Imagine the scene. You’ve just bought an existing and operational website that is currently receiving traffic. However, the website is still an entity unknown to you so you need to start recording traffic statistics that you can analyse at a later date. Only when you’ve analysed those traffic numbers can you get an idea of how the site is performing, how visitors behave on the site and how best to change the site. This article illustrates the process I use when analysing traffic statistics of a site I’ve just acquired. It’s not meant to be a blueprint that you follow religiously (though there are far worse things you could do ). Rather, it would be better to try out the ideas presented here, keep what works, throw away what doesn’t and then add your own ideas to… The Method.
Website Traffic Analysis
My tool of choice for recording website traffic is Google Analytics. There may be other analysis packages out there that are better, but my opinion is that GA is the most comprehensive free one available. The first problem we face when taking possession of a new site is that our Google Analytics tracking code is not currently on any of the site’s pages. We need to add the code to every page.
Adding Google Analytics To A Content Management System
If the site currently runs on some content management system, adding your tracking code is usually a breeze. Some WordPress themes, for example, provide an input box whose sole purpose is to accept your tracking code. Adding it there will automatically put that code on every page. If the theme does not provide such an input box, then you will have to edit one of the templates (usually footer.php) to add it there. Again, once those changes are saved the code appears on all pages of the site. Here is a handy article on adding Google Analytics to WordPress.
Another popular CMS is Joomla. Just searching the Joomla extensions directory yields a large selection of modules you can use to add your tracking code. Using a module saves you the hassle of finding the write template whose code you’d have to hack amend.
Adding Google Analytics To A HTML Site
I’ve just bought an existing website and am in the collecting traffic data stage. Unfortunately for me, the site was written in straight html. This has many advantages, but applying Google Analytics tracking code to every page is not one of them! In this case I had to systematically work down the list of HTML files, and add the code to each one. Here is my streamlined process:
- Open the next HTML file using notepad.
- Jump to the end of the file.
- Position the cursor before the </body> tag
- Paste in my Google Analytics code.
- Ctrl-S to save the file.
- Close the file.
- Go to step 1.
My new site has around 300 pages, and the total time spent on this mind numbing task was around 30 minutes. That’s 10 pages updated / min. You can use that as a rough guide if you need to estimate how long it will take you to do your site. I did this as soon as I was in control of the hosting account for the domain as I wanted to start recording statistics ASAP. The sooner the analysis stage is complete, the sooner we can actually start working on the site.
Monitoring The Site’s Traffic Statistics
I took possession of my site late on 27/07/2009 and frantically added GA code to all 300 pages of the site. That means I have 8 full days of traffic statistics to look at. It makes interesting reading. Here’s what I’ve gleaned so far.
General Website Traffic Analysis
I’m encouraged by the fact that the site received 229 visitors on the first day. So far, in the 8 days I’ve been tracking, the site has been visited by 1,532 unique visitors. That’s an impressive amount straight out of the box, without me even changing a thing on the site. The traffic distribution over sources looks like this:
The search engine share is a good amount for me, because I just love doing SEO. I have great plans for SEOing this site. Once the analysis stage is complete, it’ll be keyword research all the way and already I have ideas for new content based on what I know people are searching for when they reach my site.
The referral traffic amounts to around 1,000 unique visitors / month, which again is a reasonable amount. If the search engines drop the site (I can’t imagine this happening) I know I will still get at least 1,000 visitors / month from other sources.
The direct traffic is puzzling. I know the site hasn’t changed in 3 years, and it’s obvious that the site is pretty static, so why people have bookmarked the site or are returning via type-ins, is anybody’s guess.
Traffic Trends At Weekends
We’ve seen traffic slumps at the weekend before, and this site experiences them too.
Affiliate Sales Ahoy
The most popular page after the home page (as far as pageviews are concerned) is a page called xxx.html where xxx is the name of a type of product, like “suitcases”. You know what this means, don’t you? Yes, that’s right. That page is perfect for selling items that fall into that category of product. All I need to do is find some affiliate products in that category. I’ve checked Commission Junction and there are plenty of products I can advertise. This will be on the “things to do” list that is produced by the planning stage.
It’s more good news. The top search phrases that bring traffic to the site all support the idea that affiliate products can be sold here. Let’s pretend that the most popular page mentioned above is called suitcases.html. The top search terms are phrases like:
- travel suitcase
- samsonite suitcase
- vintage suitcase
If people are searching for a product category like “travel suitcase”, there is a good chance they want to buy one. And if they want to buy one, I certainly want to sell them one!
Average Time On Page
The average time spent on a particular page is a useful thing to know. It tells us what pages keep our audiences riveted and which pages make them run a mile. I think there is the danger of reading too much into time on page if the number of pageviews is small, though. For example, that one person visited a page and spent half an hour there doesn’t tell us much because they may just have passed out at the keyboard from boredom and clicked off it when they came to. If, however, 100,000 people visited the page and the average time they spent there was half an hour, then you know you have some mesmerising page there!
For my purposes, I’m going to ignore all pages that have less than 50 pageviews. I set the pageviews threshold so low because I only have statistics for 8 days. After a month, we’ll be able to do a more accurate study.
But… there is one page that has received 5 pageviews and that has an average time on page of 28 minutes and 57 seconds. They can’t all have passed out, surely? Anyway, I’ll look at that when I have more data.
To see the average time spent on page in Google Analytics, click Content > Top Content. To sequence the list by descending times, click on the column heading for Avg. Time On Page. Now pages whose visitors stay on the page for the longest time appear at the top. Just scanning the list of pages with PVs > 50 I can see some impressively long times:
This implies that those pages are getting read in their entirety. That means they are of high enough quality to engage the reader. This is good.
What about pages that don’t engage? Let’s click on the column heading for Avg. Time On Page to sequence the list by ascending times. Again, we’re going to ignore pages that haven’t been viewed enough times for these values to be meaningful. The first page I find with pageviews greater than 50 has an average time on page of 38 seconds. Is this good or bad? Well, the page is simply presents a list of linked articles for the reader to navigate to. That the average person spends 38 seconds reading down the list to find an article they want to read is not unreasonable. For other pages, 38 seconds might be abominably low. For a list of articles, I think it might be acceptable. Next page!
The next page has an average time on page of over a minute. Now we are in safe territory. My gut feel is that any page that keeps the visitor reading for over a minute is doing something right. As all pages further down the list are going to have longer times, let’s stop there. Conclusion, there isn’t much wrong with engagement for the pages we looked at. Of course, we only have 8 days of data, and that isn’t nearly enough to assess accurately. It gives me a warm glow, though.
Website Traffic Analysis – Conclusion
This was only a preliminary investigation to get a rough feel for how the site was performing, and to see whether there was anything seriously wrong we needed to fix. So far it’s looking good, but let’s not count our chickens. When we have a months worth of data, we’ll be in a better position to draw meaningful conclusions.
One more thing: in addition to analysing website traffic, you’ll need to analyse inbound links. More precisely, you’ll need to find and fix your 404 HTTP errors.