Ever wondered what the prominent keywords were that your visitors use to reach your site?

Google Analytics will happily show you those. With the amount of visitors per keyword of course (amongst many other metrics).

But what if you wanted to know which separate words were most used throughout all search queries? Create a tagcloud:

1. Go to the Traffic sources >> Keywords report.
2. Be sure to select a good amount of visits by stretching the date range to e.g. a year.
3. Append &limit=50000 at the end of the URL. This is some hack to get all keywords in the export and be sure to append it to the end, not before the hash (#) character or anything.
4. Click the export button at the top of the report and click CSV.
5. Open the file in an application that understands the standard CSV-format. I have to open the file in a text-editor and replace all comma’s with semicolon’s before I can open it in Excel.
6. Copy all keywords (not the numbers behind them or the general data above them).
7. Visit a tagcloud creation website and paste the data.

You can argue that the number of visitors per keyword is not taken into account but in the end that may not be what you are looking for. In case you want to know which keywords are most often used in different search phrases, this creates a pretty good image.

Below is the result of all searches that reached this website in 2009:

(click on the image to enlarge)

Thanks to Avinash Kaushik for another great post about Analytics Insights.

The name or URL strictlyPHP may not immediately reveal I have a second online passion: Web Analytics.

I am more than aware a proper site would help clarify what I do and it’s certainly on the list, but for now, a short description: besides PHP/Web development, I try to invest as much time as possible in Web Analytics. In my case summarized as: website optimization using Google Analytics.

I am extremely fascinated by the information you can derive from analysing data/traffic. I like to pin-point possible usability issues, missing information or missed opportunities to improve site conversion performance. Related to that, targeted AdWords PPC (Pay-Per-Click) campaigns are lots of fun too!

And that is what made me write this post: I just started a new Google PPC campaign for a small Bed and Breakfast in Haspengouw (Belgium) targeted to Belgian and Dutch people.

The conversion action is of course the use of the contact form and as results are coming in, it’s doing well. I did a small site rebuild in advance so not everything can be compared to the past, but the amount of new not-bouncing visitors has almost doubled and conversions (of course!) follow the same path. While traditionally March and April were their best months, July is coming close.

Conversions via phone are currently not measurable but it’s easy to ask the person that handles them if an increase is visible (audible?). You could do this properly by asking those people if they visited the website before. Logging that somewhere would allow you to combine this data without to much hassle.

Now, don’t forget to visit De Pastorie yourself! Even if it is only for a drink on their terrace.

Or even better: if you like to receive an actionable optimization report for your site or launch a PPC campaign together to increase traffic: don’t hesitate and send me a message.

Finding out why your visitors do not buy anything or don’t take any other action you want them to take does not need to be that hard.

This post will be an attempt to show how easy it is to get an insight into how many potential “converters” leave your site before they buy anything (or perform any other goal) and why they do this.

First of all: make sure you set goals in Google Analytics. In this post, I assume you only have one goal, but be sure to mark everything that is of interest to you (as the site owner) as a goal. If you didn’t do this upfront, you’ll have to wait for some time so you have enough data to work with (let’s say at least 150 – 200 conversions).

Secondly: have a look at the people that convert. Because it is a predefined segment, select it like this:

• Open up the advanced segments selection on the top right of your Google Analytics overview.
• In the “Default Segments” list, select “Visits with Conversions” and click “Apply”.

As you would expect, you now see information for visitors that completed your goal. One important note: if you have multiple goals, be sure to monitor them separately.

Have a look at the “Visitors” – “Overview” report, you will see some useful averages for visitors that converted. E.g.: you’ll notice people that convert visit at average X number of pages or stay X minutes on your site. Write them down or something (yes, with a good old pencil).

With this information: advanced segmentation to the rescue (again)! It is as easy as this:

• Open up the advanced segments selection again.
• Click “Create a new advanced segment”.
• Add one or more of the averages you have in front of you to the new segment. E.g. Page Depth greater than or equal to 15. As long as you have enough results, segmenting more precise can have advantages (as long as you don’t generalize the outcome).
• Save the segment and apply it to the reports.

You now basically look over the shoulder of people that leave your site before they convert. You can for instance notice:

• where they left your site (“Content” – “Top Exit Pages”).
• whether many of them return more than once (“Visitors” – “New vs. Returning”).
• if specific countries/languages pop out (“Visitors” – “Map Overlay”/”Languages”).
• if any OS or browser pops out (“Visitors” – “Browser Capabilities”).
• whether specific content is notably more popular (“Content” – “Top Content”).

Comparing this to the people that convert can be useful. Did not-converters

• reach you from a different source (“Traffic Sources” – “All Traffic Sources”)
• land on another page (“Content” – “Top Landing Pages”)

than converters?

Remember: there really is no end to this – segment and analyze as much as you can. One segment at a time. People in one segment could be leaving for different reasons than people in another.

Don’t hesitate to contact me if you get stuck somewhere!

Your instinct may tell you that the higher you rank on a search engine, the better the results will be.

But that is not always the case. If you’re not ranked number one for a broad keyword (e.g. for “widescreen monitor” – as opposed to “review samsung widescreen monitor”), this can give you higher quality visits than the magic top spot. When you look at bounce rates for instance you could get better results for lower (but of course still high) rankings. And it also applies to conversions.

This is explained by the broad nature of the keyword. People that are looking for something specific will (also) look further down the results and will thus have a higher affinity with your content as opposed to all visits the number one ranking generates.

It is certainly useful in light of PPC (e.g. Google Adwords) bids.
When you look at these bounce rates based on the add positions of a broad keyword, you get the picture:

It’s important to note that the Top1, Side 1, Side 2 and Side 3 all generated a comparable amount of clicks during a 2 month period. As you can see, the Top 1 position (in theory the best place) causes significantly more bounces.

So if you don’t bid for that number one spot and end up in the middle you may just be happier with the results.

The opposite is true for very specific (and/or localized) keywords.

So to conclude this post:

• Bid high for specific keywords (but highly specific keywords are usually inexpensive).
• If you bid for broad keywords, bid average.

Of course, every statement has it’s exceptions: use your analytics tool to find out what suits you best.

It was a lot of fun to track Google Analytics last year. Since I started writing this blog in mid 2008, traffic to my one-page site has been increasing.

Have a look at the visitors graph:

Can you tell when this blog was launched?

It’s interesting to see the different implications each post had. For instance, one of the first about uppercasing the first letter of a MySQL result value, has brought 313 visitors to this site (through almost as many search phrases) while the one about a Euro conversion rate feed brought none.

As a side note: close to 63% of these visitors used FireFox.

Besides that, bounce rate is at an extremely high 75%. While this is normal since the only thing this site has to offer is that once specific piece of information someone was looking for, it should be fairly easy to have visitors browse through more than one page.

If time was on my side and I would be short on work, that would probably be a fun challenge!

It’s great to use these experiences in other projects though.

You can use Google Analytics in 2 main ways: as a bloated Webalizer or as a list of possible improvements for the site in question (or both of course).

If you just start using it the second way, the number one improvement suggestion that has always been in front of your nose is the list of Top Landing Pages (which you’ll find under “Content” in the menu) and their bounce rate. Keep it sorted by “Entrances” and the only thing you then have to do is pick the page with the highest bounce rate in the top 10. Now fix that page (as clearly something is wrong).

Make sure people that hit that page find ways to lengthen their visit to other parts of the site. Clearly visitors to that page aren’t noticing your navigation or other links.

You can use this knowledge to direct them to the content that you want them to see or to the content that they are looking for. To get a hint of the latter, click on the page in question in your report and click on the “Entrance Keywords” under “Landing Page Optimization”. The list of search keywords used to reach that page, for instance, could set you off in the right direction.

Of course you can continue with other high bounce rate-pages but it may be a good idea to monitor if your changes had any effect first.

The more data available, the more relevant the Top Landing Pages report is. But even if your site only attracts a few visitors per day, just stretch the time span your report covers to have a larger data sample.

Hopefully this helps to improve retention bit by bit.