea of how well your site is performing for one subset of the people who buy your products.

It’s important to remember that analytics data is representative of real people, and so you have to look past the numbers and try and understand what they are telling you about the state of mind of the people who are using your site. This is where the art of interpreting the data comes in, and although it takes time and experience, it is well worth it. If you view your site as though you were a user rather than an owner then you can start to think about how to make the site better for the people who want to use it, and the more you learn to think as they do the more effective the decisions you make will be.


It can be difficult to know if a change has had a positive effect, or if there is some other factor that is affecting your site’s performance. The way to approach this problem is to use a system like Google Optimize to run experiments on your site, or use Google Ads to run experiments with your adverts. The most basic type of experiment is called an A/B split test. The idea is that you decide that a change would improve a page’s performance. You tell the A/B split testing system that you want to test this change. The system alters your website’s behaviour such that some of your customers will see version A of a page, which contains your change, and others will see version B, the original page. The Split Tester will track how each version of the page performs and show you the difference, which means that other factors are controlled for.

The split test will show you how well your new version performs against your old version, and if it shows an improvement, use the new version in place of the old one.

Back to the Garage

In my last post, I used a car breakdown service as an example of a service with a very simple customer journey. We deduced that the minimum viable strategy was to use a Google click to call ad linked to a Google places account. So how could we use analytics to help us optimise the effectiveness of the advert?

The first thing to do is to write an advert, for which you can gain inspiration from careful thinking and looking at what your competition is doing. Then you set the advert live and Google Ads promotes your advert.

Each time someone searches for “car breakdown recovery Anytown” or similar in your local area the Google will show them the advert. This is called an impression. Hopefully the user will click on the advert (and generate a call), but it will probably have to be served a few times before someone clicks on it. As time goes on, and your advert is repeatedly served, a report will emerge which shows that for n impressions you have received x clicks. This information is used to generate the Click Through Rate (CTR = x/n * 100).

Google Ads collates and displays all this information for you. We observe that our breakdown advert attracts a CTR of 1.3%, which isn’t bad, but we would like to improve on this. We suspect that the strapline “Garage Services Anytown” is perhaps too vague and so we want to change it to something more specific. In order to ensure that we don’t make an expensive mistake we decide to run an A/B split test (referred to as an “Experiment” in Google Ads) which will tell us if our new version is an improvement.

To do this, we create the new advert and tell Google to serve it to 20% of the people who would have been served the old ad in all future searches.

We run the test for a month and find that the new version has a CTR of 1.8%. Big improvement! We then drop the old advert and start using the new one. We then start to consider how to improve the current ad and devise new experiments, and so the process continues. Note that the old advert also seems to have a better CTR during this experiment. This is why we run experiments instead of making an informed guess. Things beyond the advert can affect how well it performs. Maybe a competitor pulled out and so your ad ranked better than it did before, or perhaps you received a boost from another activity you’re doing elsewhere. The experiment controls for these factors by comparing both adverts under the same conditions.  

So, in summary:

  1. Find a good analytics platform
  2. Learn to use it properly
  3. Consider your data from the perspective of your customer profiles
  4. Work out how to optimise the site and improve it incrementally using split testing.

That’s nearly it for this set of posts! There was quite a lot to cover, and this is only the basics. In my next post I’ll give some general thoughts that should help you to be successful in your online marketing.