Using PadiTrack to Measure Site Navigation

PadiTrack is an easy to use, free tool that enables you to extend Google Analytics with more flexible goal funnels and other navigation path tracking.

PadiTrack offers several advantages over Google Analytics for setting up goal funnels:

1. Historical data – with Google Analytics, setting up a goal funnel means the data is reported starting the day the funnel is set up, so no historical data is available. With PadiTrack, you can set the date range as far back as your Google Analytics data goes.

2. Segmentation – advanced segments can’t be applied to funnels in Google Analytics, but PadiTrack provides the opportunity to apply segments, including custom advanced segments that you have set up in your Google Analytics profile! (Late-breaking: the *new* Google Analytics now in beta allows for segmentation of goal funnels – but it’s not fully rolled out yet.)

These strengths make PadiTrack a powerful tool for monitoring progress down the goal funnel, but also make it very useful for understanding navigation from one page to another.  This especially applies if you have a dynamic site and want to understand the movement of visitors from one page ‘type’ to another. For example, you may have ‘product category‘ pages with URLs like ‘www.site.com/category.php?cat_id=111‘ and ‘product detail‘ pages that have URLs like ‘www.site.com/product.php?prod_id=123‘. If you want to know frequently visitors go from category to product pages, you have several options:

  1. Simple calculation: If the only way a visitor can get to a product detail page is via a category page, the calculation is straightforward, based on unique pageviews of each type in the Content Report. But modern sites are rarely so strictly laid out, since we usually want prospects/customers to be able to get to product detail pages via search (off-site or on-site) or other convenient means.
  2. Set up a Google Analytics Goal Funnel either with with product detail pages as end goal, or as part of larger goal. Sensible approach, but doesn’t help if you want to see last month’s results in order to make a decision on testing priorities. Also, only the first step can be set as required, so it is not uncommon to see leakages in to the funnel steps from pages that are not previous steps. And you can’t segment the funnel data – at least not until you have access to the new features currently in Beta.
  3. Use the ‘Navigation Summary’ report, which is helpful in getting a sense of flow through a given page. But it only applies to individual pages (including previous and next), is based only on clicks, and only shows a limited number of pages in the ‘next’ list. The ‘class’ of pages we want to measure may be dispersed over a large number of individual pages.
  4. Use ‘In Page Analytics’: again, only one page at a time and, while it can be a helpful visualization, tends to be an unreliable data source.
  5. PadiTrack gets around all these issues! You can use various match type options or regular expressions to identify page types by URL or Page Title (bonus!), select your desired date range, and *boom* dat’s it. For more granularity, you can filter by top referrer or top keyword or apply GA advanced segments.
Paditrack select page

Example: setting up first step in PadiTrack

Once you set up your steps – as few as 2 or as many as 5 – PadiTrack will create the funnel on the fly for your chosen date range:

PadiTrack conversion funnel

PadiTrack Funnel/Navigation Path

So we can see in this case that about 15% of those visiting a category page went on to a product detail page during their visit. Depending on our expectations/goals, this may warrant testing changes to the category page design in order to improve flow-through to product details. Or you can extend the funnel by adding more steps (up to a total of 5) to assess further progress toward the end goal. Or you may compare this to other key navigation steps on your site to prioritize testing efforts. All easy to do, with results available in minutes.

Care to share any thoughts or experience with PadiTrack or other conversion funnel/navigation tools? Please do!

Advertisements

Google Analytics: Rewriting Page URLs to include ‘WWW’

Avoid disaggregating page data in Google Analytics by applying a filter to force all ‘www’ domains to be displayed in Content reports as ‘www’ version (http://www.mysite.com/page.html as opposed to mysite.com/page.html).

Out of the box, Google Analytics displays URLs in Content reports without using the domain name (‘page1.html’ instead of ‘http://www.mysite.com/page1.html‘) . This makes sense in most cases, since it is redundant and you probably know what your domain name is. However, there are a variety of situations where you may want to display the full domain name, most commonly when your site is spread across multiple subdomains. (http://www.mysite.com, store.mysite.com, blog.mysite.com, etc.)

It is easy enough to add a filter to your profile that will cause the full domain/subdomain to show up in your Content reports. (And, of course, if you want to track across multiple domains and subdomains, you’ll need to modify your GA tracking code to accommodate this.)

Potential Issue: Page Data Split Between Two Versions

All well and good, but there is an issue that arises when the full URLs are displayed in Content reports on sites where visitors can access the site at both ‘www.mysite.com’ and ‘mysite.com‘. As a result of these two versions of  the domain, the same page may be reported on separately, in the ‘www’ and ‘non-www’ versions:

Page data split between www and non-www

In the example above, the same page is shown in two separate versions, one with 16 pageviews and one with 6 pageviews. Not cool.

For search engine optimization,  this causes canonicalization issues and is best dealt with via 301 redirect. However, this may not always be possible – at least in the near term, particularly if you don’t have access to your server settings – and you may want to have your data as accurate and relevant as possible NOW.

Solution

By applying an additional filter ahead of the filter that adds the domain to the adds the domain to the URI, we can force Google Analtyics to include the ‘www’ at the beginning of the domain in cases where it is not already present. The desired result:

Page data consolidate as 'www'

Here we can see that the data for ‘default.aspx’, previously split between two ‘pages’ in the report, is now consolidated to give us a more relevant picture of what is happening with visitors to the site: 22 pageviews of this page (16 + 6). Aaahh…that feels better!

Two Filters Used: one to add ‘www’, one to show full URL

This solution was reached by applying two straightforward filters to the GA profile:

1. A filter to recognize situations where the hostname starts with the ‘raw’ domain without ‘www’ (‘fig4.com’ in this example) and then adds ‘www’ at the beginning of the hostname in these situations. This filter will not add ‘www’ in cases where it is already present, nor will it add ‘www’ in cases where the page is on a subdomain. It does assume that you have a single domain, so it would have to be modified in the case of cross-domain traffic. It also assumes that you want to add ‘www’ to all URLs, as opposed to removing ‘www’ from all URLs. If you prefer no ‘www’, just flip the fields around.

Filter 1: Add the WWW

2. The usual filter for displaying the full URL including domain. I have included a leading ‘/’ in the Output To -> Constructor – this is not typically recommended in official documentation, but I did see it recommended somewhere by one of the big names in the field, so I figured I should try it and have seen no adverse affects.

Filter 2 - included domain

That’s all there is to it. Works for me and I hope it will work for you, but let me know if you have any feedback.

This post goes out to my friends at www.CIGNA.com (or CIGNA.com if you prefer 🙂 ).


#Fail:

One approach that I had high hopes for didn’t pan out. Not sure why:

Filter attempt with search and relpace that didn't work

Web Analytics Usage: Recent Data on Tools & Tagging

Web analytics landscape: Audience poll in a recent webinar shows Google Analytics at 75% market share, while another webinar highlights the prevalence of sites with multiple analytics tools implemented.

Web Analytics Tool Usage

Mediative (formerly Enquiro) recently presented a webinar called ‘Analytics and Reporting – How to Define Metrics…‘ as part of its B2B Expert Series.  Insightful presentation, as always. During the broadcast, viewers were offered the opportunity to complete a poll and indicate the analytics platform they are using. Here are the results, from January 25, 2011:

Mediative Webinar Web Analytics Market Share

Source: Mediative webinar 'Analytics and Reporting'; Jan. 25, 2011

Not surprisingly, Google Analytics is the dominator, being a widely known and freely available solution. The result is a 75% market share for Google Analytics with this audience. The major paid solutions fall into line pretty much as expected, with Adobe’s Omniture at 10%, Webtrends and 4%, and IBM’s Coremetrics on less that 1% of the sites polled. Red-headed step-child Yahoo! Web Analytics was not offered as an option in the poll, but presumably comprises some portion of the 4% in the ‘Other’ category, along with Unica and…others. A little disturbing to see in this day and age that 6% of the sites of the participants in the poll have no web analytics.

I don’t know the size or composition of this audience, so it’s hard to say how representative this data is. Mediative webinars usually attract an audience in the hundreds, from a fairly broad cross-section of industries and company sizes, so this is probably not too far off the mark for North American business sites in general.

One thing poll respondents weren’t offered was the chance to indicate more than one analytics solution. Would be interesting to see how that option would change results – I suspect it would skew the results even more toward Google Analytics, since it is not unusual for sites with a paid solution in place to use GA as a back-up/alternative.  The ‘Other’ category would probably get a boost as well, especially if people included things like voice of customer or audience measurement tools.

Site Tagging for Web Analytics

Coincidentally, in another webinar on the very same day, Eric Peterson (Web Analytics Demystified) referenced data on the number of sets of tracking tags that companies nowadays typically carry on their sites. This was in a presentation called ‘The Myth of the Universal Tag’, hosted by Josh Manion of Ensighten and also featuring Brandon Bunker (Sony). Lots of thought-provoking info on the value of tag management systems, but the data in question showed that almost 1/2 of the top 100 internet retailers have 3 or more sets of web analytics tags on their sites, and 85% have 2 or more sets. Since these are large, presumably sophisticated sites, they may be expected to have more tracking tags placed on their pages, but the overwhelming extent to which this is happening suggests that it is likely common on smaller sites, as well, which is supported by observation.

ObservePoint, Web Analytics Tagging, Top 100 Internet Retailers

Source: Whitepaper 'When More is Not Better: Page Tags' by Eric Peterson, sponsored by ObservePoint

This data originated from ObservePoint and was previously published in a paper called ‘When More Is Not Better: Page Tags’, written by Eric Peterson and sponsored by ObservePoint.

Conclusion

So a couple of snapshots of the current state of the web analytics landscape: lots of solutions in play, and lots of sites leveraging multiple solutions. No doubt further consolidation will proceed in the industry, although it is likely that new tools will also emerge, so it will be interesting to see how the market share relationships change over time.  Tagging of sites with multiple tools will no doubt continue and possibly expand, so it will be interesting to see how the use of tag management solutions evolves to deal with this proliferation.

Google Analytics Content Drilldown – More Useful Than It May Appear

For a lot of websites, the ‘Content Drilldown’ report does not appear to be particularly useful. In fact, compared to the ‘Top Content’ report, it often seems…completely redundant. For example, the only difference in the reports shown below is the name. (Aside from the difference in the number of pages – which we’ll get to later.) Furthermore, when you drill into an item on either report by clicking on it, you get to the same detail page. But, as I have (finally) come to realize, there is in fact more here than meets the eye, and for some sites the ‘Content Drilldown’ report may provide crucial, under-used perspective on site usage.

GA top content report

GA content drilldown report

(For many sites, there is no noticeable difference between the ‘Top Content’ and ‘Content Drilldown’ reports.)

The difference, as demonstrated in the screenshot below, is this: the Content Drilldown report (as its name admittedly implies), shows activity at the folder level, not just page level.  Depending on the structure of a given site, this can provide a very useful aggregation of data by folder that allows for easy comparison of performance between different sections of your site.

GA content drilldown report with folders

This example is for a site that is almost entirely structured in folders. Other than the home page (/index.asp) everything in the Content Drilldown report represents different sections of the site, organized by content.  ‘Index.asp’ shows up in the report because it is at the root level of the domain. So by rolling up all the pageviews in a given section, we can see in this example that while they are doing pretty well on the ‘shopping’ section of the site, in terms of exit rate, they are losing a lot of visitors in the ‘ideas’ section. Could be a good area to focus on for improvement!

And from here, when you click on one of the items in the report, you drill down into the next level of folders/pages. (Note that at each level, the number of ‘pages’ viewed refers to folders and/or pages at that level – hence the difference between the number of pages in the ‘top content’ and ‘content drilldown’ reports in the first example above.)

It’s also worth noting that if you are tracking different subdomains within your site and you have a filter in place to show full domain names in content reports, the Content Drilldown will start at the subdomain level.

With the advent of more flexible custom variables, site sections can also be tracked by applying these variables to pages. That approach has some advantages, but it involves changes to the GA tracking code. (And warrants a separate blog post!) Meantime, if you have your site architecture in order, the Content Drilldown can get you a long way right out of the box.

New Ways to See AdWords Search Query Terms

AdWords Report CenterAs part of Google’s phasing out of the AdWords Report Center, the traditional ‘Search Query Performance’ report is no longer available. But there are new and better ways to get this critical data!

Using Google AdWords reports to track performance of the keywords you are bidding on is fundamental to search marketing success, and the ease with which this can be done is a source of delight for search marketers, especially when compared to measuring traditional forms of advertising.  If you are bidding on ‘industrial supplies’, for example, it is easy to get metrics like impressions, clicks, conversions, cost per conversion and then make decisions that will lead to better performance.

But most folks also realize the additional value of being able to look behind the scenes to see the actual search queries that users are typing in and using that information to refine campaigns.  Because if you are bidding on ‘industrial supplies’ on anything other than exact match, your ad will be showing up for a lot of terms beyond strictly ‘industrial supplies’. These could be: ‘cheap industrial supplies‘, ‘industrial supplies oregon‘, ‘industrial equipment and supplies‘, ‘industrial painting supplies‘ and thousands of other variations. The value of having this deeper insight into the terms triggering your ads is obvious.

Not so very long ago in the short history of search engine marketing, this was data was not as accessible as many would’ve liked, served up only in a special report, with large chunks of information missing. (The infamous ‘other unique queries’.)

Times change (quickly) and this industry evolves (rapidly) and now more complete search query data is available from a couple of sources. At the same time, the traditional Search Query Performance report has been phased out of the AdWords Report Center – along with just about every other kind of report. So let’s look at how we can get at AdWords search query data now:

1. Within AdWords Keyword data in UI: Recent enhancements to the AdWords user interface include the ability to generate an ‘on the fly’ report on search terms.  From the ‘Keywords’ tab, you can select the ‘See search terms’ button to go to a report on search terms.  This shows all the search terms used, with indications as to which ones are currently being bid on.  The nice thing here is that you can instantly add keywords or even add negative keywords directly from this report.  And you can also download the report in CSV format.

AdWords search query terms
3. Google Analytics AdWords data: With the new AdWords reporting enhancements to Google Analytics, it is easy to get a look at actual search terms being used. Simply go to ‘Traffic Sources’>’AdWords beta’>’Keywords’ and then use the second dimension box to select ‘Matched Search Query’. Here you get a nice side-by-side listing of keywords you are bidding on along with matched terms.  The additional advantage, of course, is that you get all that juicy post-click behavioural data, such as bounce rate, goal completions, and – if you have Ecommerce tracking set up – revenue.

GA search query

So there are 2 ways that I know of to investigate AdWords search query data and use it to improve performance of your keyword advertising. Both of these methods offer some advanced flexibility and power compared to the old AdWords Search Query Performance report.

Careful with that Bounce Rate, Eugene

Bounce rate is a useful metric but can be misleading if not used in context

We all know that bounce rate conveys vital information about site performance and can provide very useful clues about where to look for improvements. Especially since Avinash nominated it as ‘sexiest web metric ever‘ back in ’07. As is often the case with web metrics, though, the numbers shouldn’t just be taken at face value. It can be important to consider metrics definitions and their use by web analytics tools in order to make well-informed decisions.

Here’s a case in point that I encountered recently in a site’s Google Analytics ‘Top Content‘ report:

Bounce Rate in Content Report

Clearly there is a problem here that requires urgent attention: high traffic pages with huge bounce rates! It’s important to keep in mind, though, that bounce rate is calculated only for visits that start with the page in question.  Being the ‘Top Content’ report, the pageviews account for all views of that page, whether the first page in a visit or the 50th.

So the report is NOT saying that of the 2,215 views of the first page in the list,  2,076 (93.75%) of them resulted in bounces.  It IS saying that of all the visits that started with this page, 97.5% of them went no further.  This, then, begs the question: “well, how many visits started with this page?

For the answer to this question, we turn to the ‘Top Landing Pages‘ report in Google Analytics (or equivalent in your tool of choice). And here we see a story that puts things in perspective:

Bounce rate from Landing Pages report

Yeah, that first page in the list has a high bounce rate…but it is based on only 16 entrances compared to over 2,200 pageviews in total! (I know we’re mixing metrics here, but the story is valid.)

Depending on the purpose and value of that page, this still may be cause for concern. But chances are there are bigger fish to fry.

There are other ways that bounce rate – and indeed most web metrics –  can be twisted by context, so it is prudent to keep in mind the nuances behind the data being presented by our trusty tools.

(For those interested, there is further discussion of the ins-and-outs of bounce rates in Google Analytics here on the Google Analytics Help Forum – particularly relevant if you are wondering how to reconcile content and navigation reports.)

Entrance Sources in Google Analytics: Don’t Go There

The ‘Entrance Sources’ report in Google Analytics offers little, if any, value but holds great potential for confusion.

We all love Google Analytics for the super-intuitive interface that makes it easy to navigate around and generate reports that quickly tell us what we need to know.  Especially helpful for those who might not be in there every day but still have business questions that need answering.

But, there are some places in Google Analytics where the terminology tossed around is maybe not so intuitive and can actually be downright confusing. Case in point is the ‘Entrance Sources‘ report. This came to light in a recent client scenario:

1. A micro-site (let’s say ‘xyzsite.com’) was created for a promotion to drive traffic to a specific page on the main site (let’s say ‘/community/…’).

2. The client was looking at content report to see how the page was doing in terms of traffic, and saw that it had 2,768 unique pageviews, indicating that there were 2,768 visits that included a view of this page.

GA Content Report

3. The client wanted to know how many of these visits came from the micro-site, so she did what seemed like a logical thing: clicked on that page in the report and then ‘Entrance Sources’ under ‘Landing Page Optimization.

GA Content Details

Here’s what she saw:

GA Entrance Sources

So now total unique pageviews have seemingly gone from 2,768 to 7,134 of which the micro-site xyzsite.com is accounting for 6,611, well above the number of pageviews shown in the previous report.  Clearly, something amiss.  And yet, if a reasonably intelligent person steps back and takes a look at it, what else could this report mean?  Other than it shows the  pageviews/unique pageviews/avg. time on site/etc for the page indicated in the Content box based on the Entrance Sources listed in the ‘Source’ column? Especially given the large bold heading that yells, “This page was viewed 8,722 times via 25 sources“.

In fact, it means something quite a bit different, although there are no clues. You just have to be in the know. 🙂 For those who are in the know, this report actually indicates the total number of pageviews/unique pageviews/etc throughout the site for visits that a) arrive from the source indicated and b) land on the page shown in the Content box – but then may continue on to other pages.  And that’s the catch: all the other pages are included in the count of pageviews and other metrics.

So for starters, the large bold heading should yell something more accurate like, “This page (and subsequent pages) were viewed 8,722 times via 25 sources”.

The current presentation may strike some as bizarre, misleading and possibly even useless. There could be some method to Google’s madness – but maybe not. The best thing I can think of was that this report could provide some insight into the site-wide impact of different source/landing page combinations that could inspire a person to try to direct more traffic from a given source to a particular page rather than others that may have relatively less flow-through.  But really, there’s easier ways to get this kind of direction.

Which brings us back to the client’s initial question: how many visits came from the micro-site and entered the main site on the /community/ page? There are a couple ways to answer this question:

1. Top Landing Pages: breakdown or pivot by Source. In this pivot table view we can easily see that the xyzsite.com drove 2,166 visits to the /community/ page.  And with only a 11% bounce rate – not bad!

GA Landing Pages

2. Traffic Sources – Referring Sites: breakdown or pivot by Landing Page. From this view we can confirm that xyzsite.com drove 2,166 visits to our /community/ page, of which 75% where new visits. Great confirmation that our micro-site is attracting new potential customers.  And from this report we can easily access Goal data to gain some insight into the quality of these new visitors.

GA Traffic Sources

So these views help us get to the larger question at work here: is the micro-site project having the intended effect and providing the desired return on investment? Although we don’t have all the data we need to answer that question (amount of investment, target return, baseline data, etc.) we can certainly see that the micro-site is having a positive impact in terms of both quantity and quality of traffic being generated – and that’s a great start.

Hopefully, this clears up the confusion that can be caused by the Entrance Sources report. My advice: don’t go there. But I’m open to ideas if anyone else has profited from this report.

Note: the foibles of this report are also discussed in Google Analytics Help here.