Archive for the 'Analytics' Category

On-Site Search Box Text Confuses Users

Do you have a search box on your website that contains a phrase like “Enter Search” or similar? Are you using Google Analytics to track Site Search?

I’ve noticed this on several sites for a while, so thought I would post about it. In most cases where there is text already in the search box, instructing the user what to do, that text tops the list of keywords searched for on the site. Take this example:


As you can see, the search box in the top left of the page has the text “Keyword/Code Search…” inserted by default, and it disappears when you click in the box. Can you guess what the most popular keyword used to search on the site is?


Yep, “Keyword/Code Search…” by a long way! What does that tell us about this use of text in the search box on a website?

My opinion is that it isn’t sufficiently clear to the user what they are supposed to do. So they click the arrow next to the box, expecting it to take them to a full search page, but instead, it gives them the search results from the site for “Keyword/Code Search…” In this site’s case, that gives you a full list of all the products in the catalogue, but not a search page – you get the same box and text again.

My take on this is that designers need to be more instructional about what to do with/how to use the search box on a web page. Just putting “Keyword” in the box is not telling people how to use the function, but “Type what you’re looking for in this box” might just work.

Google Analytics Campaign Tracking vs. ASPX

Google-Analytics-LogoAs a result of the MyDeco experience (see earlier post), we found that the site in question wasn’t recording campaign tracking (although obviously we can see referring websites). In case you’re not aware of Campaign Tracking, there’s a guide here.

MyDeco are keen for retailers to use campaign tracking to ensure more accurate results with a better quality of data. This is usually done by appending “?partner=mydeco” to the end of any link to the retailer, so that it shows up in their logfiles. If you’re using Google Analytics, this won’t do, as Google wants campaign tracking to be done in its own UTM format (see the guide linked above).

So, we tried this with the site in question, which is hosted on a Microsoft IIS server and written in ASP .NET (.aspx). This had the effect of causing an error – the pages really didn’t like having a query string put on the end of the URL, which is what putting a “?” means. So, we needed a way to get Analytics to accept an alternative character to replace the “?” and thereby stop the website from throwing errors.

The solution, after some searching, was to use the anchor signifier “#” instead of “?”, which the website is happy to accept. However, you can’t just make campaign URLs with “#” instead of “?”, because by default Analytics won’t know what it means. You need to add this line of code to your Analytics tracking code (the code inserted into every page of your website when you set up Analytics):


This line of code should be inserted as follows: Find the Google Analytics code in your webpage and add it like this:

var pageTracker._gat._getTracker(’UA-xxxxxx-x’);

I found this tip courtesy of Digital Notions, so hat-tip to them. :)

Google Analytics Adds Event Tracking Segment

Google have added an Advanced Segment item for Event Tracking, so you can compare the behaviour of visits with an event to any other segment you choose. You can read about it and see a video here.

This reminds me to remind you – if you have any events on your website, whether it’s a document for download, a video to watch or anything else, make sure you add the event tracking code to it so you can see what’s happening in Analytics. Have a read of the two links in the blog post I’ve linked to above: Implementing Event Tracking and Event Tracking Best Practices, which will tell you how to do it.

MyDeco Click Reporting – How Retailers Can Get Ripped Off By CPC

I had a click report from My Deco sent to me by a client on Friday, asking for my comments. MyDeco is a comparison shopping engine in the home interiors vertical market; they work on a combination of affiliate comission (CPA) and pay per click (CPC) to generate their revenue, so retailers pay for clicks through to their website and a percentage commission if a sale is made.

This particular client has been very disappointed by MyDeco’s performance, with no sales being generated and a very small number of visitor referrals. So it was something of a surprise to see the click report claiming hundreds of visits with a resultant bill of three figures, when Google Analytics reported visits barely into double figures from MyDeco in the same period.

The accompanying email from MyDeco claimed “We have excluded Google, Yahoo and other known bots in line with industry standards, as well as repeat/multiple clicks from the same IP (double clicks)… We would also like to point out that some CPC retailers have experienced differences between their Google analytics reports and the volume of clicks in the attached reports.  This is due to the inherent limitations of Google analytics software, to identify trends rather than 100% accurate datasets.” Well, I think Google might have something to say about that, but it becomes pretty clear why they’re making excuses when you open the report.

Inherent “limitations” of Analytics accetped, do you really think that it is so limited that it would fail to report 97% of all visitors from MyDeco!? Seriously, you think Analytics is likely to be that wrong? I can’t see it.

But then we come to the real crux of the matter. The report provided, by default, opens to show clicks listed per day for the period. If you expand each day, it tells you the time of each click and the product that was clicked on. I found several instances where there were five clicks from the same product within a minute of each other. The charge was 36p per click every time. What do you think the chances of five different people clicking on exactly the same product within the space of a minute are, taking into account that we are talking about a three month period with less than 300 clicks in total? They must be the same person each time. So what happened to excluding “repeat/multiple clicks from the same IP”?

This kind of rubbish is disgraceful and gives CPC a bad name. How many retailers would just open the report, look at the number of clicks and pay the invoice? If you’re a MyDeco retailer, I urge you to examine your click reports in detail and see if it tallies with your Analytics and log files. In this case, the discrepancy is huge and the cause is quite clear – complete failure of the MyDeco click cleaning software to do its job, very much in MyDeco’s favour.

Canonical tagging and Google Analytics

Apologies for the paucity of posts recently, I’ve been very busy delivering training and consultancy for the eBusiness Programme. Presenting my new Google Analytics & Conversion workshop (four times in a week!), Analytics is very much on my mind and I spotted this on Erik Vold’s blog, relating to the previous post here about the Canonical tag and preventing Analytics from double-counting pages:

Erik has created some Javascript and amended ga.js to stop Analytics counting canonical pages in your stats. Very handy! :)