CxFocus

Tag: google analytics reports

  • Get an Instant Checkout Health-check With This One-stop Report

    Screenshot showing Google Analytics Dashboard with Checkout Abandon RatesSee if anything is wrong with your checkout at a glance with this Google Analytics Custom Dashboard.

    Important Update: September 2012 – Combine all data in one chart

    If you have an ecommerce web site you know that the performance of the checkout is critical. If anything goes wrong with the checkout it will be costing you money.

    That’s why first thing every morning, before I even look at the sales figures, I look at my clients ‘Checkout’ custom dashboards.

    It gives me an instant health check. I can clearly see if something has gone wrong with any of the stages in the checkout, such as the payment gateway or the address lookup system. Or if something is wrong with a promotional offer codes and people who use it are unable to buy…

    All I have to do is to take a quick look at this one-stop dashboard to see if any of the ‘sparkline’ charts has an unusual looking spike. I can instantly see if something is out of line, and exactly which part of the checkout has a problem. I really like reports like these. They’re what I call ‘read and react‘ reports.

    Using Google Analytics to Monitor Checkout Abandon Rates at Each Stage (mp3)

    Build a report like this and you’ll find it’s very powerful. You may already have read enough to know what to do, but I’ve explained how to do it below. Follow these instructions and you will be able to spot if there’s anything odd happening with your checkout by looking at one clear report.

    Which means that… You can jump in and fix the problem before too much money has been lost.

    How To Make Your Read and React Checkout Dashboard

    The secret of this technique is to set up a sequence of Google Analytics funnels, one for each stage, and then build a dashboard showing the abandon rate for each of those funnels.

    Screeenshot: how to set up Google Analytics goals for horizontal funnel from LunametricsThe idea of using a series of separate funnels has been around since 2010.

    The technique is called building a ‘horizontal funnel’ and is mostly used to allow you to use segments with funnel abandon data. This idea was recommended in a Lunametrics article by John Henson: http://www.lunametrics.com/blog/2010/06/04/segment-goal-funnel-google-analytics/ [link opens in a new tab]

    The original concept involved using a Google Analytics Custom Report to present the information in a way which made all the important numbers easy to see, but it did not include any visualisation of the abandon rate from the different stages.

    But now we can use GA Custom Dashboards to show all the individual abandon rates, trended over time, on a single screen.

    Being able to spot the changes in the trend compared to normal performance adds the final touch which transforms ‘detailed analysis’ data into a ‘read and react’ operational report. You also get a bonus of being reminded of the relative abandon rates from the different stages so you can see where the friction points are.

    Each time you look at this dashboard you can see what’s going on in a split second. You can instantly tell if you need to fix anything.

    At peak periods it’s a good idea to check back later and set the date range to include ‘today’ so you can keep an eye on things.

    Once you’ve tried this technique you can apply it in lots of ways. In the example here, I’ve extended the concept to show some extra very useful data which is relevant to checkout problems.

    Screenshot: Checkout GA Custom Dashboard

    In the first column I’ve got the ‘alarm bell’ step by step abandon rates. These are all ‘metric’ widgets which include the sparklines trend charts which are the key to this technique. I’ve arranged the widgets in order of the likelihood of trouble, not the real life order of the stages. You can just drag and drop them to suit your preference.

    The middle column contains a ‘table’ widget which might also provide evidence of something odd, or at least interesting, going on with the checkout. It shows the overall checkout abandon rate by source/medium of the visit. I find it useful to be constantly aware of how some channels bring visitors who are far more motivated to make it all the way through the checkout than others.

    In the third column I’ve got some ‘line chart’ widgets because they allow me to compare two metrics. In this example they are comparing the abandon rate from just the cart page with the abandon rate for the rest of the checkout. Or the abandon rate from just the cart page with the abandon rate of the checkout as a whole.

    You could extend the idea behind the middle ‘table’ widget further. For example, on many sites it would be useful to have another widget showing the abandon rates from just the cart page, broken down by source. If your cart page contains an offer code box, for example, a sudden spike in the cart page abandon rate from one source is a strong indication that there’s probably a ‘bad’ code being promoted there. In that situation you can either create an extra version of the code which matches the one which people are trying to use, or try to promote the correct code via the same source.

    In the video below you’ll see me building another variation on this layout. Once again the first column contains the ‘alarm bell’ sparklines, but the table widgets in the middle can be used to gain a wide range of operational insights which you can use to modify or correct promotional campaigns straight away.

    You can include up to 12 widgets on each dashboard. What you choose to show in the tables should be decided on the basis of what you’re actually putting your resources into now. This is a living dashboard: you should change it to align with what matters now and what you can still change.

    Can you see the theme there? This dashboard is not just a read and react tool for spotting technical errors. It’s also a very useful tool for getting an understanding of the motivation of the visitors from different channels. The biggest variations in checkout performance, in my experience, have very little to do with the technicalities of the site and everything to do with the strength of people’s desire to buy. Persuasion and motivation can have more influence on abandon rates than the position or wording of the ‘checkout now’ button.

    The original Lunametrics post gives you excellent instructions on how to set up the sequence of funnels.

    And this video shows how to use those funnels in a custom dashboard like the one above:

    The example shows the abandon rates from a Magento one page checkout, as reported using the GoogleAnalyticsPlus Magento plugin from Fooman.

    Updates on Checkout Abandon Rate Dashboards in Google Analytics

    [September 2012] Major Update: Combine all your checkout abandon rates into one chart!

    One of the problems with dashboard widgets is that you can’t control the y axis scale. This means that there’s no easy way of making a visual comparison if the abandon rates from different stages are widely different — for example 40% from the cart page, but 5% from delivery options.

    An interesting way round this is to use the abandon rate data from your ‘horizontal funnels’ outside of GA itself, using the GA API.

    Here’s an example of a chart which:

    1) Uses the same y axis scale for all the abandon rates
    2) Combines them all in one chart (even better)

    Example of google checkout abandon rates by stage combined in one chart

    For full instructions on how to do this see this Google Analytics blog post:

    http://analytics.blogspot.co.uk/2012/08/automate-google-analytics-reporting.html

    and here’s the video version

    The metrics you need will be something like

    ga:goal1AbandonRate,ga:goal2AbandonRate,ga:goal3AbandonRate,ga:goal4AbandonRate

    where the goal number matches the actual goals you have configured on your site.

    [January 2012] Someone has just pointed out that it can be extremely useful to look at checkout abandon rates by browser for each stage. At the moment there’s no easy way to do this in the dashboard itself. You can’t use Advanced Segments in dashboards. And even if you could, you really need to be able to segment by browser version, not browser, as this screenshot shows:

    screenshot: google analytics cart abandon rate by browser

    Having to configure each segment first and only being able to see four browser versions at a time would make using segments tedious, even if it was possible.

    So the way I do this is to configure the widget to link through to a Google Analytics custom report containing a tab which allows me to see the abandon rate broken down by browser and then by browser version. I also have a tab showing a source/medium breakdown, which is the one I use more often.

    You can import an example of such a report into your own GA profiles by logging into GA and then clicking this link:

    https://www.google.com/analytics/web/permalink?type=custom_report&uid=pmQVC9ZyRj6SYJ565Eu9kA

    This example shows the rates for two goals (14 and 15 in this case) so that it can be used as the link from two widgets. That suits the way I work. This is intended as an example to get you started: you must edit the report to show the goals which you have configured in the relevant profiles.

    Resources for Horizontal Funnels and Custom Dashboards

  • How Multi-channel Funnels Give You More Insight Than Ever Before

    Screenshot: SEO Keyword Grouping GA Multi Channel FunnelsUpdated: March 2012

    Multi-channel Funnel Reports in Google Analytics bring us a whole new depth of understanding to how people come to web sites over the course of several visits.

    Now we can see how some channels are more valuable than we thought up to now. They’re helping make sales which are being credited to something else.

    And there’s a bonus: we can also start working with our SEO keywords and other sources of traffic with a level of control which was previously only available for paid advertising and other campaigns.

    In real time. Retrospectively. With no pre-configuration effort (so not like campaigns, then).

    The most stunning of these reports, in my opinion, is the Top Conversion Paths report. For the first time ever it shows the history of a series of visits.

    What’s more, it groups them by channel so that you can see the most common sequence of routes to the site. For example: social, followed by search, followed by email.

    What’s even more, you can create your own custom groupings which means that you can start working with your organic search keywords in groups, like ‘head terms’ and ‘long tail’ or ‘brand’ and ‘generic’, just as you would with paid search. Or you could group your social sites according to different types, or separate different classes of referring sites, and so on.

    The standard groupings include things like:

    • ‘Paid Search’ — PPC advertising
    • ‘Display’ — banners etc
    • ‘Other advertising’ – other paid advertising
    • ‘Organic search’ — unpaid search visits
    • ‘Social’ — Facebook, Twitter and a long list of other social sites (take a look)
    • ‘Referral’ — referred visits
    • ‘Email’ – visits where the medium has been tagged as ’email’ (exact)

    This allows for a very powerful summarised view of the pattern of visits like this:

    Screenshot: Google Analytics Top Conversion Paths report

    But this default view only scratches the surface of what you can do with these reports.

    The real power lies in the fact that you can create your own custom groups. The interface for doing this is great and you can (should!) use the basic groupings as a template to get you started.

    The first thing I would do is to split those ‘Paid Advertising’ and ‘Organic Search’ groups down so that you can separate searches for your brand keywords from generic searches for more general terms. Those are two hugely different types of search.

    In order to get new customers you need to be able to clearly understand the steps which someone takes from doing a generic search for something you supply through to the point where they know about you and your brand.

    At this point I should pause and thank Matt Trimmer at iVantage who pointed out that Custom Channel Groupings has the facility for doing exactly what we need.

    Here’s an example of what such a report looks like. This is much more revealing than the default version.

    Screenshot: Separate Brand and Generic Search Groups in Top Conversion Paths

    Suddenly it’s obvious that most of the search visits are from people who already know about the brand. Generic search doesn’t appear until much lower down the list.

    The Custom Channel Groupings interface makes it very easy to copy and then edit the basic group configuration.

    The slowest part of the process is working out what your brand keyword variations are. There are full instructions in an earlier article on How to Start Working with Brand Keywords in Google Analytics

    Once you have your list of keywords it’s all point and click from then on. This 6 minute video shows you what to do:

    Separating brand from generic search is the first thing I would do with these reports. But the potential to go further is huge. This approach means that you can start working with groups of unpaid keywords in the same way as you do with paid keywords. For example, you can start breaking those generic search terms down into ‘head’ and ‘long tail’ groups.

    Things to Watch Out For with Multi-channel Funnels

    • You have to have goals or ecommerce configured for these reports to work
    • The default view takes all goals into consideration. I think it’s better to be selective about which goals you are reporting on, particularly if you have many micro-conversion goals set up. For most ecommerce sites, I would tend to consider the ecommerce transactions on their own and then look at key micro-conversions separately. It stands to reason that you might get a series of micro-conversion visits before a final transaction visit, so you need to keep them separate.
    • These reports only work with goals configured and they only show visits with goals. But goals could include non-bounce visits if you so wished.
    • Experiment with the drop down menu which allows you to choose whether to see all the visits, even if they only have one step, or more than 2, and so on. The ‘more than’ options are the most interesting when you are looking for patterns. But don’t treat those as the default or you risk missing the fact that only some visits have any history at all.

    That last point is an absolutely critical one. The history of multiple visits is based on cookies. People now use multiple devices to visit sites and some systems also delete cookies, which means that many real-life returning visitors are counted as new visits by Google Analytics and all cookie-based systems. So what we have here is a sample. These reports provide much more insight than we have had up to this point. We can now see the history of a series of visits. But it’s not the complete picture: this is great for assessing relative numbers  and trends, but the actual numbers should only be treated as indications.

    Resources Relating to Multi-channel Funnels

    Here’s a beginner’s introduction video from Google:

    There’s a more serious one here: http://youtu.be/rZ2RbGsuy3U [Opens in new tab]

    [Other good sources are appearing all over the place at the moment, so I will continue to update this list.]

    Updates

    • [March 2012] Google have introduced new standard channel groupings in Google Analytics. I think it’s wise to take a look at the actual definitions (start to create a new custom set based on the standard definitions, but don’t save). You may find that some of them won’t work for you. For example I work with clients who use several variations of medium for their email campaigns, such as ’emails’, so the standard exact match for ’email’ is no good. A ‘contains’ match would be better for them. Here’s the official guide to the standard Google Analytics Channel Group definitions.
    • [November 2011] The introduction of secure search by Google means that a growing proportion of organic search visits do not have any keyword information. These will show as ‘Not Provided’ in your keyword reports. Since you do not know whether these searches were for ‘Brand’ or ‘Generic / Non-Brand’ terms, it now makes sense to add a third organic search Channel Group for ‘Not Provided’.

  • Download Custom Report to Compare Days of the Week in Google Analytics

    Screenshot: Google Analytics Daily Report by Day of WeekHere’s a great custom report for you to try. There’s a link below to import the report straight into GA, but first let me explain why I think this is so exciting.

    Retailers and other web sites with strong weekly visitor patterns need to be able to compare Mondays with Mondays and Saturdays with Saturdays.

    Comparing the same day of each week over a long period is also very useful for any site planning emails or any other form of on-line campaigns.

    So I was really excited when a comment on Linked In from Mike Sullivan of Nextanalytics provided me with the key to doing this.

    Although I’m a huge fan of Custom Reports, for some reason I had not noticed that ‘Day of Week‘ is available as one of the dimensions.

    In v1 of Google Analytics you used to be able to select all the ‘Mondays’ in a month by clicking on the top of the column in the calendar. But you could only do one month at a time and even that ability disappeared in later versions. It became necessary to do this kind of reporting outside of GA using Excel or a dashboard client.

    Now you can do it all in a custom report, so I’ve made an example ‘weekday‘ report which you can download using a link at the end of this article.

    Using Custom Reports is so much better and allows you to experiment with using secondary dimensions at the top level of the report. ‘Dimensions’ is the GA word for ‘what the data is that you’re looking at the numbers for’ — the information in the first column of the table. Dates or Days in this case. So you could add the source/medium for each day next to it in a second column, for example.

    But first take a look at this standard version:

    Screenshot: Google Analytics Custom Report by Day Tips

    But my favourite view at the moment is to use ‘Day of Week’ as the top level dimension, and then drill down to ‘Date’. By using Custom Reports you can add a new tab to view the data that way round:

    Screenshot: Google Analytics Custom Report by Day of Week

    I’ve added the names of the days to the screenshot so you can see what’s happening. The report just shows the number of the day: 0 for Sunday, the first day of the week and 6 for the last day, Saturday.

    This version of the report means you can see the average figures for all 7 days of the week compared to each other over an extended period. You can then drill down to see how each of the ‘Tuesdays’ compare with each other. That’s the insight which interests me most. The chart also updates to match, so you can choose whether to leave it with a single spike for one day each week, or change to ‘Graph by week’ view to see a continuous line.

    Screenshot: Google Analytics Comparison ViewThese reports work really well in ‘Comparison View’ and ‘Performance View’, so be sure to experiment with those. But I’ve discovered that you can’t sort by the day of the week when you’re using that view, so you’ll need to be careful. There have been changes to the sorting functions during the last few weeks (weighted sort appeared and then disappeared), so this situation may well change.

    How to Make a Day of the Week Custom Report in Google Analytics

    The simple clue that ‘Day of Week’ is available as a dimension will have been enough for some people.

    If you’d prefer to see how it’s done, I’ve made a 5 minute video showing each step.

    Or if you’d prefer to save some time, use the link below the video to import a copy of the report and then edit it to match your business needs.

    You’ll see in the video that I ran up against a problem when I went for overkill and tried to include all 20 goals and a load of extra ecommerce metrics. I was doing that because I thought it would be helpful to make a report which would work for everyone no matter which goals they use.

    The final report does not include the final few goals, but you can download it into your own GA and see for yourself. Have a play with it and edit it to suit.

    How to Import the Days of Week Google Ananlytics Custom Report

    1. Log into GA
    2. Click this link: [opens in new tab] to import the report
    3. Don’t forget to scroll down to choose your profiles and then click save!

    Things to Watch Out For

    When you’re doing this kind of comparison it makes sense to use a date range which includes full weeks

    If you’re using the ‘View by Day of Week’ approach it is vital to choose full weeks

    The default sort order in GA is based on the first column of metrics in the data. I always repeat ‘visits’ in each metric group because it’s important to always be aware of the size/significance of the group of visits you’re looking at. But in these reports you will almost certainly want to sort by the dimension, not the metrics (the date or the day of the week). Don’t forget to click at the top of the dimensions in the table to change the sort order.

    In the screenshots I’ve added the names of the days of the week. GA just gives you the numbers! 0 is your first day of the week, 6 the last.

    Resources Relating to Custom Reports and Days of Week

    Before you go, please give one of these sharing buttons a click:

  • Learn the Secret Power of Google Analytics Custom Alerts

    Abstract photograph of light shining on wallHere’s a quick way to get Google Analytics to shine a light on problems on your site and help you sort them out. You can also use this tip to spot promising sources of new customers and other good news like that.

    In the practical example you’ll learn how easy it is to set GA to alert you when new visitors are landing on a 404 (“Page Not Found”) error page and give you a one-click route to the source of the problem so you can head over there and fix it.

    That’s powerful stuff. If someone comes to your site and all they get is an error they may not give you a second chance.

    To adapt a well-known analytics phrase: “They came, you slapped them in the face, they left”. So a quick fix for these problems is very valuable indeed. Using this approach, GA gives you a great way of sorting things out fast.

    Once you’ve tried this very practical example, you’ll probably find that you want to apply the same method to other key events on your site, including good ones. It’s and easy but powerful technique — let Google Analytics do some of the tricky work for you!

    The secret is to combine micro-conversion goals with GA Custom Alerts. The hidden super-power of this combination is that it uses the cleverness of Google to give you clues as to ‘why’ something unusual has happened, not just to flag up the fact that it ‘has’ happened.

    First I’d like to thank @mattycurry for really driving this point home in a @boagworld podcast I heard a while back.

    Example: How to Spot and Fix Broken Links from External Web Sites

    Use this simple technique to make a “Read and React” report for spotting accidentally broken links from external web sites. In the age of user-generated-content there’s a lot of this about. I often see it with links being posted in forums on voucher or coupon code sites.

    Using Custom Alerts I can not only see that there’s been a sudden increase in 404 pages:

    Screenshot: Google Analytics Custom Alert Lightbox Overlay

    But GA also provides me with an instant analysis of the source of these visits:

    Screenshot: Google Analytics Custom Alert Major Contributors

    And, since these visitors are being referred from another site, there’s also a handy button to take me to the referring page with the dodgy link. So I can click straight over there and try to sort things out.

    In short: GA tells me something’s wrong, suggests what the cause may be and even helps me on my way to providing a fix. Nice.

    How to Set Up a Custom Alert Based on a Goal

    The first thing you need to do is to make sure that your 404 page is being tracked in Google Analytics. I’ve written about this already in another post, which also contains an interesting discussion in the comments about how to get the best out of error tracking if you’re starting from scratch. But this time round I’ll assume that you can identify your 404 page as either a pageview or an event in Google Analytics.

    Once you know how your 404 page is being tracked, you need to set up a Goal for it in Google Analytics.

    The next thing to do is to set up a Custom Alert based on that Goal. It’s true that GA will show you automatic alerts for unusual Goal data without you needing to configure anything. But I currently recommend using a Custom Alert for two reasons:

    1. Custom Alerts allow you to fine tune the timing and scale of the change which will trigger the alert
    2. Custom Alerts generate the oh-so-helpful ‘Major Contributors’ report

    Here’s a video showing you how to do those two stages:

    And that’s it.

    Now all you need to do is check the Google Analytics Intelligence reports each day and watch for those alerts. When you spot one, look at the “Major Contributors” then click through to the source and try to fix the problem. If it’s something someone has posted in a forum you’ve got a golden opportunity to sort out the problem and generally demonstrate your willingness to listen and to help.

    I’ve seen discussions with broken links to a promotional offer start to turn hostile because people assume that the offer has been suddenly withdrawn. This situation is a classic example of where you can turn the impression of your brand all the way from negative, through neutral, and right round to positive just by showing up and being helpful. You can rescue thousands of visits by using this technique if you move fast and handle things well.

    Final point on ‘what to do’ — don’t forget to add a Google Analytics Annotation to note what happened:

    Screenshot: Google Analytics Annotation for Custom Alert

    Why You Shouldn’t Rely Just on Email Alerts

    To repeat one point from the video: at the moment I think you need to check the reports rather than rely on the automatic email and SMS alert options.

    The Intelligence processing takes time and (in my UK time-zone) those emails don’t arrive until more than a day after the trouble began. With this particular alert you need to react as quickly as possible.

    In fact I also add the 404 Page Goal Conversion Rate to my morning dashboard so that I can spot anything nasty as soon as I start work. I use a desktop dashboard client which is configured to include the current day as well as the last last month so that I can spot what’s happening in almost real time. That’s good in case something really bad is happening. But then I need to do some manual work right away to track down the source because I don’t have access to that handy one-click route via the “Major Contributors” report.

    Use the Custom Alerts and Goals Combination Again and Again

    The example I’ve given above is a very specific and practical one which should be relevant to many sites and which can help you take action to fix something which is broken.

    You can also use the same technique to tell you when something good is happening and suggest why.

    For example: you may well already have a Goal set up for the “Add To Cart” micro-conversion. This approach could point you at the source of an unexpected increase in purchases.

    Better still, if you have a goal which matches the start of a “new customer” checkout, the “Major Contributors” report could alert you to a promising new referral source of business. This is the kind of thing you might not normally see buried in all the campaign activity to recruit new customers.

    Google Analytics Intelligence is a really powerful tool for spotting things that you may not be actively monitoring. It shows you things you weren’t looking for. You can take a look at that unexpected source and see if there’s anything you can do to get more of those people interested in you. In the age of social networking, simply showing up and showing you care can bring big rewards. Intelligence Alerts can help you take an interest in things you would not normally have noticed. Give it a try!

    Resources Related to Using Google Analytics Custom Alerts and Goals

  • How You Can Put Google Analytics Site Search to Work for You

    In part one of this series I explained why site search reports are so valuable. These reports contain the words actually typed by your visitors, so they give you valuable insight into the intentions, needs and language of your market.
    Illustration showing stick men lifting chart column
    The site search reports do a lot of the heavy lifting for you: they give you lists of words you should be adding to your site and they show you where you should put them. The search terms also provide quantifiable information about the level of interest in new products and items which have been discontinued too early.

    You can use this information to

    • Attract people to your site by using the words they use themselves
    • Make it easier for them to find what they want when they’re on the site
    • Gain valuable insight into demand for new products and other market intelligence

    In this post I’ll explain the specifics of how to do all that.

    This post is part two of a series:

    1. Previous Post: Why Site Search is So Important
    2. This Post: How To Use Site Search Reports in Google Analytics
    3. Coming up: Extras

    I’ll use Google Analytics for my example because it is widely used and because its Site Search reports work very well in this context.

    The two reports which are most useful are “Site Search: Search Terms” and “Site Search: Start Pages”.

    I recommend that you begin with this report because it gives you an overview of what visitors are searching for on your site. When you’re new to this process, the biggest gains will probably be found here.

    Screen grab: where to find Google Analytics Site Search Terms Report

    The moment you look at this report you may instantly see some surprising and useful information.

    Examples of things to watch out for include:

    • Terms which are different words from the ones which you use on the site
    • Words which are on the site, but people are using search to find them
    • Searches for new products which you don’t actually stock (but might)
    • Searches for discontinued items

    In the next sections I’ll show you some of the things you can do with the clues you’ve found.

    Your Customers Use Different Words

    It’s very common for manufacturers and retailers to use industry terms and jargon which are very different from the language used by shoppers in our stores or on our web sites. Those industry concepts affect our classification of products and the category navigation of our sites.

    • We design our sites to suit industry insiders, not the people using the site
    • But on-site search reports teach us the words our visitors actually use

    External search reports can do this too, but to a more limited extent — they only show the keywords which did result in a visit to the site. Which means that the relevant words probably are already on the site. These reports don’t give you information on those other words, the ones people are using which are not the ones on your site and which are not bringing visitors. Those potential customers don’t even get to your site. (By the way don’t forget: Google Insights for Search can provide some very useful on information on such phrases and the volume of people who’re searching for them.)

    The fix for this is obvious but it may be easier said than done.

    You need to change the language used on the site to match the words your customers use. Those trigger words are the ones they are scanning for when trying to find what they want to buy. These search clues may turn out to be the tip of an iceberg of lost sales.

    Many other people may have scanned for those words and decided that you don’t stock what they want. Search is the navigation method of last resort: those who actually searched were giving you one last chance. Other people would have gone back to their Google results and on to your rivals.

    • Changing the language in your product copy may be hard in between catalogues or seasons. But you need to do it
    • Changing the language used for categorisation and navigation is probably even more of an upheaval. But you need to do it

    As ever, the secret is to test. Choose the search term with the highest volume of searches. If possible configure a GA goal to match the URI of the relevant destination pages (regular expressions make it possible to match several different pages). This way you can establish a benchmark conversion rate for this micro-conversion.

    Make the change: add these words to your navigation, product names and description as appropriate. Then see if the volume of search for that term goes down and, more to the point, the traffic to the relevant pages via navigation goes up. Keep an eye on your organic search traffic as well. Since you’ve added a popular search term to your navigation structure, as well as including it in the copy, you should pick up some extra external search benefit as well. This could well start to form a virtuous circle, attracting more people who will see the trigger words prominently on the site and will be more likely to buy.

    Illustration: add the right words, more people will buy and search engines will send more people

    Your Site Navigation and Copy Use The Same Words, So Why Do People Search?

    If you spot a large volume of searches using words which are the same ones as you use on the site, you need to find out why people are needing to search for them.

    By ‘large volume’ I mean anything which stands out as having an unusual share of search when compared to the spread of other terms in the report, or compared to your expectations of the market. You can use the GA ‘performance view’ of the data table if you would like some percentages calculated for you — but it’s usually enough to eyeball the standard table to spot any interesting cases.

    Next you need to find out where those searches were taking place. Click on the search word to drill down to next level of detail. The default view will show the ‘Destination Page’ as the dimension in the table rows. For our purposes we need to see the ‘Start Page’ instead. You’ll find it within the ‘Other’ dimensions menu, which is easier to show than to describe:

    Screen grab: how to change site search dimension to start page

    On the left of this report you can see a list of the pages where people searched for that term. Where were people when they decided to search for whatever-it-is.

    [At this point I’ll mention something that you may find puzzling. It is not unusual to find that many searches start with an ‘entrance’. That means that the first page view recorded in the session was the search results page. A common reason for this would be if someone has used a link to a particular search results page on another site. Sometimes you may even choose to use such an approach for PPC advertising so that you land people on a specific dynamic set of products without needing to build a special landing page or category page for them. These pre-built searches do not involve the visitor personally starting a search, so they are not relevant here. If you discover that the term you are looking is one of these pre-configured searches, ignore it for now.]

    Next go to the page on your site where the term you are interested in was used. Ask yourself why someone looking at that page would decide to enter that term into search. Later you might need to get into the details of where the people who searched had come from, because that might have an influence on why they would expect to see ‘that term’ here. But it’s often enough to just accept the simple fact that a load of people were on this page and typed what they did into the search box. They were on the page. They wanted to get to something related to ‘that term’ and they could see nothing to click.

    In this case, where people are searching for words which are already on the site, you know what those visitors should have done to get to the right place. But it wasn’t obvious to them.

    The solution here is to make the route to the relevant page more prominent. That may not be easy to do if it would require making changes to the permanent navigation of the site. So it would be a good idea to do a test first in order to measure whether you can get an improvement and whether the return on investment would be justified.

    For the purposes of this test you need to create some form of manual change to the start page to add a prominent extra link. If it’s the home page or a category page perhaps you might have content-managed promotional areas where you could fit in some form of banner-type link. If people are ending up on a product page and then consistently searching for something else, then you’ve probably got a cast-iron case for using one of the “you may also like” slots, if they’re a feature of the site!

    Somehow or other you need to manually add that prominent route to whatever it is people on that page are searching for. Then monitor to see whether there is a drop in the level of search for that term and whether the micro-conversion rate for people reaching the relevant page goes up. If you set up a GA goal for the relevant page you could even use the “Reverse Goal Path” to see whether people were coming from the page where you added the manual link. And you need to track the increase in revenue from the relevant products to see whether the investment is justified.

    Market Intelligence

    Those points about “New products you don’t actually stock” and “Discontinued items” are important too. They’re another classic example of how web analytics can provide market intelligence. This work is not just about improving your website: the website should be a valuable source of information for the whole business. These search reports can give you information about the demand for different products.

    The value of the start pages report is that it matches the way many of us see our own sites. The structure is based on pages which is often the perspective of someone working behind the scenes. This can make the Start Pages Report a very practical tool for actually working on the site.Screen Grab: Where to Find Google Analytics Search Terms Start Page and Switch to Performance View

    The first thing to do on this page is to switch the view of the data to ‘Performance’ so that you can see the start pages listed in descending order with Total Unique Searches expressed as percentage share as well as actual numbers. You’re mostly interested in the share, but keep an eye on the numbers too. Search is not heavily used on most sites, other than ones in markets like books, music and technology which are particularly suited for searching, so you need to be aware of when to cut off and avoid wasting time on items which affect only tiny numbers of people.

    It’s likely that the homepage will be the top search-start page. If there’s a very steep drop-off, so that all the other pages have low single-digit shares of internal search, then that suggests that people are resorting to search at various times in various places. That’s generally a healthy sign. What you’re looking for are examples of other pages which are getting a higher share of search than similar pages on the site. That’s not happening in the example below, where the start pages are spread very widely throughout the site, with only the home page standing out as prominent.

    Screen Grab: Google Analytics Site Search Terms with Steep Drop off after Home Page

    Keep an eye out for pages in the top positions with a relatively large share of search.

    Category pages are worth a careful look. Even if the highest category page in the list does not stand out as having a particularly high rate of searches, it might be worth digging into this subject just in case. Use the inline search filter (now located above the data table in GA v5) to include only the category pages. That way you will be able to compare the share of search between category pages to see whether any are particularly bad. In order to do this, you’ll need to work out what pattern in your site’s URI appears in all the category pages. If you can’t do that, you may not be able to filter the report and will just have to study the results as best you can.

    Screen Grab showing how to use GA In Line search to filter results

    Category pages deserve this special attention because they are ‘routing’ pages. Their purpose is to help people find things via browsing navigation. So if people are switching to search on a category page, it suggests that browsing navigation has failed and the customer is becoming frustrated.

    You now need to perform the reverse of the process described in the ‘Search Terms” section above. This time you know where they started the search, so you need to drill down into that page in the report to see the internal search terms people used.

    Then you need to go and look at that page on the site and ask yourself why people on that page would search for that term, and what you can do to help them.

    In the case of category pages it could be that the customers do not have the same expectations about how different types of product should be grouped. This is often the case in my experience. The Search Start Page report can throw light on this by highlighting cases where a significant number of people expect a particular item to be in a category only to discover that it’s not there.

    Please treat these reports as powerful clues to navigation problems, not definitive answers. Moving a product from one category to another might please one group of customers but make life hard for those who have had no problem finding it in the original location.

    Why Do People Search For Things Which Are Already There?

    There’s another kind of category page issue which might be more surprising, but is less troublesome to fix. You may find that people on a category page are searching for items which are already displayed on that page. For some reason they have navigated to that page, or landed on it from outside the site, expecting to see some ‘search terms’ but they haven’t spotted them, so they search instead.

    Product names can be a problem here. If the product name does not include what the product ‘is’ and the category page only displays the name, then the word the customer is looking for is nowhere on the page. (You’ve also thrown away a Search Engine Optimisation opportunity as well, but we’ll concentrate on the needs of real people, not search engine spiders, for now.) You might have some short persuasive copy for each item, which can be a great sales tool — but that’s more text and less likely to be read, even if the relevant word is in there. You need to fix the name itself.

    If the name doesn’t include the trigger words, you’re depending on the customer recognising the product in a quick scan of the pictures. Which in turn depends on the customer knowing what the item looks like, which is not something you should assume in all markets. Even if the customer has an idea what they’re looking for, their expectation might not match the way you display your products. Perhaps you show them as part of groups, perhaps those groups are in a context such as a ‘lifestyle image’?

    People scan very rapidly when they are browsing through a site. The headlines and the images get a brief glance. The smaller text scarcely registers. There are loads of reasons why something which is obvious to you might not be obvious to your customers. The start pages report is telling you: “they came, they didn’t see, they searched…”

    As ever, the solution is to test. Select a problem page and see whether you can make a difference. I would start with the words, because they’re almost certainly the easiest to change. But if you can manage to include different types of product picture in your early tests, then you will learn lessons which you can apply to all product photography next time round.

    Once you’ve been working with this data for a while you will become familiar with the internal search terms used by your visitors. Ideally you will have taken action to make it easier to find things, by adding those words to your navigation or as on-page trigger words.

    But the search terms report has a continuing use once the first round of improvements is in place. You should monitor the report for increases in searches for new terms.

    Watch out for the sudden “out of the blue” appearance of a new term. This usually means that something is driving people to the site with an expectation that they are going to find something to do with that word. And they’re not finding it.

    A classic example would be a promotion you have launched, with a particular name or perhaps an offer code attached to it. If something goes wrong and the landing page for the campaign does not carry matching material and trigger words (the ‘scent’ of the promotion), then you may well spot this in the internal search keywords report.

    Another example would be people starting to search for seasonal products earlier that you would have expected. This could provide evidence that you should be adding your seasonal ranges to the site earlier and maybe launching some of your seasonal campaigns earlier.

    Perhaps at some point Google’s ‘Automatic Alerts’ and ‘Custom Alerts’ will start working with external and internal search keywords? It would be great to know about sudden increases in the volume of searches for various terms. But for the time being it’s useful to keep an eye on the internal search terms report to try to catch this happening.

    When you spot the sudden appearance of a search term like that, drill down into the search terms report and then use the ‘viewing’ option above the data table to change to look at the medium, source or other campaign data to find out where these people are coming from.

    If, for example, you find something like ’email’ is bringing visitors to the page but they are then searching for some promotional term, then the chances are that your email is promoting that term but the relevant trigger words are not on the landing page. Fix the page. ‘Email’ is just one example. Affiliate voucher and coupon campaigns are also prone to this kind of mistake. It even happens with paid search.

    Examples like the ones above can be expensive mistakes. So checking your site search terms each day can be a wise idea. These days that’s a doddle to do. The latest version of GA allows you to add Widgets to your dashboard, so you can have a list of your top site search terms right there for you to glance at each day.
    Screen Grab: How to add search terms to Google Analytics dashboard

    That’s it for now. This long article comes down to one simple point. Listen to the language your visitors are using. They’re probably telling you things you ought to know.

    In the final part of this series I’ll give a few more tips and some further sources of information and guidance.

    Meanwhile, here’s that navigation menu again:

    1. Previous Post: Why Site Search is So Important
    2. This Post: How To Use Site Search Reports in Google Analytics
    3. Coming up: Extras