Continuing our series reviewing browser testing tools we are having a look at CrossBrowserTesting, which allows you to check your website live across the major web browsers and operating systems and now also includes several mobile devices.

Previous browser compatibility testing reviews that you might want to have a look at include Mogotest and Adobe Browserlab.

The main difference between CrossBrowserTesting and other browser checking tools like Mogotest or Browserlab is that with CBT you can actually test your website ‘live’ by using the browser instead of receiving screenshots or a report showing you what is wrong.

This means that you are able to pick up issues regarding JavaScript menus or AJAX functionality that you wouldn’t see with a simple screenshot.

Of course, this additional offering doesn’t come cheap. Pricing starts at $19.95 per month for just 150 minutes of use. However, if you consider that the alternative could be to put together your own testing computers with all the operating systems and browsers you require then $20 a month might not be so bad.

Let’s see if CrossBrowserTesting is worth it.

Signing Up

CrossBrowserTesting is a paid service but there is a 7 day free trial with each package, which allows up to 1 hour usage for 1 person.

Unfortunately, in order to sign up, you do have to input your credit card details but if you cancel before the 7 days are up then you won’t get charged.

Once signed up you are dropped into your Test Center and have the option of running a live test, taking screenshots of a URL or installing the bookmarklet (more about that later).

As the main purpose of CrossBrowserTesting is the ability to run live tests, we’ll look at that first of all.

Live Testing

Clicking on ‘Run Live Test’ opens a new window with all the available operating systems and browser combinations plus the screen resolution can be picked from a list.

You can also view what additional software is installed on each operating system if you need applications such as Adobe Acrobat Reader, Flash, Silverlight, Quicktime, etc. This is extremely helpful so you can test the full website including animations, PDF downloads and video clips.

The availability of browsers and operating systems is pretty good with all versions of the main browsers available including Internet Explorer, Google Chrome, Firefox, Opera, Netscape and even older browsers going back to Firefox 1.0, Internet Explorer 5.0 and Netscape 4. You don’t have to test your website in these seldom used browsers but they are there if you need them.

Newer browsers are also present including Internet Explorer 9 RC and Firefox 4 Beta, not yet fully released but very useful to start testing your sites on.

I chose a Windows 7 OS that had Internet Explorer 9 RC installed at a resolution of 1024 x 768 in order to test the WebDepend website on.

The desktop loaded in pretty quickly and I double clicked on the Internet Explorer icon to get started straightaway. Browsing to www.webdepend.co.uk was also fairly quick, there is a slight delay as you type the web address into the address bar and it took a couple of seconds for the WebDepend website to appear but first impressions were that the system was fairly responsive.

There is an additional bar at the top of the screen, which shows the operating system you are using and also allows you to end the testing session, take a screenshot of the page you are viewing and record your sessions to playback later on.

I wanted to test the dropdown navigation on the WebDepend site, as this can’t be tested using screenshot applications, and so I hovered the mouse over each menu item to see how the navigation displayed once it had opened out. I did have to wait a second or two each time for the browser to catch up with where my mouse was pointing.

Using the scrollbar also took a couple of seconds to scroll to the part of the page I wanted to view and so I can see a lot of browser testing starting to become quite tedious in this respect (but browser testing can be fairly tedious anyway I suppose).

Another area that you can’t properly test with screenshot tools are enquiry or contact forms. I wanted to fill in an enquiry form to make sure the validation and submission worked correctly in IE9. The form was completed pretty swiftly and submitted fine, I received the test email within seconds so there was no delay in being able to test filling forms.

Ecommerce websites could make great use of CrossBrowserTesting to fully test the shopping process in each web browser and operating system configuration.

Taking A Snapshot

Whilst testing a website live, simply click on the snapshot icon and it will grab the screen that you are on and save it in your test center. You can then add notes to the snapshot plus what browser you were using at the time for future reference. A publicly accessible link to the screenshot can be shared on social networks.

One useful enhancement to this would be the ability to send snapshots straight into a bug tracking tools such as Unfuddle, Pivotal Tracker or Fogbugz and create a new bug. I would certainly appreciate being able to create new bugs in Pivotal Tracker, which we use at WebDepend, in this manner.

Recording A Video

Once you have recorded a video you can watch it back and add notes plus the browser you were using for future reference, as with snapshots. A publicly accessible link to the video clip can be shared on social networks.

Mobile Testing

Fairly new to CrossBrowserTesting is the fact that you can test how websites appear on an iPad, iPhone or an Android 2.2 device.

Click on ‘Run Live Test’ as with the Live Testing carried out above but choose one of the mobile operating systems to launch.

For this test I chose Android 2.2, which has the Android 2.2 web browser at a resolution of 320 x 480. I waited for around 60 seconds whilst my configuration was being prepared but then an on screen error message appeared to say that it had not successfully launched, the technical team had been informed and to try again.

Once I did try again, after waiting another 60 seconds, a virtualized Android operating system appeared and I could launch the Android web browser.

As described above, there is a slight lag whilst the browser interprets your mouse click or keyboard action but it works pretty well and you can successfully browse the website you want to test, using the Android device including navigation buttons and keyboard.

You will probably want a greater range of mobile browsers and operating systems if you intend on doing a lot of mobile device testing but having iPhone, iPad and Android represented is a good start.

Automated Screenshots

CrossBrowserTesting does also include automated screenshots with every pricing plan. In your test center click on ‘Take Screenshots’ and then input your URL including any advanced options such as whether a login is required or if you want to delay the screenshot by a couple of seconds.

Then select the browsers you need screenshots take for, you can customise the list or selecting all browsers will give you a pretty healthy list.

It is also possible to specify the resolution for each browser, set to 1024 x 768 by default or 640 x 960 for the iPhone 4. At present it doesn’t appear possible to have screenshots taken for an Android 2.2 device.

The final step is to click ‘Take Screenshots’ and CBT will start churning through all the screenshots and presenting them to you as they are taken.

Screenshots are taken pretty quickly with all 41 browsers returning a screenshot after about 3 or 4 minutes. I was running these tests at 2pm UK time, which is 8am where CrossBrowserTesting is based. Hence, all times are recorded as being 8am, which might confuse me when I come to refer to results and figure out when I was carrying out the testing.

Checking through the screenshots I spotted a small layout issue on the WebDepend home page when viewed on an iPad (I don’t own an iPad unfortunately and so hadn’t tested on this device previously) so the test highlighted that bug to me, which was useful.

But be careful as using the screenshots facility adds to the overall minutes used. I managed to rack up 31 minutes of usage in checking through the screenshots produced without realising.

Against each screenshot is a handy link to launch a live test session for that specific browser and OS combination so if you spot a problem you can quickly investigate in more detail plus test the rest of your website to find any further problems.

Bookmarklet

There is also a bookmarklet that allows you to take screenshots or launch a live testing session on any web page that you are visiting. This allows quick and easy access to starting your browser compatibility testing.

API

One of the areas not utilised as part of this review is the API, which allows you to automatically trigger screenshots of a URL and then email them to an email address to be checked. I suppose it could also enable you to send screenshots into a bug tracking system as described above.

At present, I believe only the screenshots functionality is included in the API but there are plans to extend that to the live testing side of the system too.

Conclusion

Overall, CrossBrowserTesting is a very useful service for any browsers or operating systems you don’t normally have access to. For me this would be several browsers including IE6, older versions of Firefox, Safari, Netscape, Opera and mobile browsers including the iPad, iPhone and Android 2.2. That’s quite a number and whilst I might not test everything on all the older browsers, I do want to make sure everything displays and works on mobile browsers plus some clients still require IE6 compatibility.

Another advantage I can see is for website developers and apps companies to react to users problems and try to reproduce the issue on the specific OS and browser combination that user has installed. For the smaller organisations they may not have all these combinations to hand and so a service like CrossBrowserTesting becomes essential.

I can also see scenarios where I am out of the office and away from my normal testing environment. Being able to test on up to 41 web browsers from anywhere I can get online is a huge bonus.

CrossBrowserTesting is very easy to use and get to grips with straightaway. There are little details that make browser testing easier, such as the ability to go into a live test from a particular screenshot you had taken. Having the bookmarklet and API add to the overall sense that the application is easy to use however you want to use it.

On the downside there are small delays in using the live testing so over a full session that means browser testing takes longer than it would if you were using the machine locally.

The main drawback is that minutes soon rack up, whether you are using Live Testing or Screenshots so the base package of just 150 minutes per month would get used up very quickly. I wonder whether the ability to purchase extra minutes or a pay as you go service would be something for the folks at CrossBrowserTesting to consider.

However, even with the slight negative points, using CrossBrowserTesting has to be less hassle, cheaper and far more convenient than getting together all the operating systems and browser combinations available plus the maintenance and space required to house all the equipment.

I certainly know what I prefer.

Adobe Browserlab takes screenshots of your web page in several different web browsers

Adobe Browserlab is a browser testing service that allows you to check how each page appears in a variety of mainstream web browsers.

The web-based application allows you to compare your web page in 2 web browsers side by side and in that respect, shares a similar principle with Mogotest, which I reviewed recently. Read the Mogotest review.

One benefit to Adobe Browserlab is that it is free for 1 year if you sign up before 30th April 2011.

Sign up

First of all, you have to sign up to Adobe CS Live in order to gain access to Browserlab. By signing up to CS Live before 30th April 2011 you get 1 year complimentary access but we are not sure what the costs will be once the first year is finished.

CS Live includes other products including CS Review, Acrobat.com, Adobe Story and SiteCatalyst NetAverages.

Features

Once you have signed up to Browserlab, you can check your web page by inputing the address and selecting the browser you would like to view it in.

Adobe Browserlab then fetches the screenshot in that browser and presents it to you.

There are several different views available to you, as follows:

1-Up View – the default view, this allows you to see the screenshot of the web page in a single web browser.

2-Up View – the view I use the most, you can compare screenshots from two different browsers side by side to see where your web page may differ in one browser over another.

Onion Skin View – an interesting feature and places the screenshot of one browser over the other in a semi-transparent view, allowing you to see more easily the differences between each browser.

Additionally, there are the following features:

Smart Align – allows you to select a specific point on your screenshot and then determine which of the other web browsers display the screenshot at exactly the same alignment. This helps to pick up graphics or other elements that may be slightly out of line from browser to browser that you cannot easily see.

Delay – sets a delay for any animations that come in or start after a certain number of seconds.

Integration – Browserlab integrates with CS5 but I am not sure how, as I don’t own a copy and so have not covered that particular aspect in this review.

Screenshots rather than full testing

Adobe Browserlab only takes screenshots of your web page to display in each browser. This means that you can only realistically check for layout issues rather than being able to actually browse the website with that web browser. It also means that you will not be able to pick up JavaScript issues or items that may occur as you navigate around the site.

For full testing you need to use something such as CrossBrowserTesting.com or Spoon.net/browsers.

Browser Sets

This allows you to set up groups of browsers that you would like to test your web page in. You could create browser sets to test in all the IE browsers for instance or a list of browsers that you support as standard or any browsers that you don’t have readily available.

Supported Browsers

Adobe Browserlab supports all of the main web browsers including the following:

Windows:

• Internet Explorer 6
• Internet Explorer 7
• Internet Explorer 8
• Internet Explorer 9 Beta
• Firefox 2
• Firefox 3
• Firefox 3.6
• Firefox 4 Beta
• Chrome 7
• Chrome 8

Mac:

• Firefox 3
• Firefox 3.6
• Safari 3
• Safari 4
• Safari 5

The range of browsers supported is better than Mogotest but you do have to check each one manually for issues rather than being given a report in the case of using Mogotest.

Conclusion

Adobe Browserlab is a very useful browser testing tool. You gain access to browsers and browser versions you may not have easily available to you, in my case I don’t have Internet Explorer 6 or older versions of Firefox or Safari installed on any of my machines.

Adobe have created a tool that is pretty easy to use and if you have CS5 then Browserlab is probably even easier to use. The 2-up and onion skin views allow you to easily compare one browser with another but I can’t see myself using the smart align feature that much to be honest.

Browserlab is a good free (for now) service but we’ll see what happens if and when Adobe starts charging for it.

Sometimes however the issues can be much more important – whole pages can be rendered incorrectly by certain browsers and, if not checked, it may mean that users visiting your website with that browser cannot use the site.

Most designers, developers and project managers manually check the website in each of the main web browsers, which is time consuming and requires detailed and concentrated testing of each page or page type within the website.

Introducing Mogotest

Mogotest aims to make browser compatibility testing easier

However, all is not lost, a new web app aims to make the process much easier. Its called Mogotest and takes the strain out of browser testing by comparing how your website displays in each browser and giving you the results.

Free Test

To see Mogotest in action you can run a free page test from their home page, which tests a single page and shows you the results once completed so you get feedback straightaway. This is a great introduction to the Mogotest service and gives you instant access to many of the features of a paid plan. Mogotest includes up to 3 page tests for free.

Pricing

Starts at $15 per month for a personal plan, which gives you a single user account, up to 50 tests each month across a maximum of 3 sites. You don’t get certain features such as being able to test authenticated pages or being able to schedule tests. These features come along in the freelancer plan, which is $45 per month. Pricing goes up much more steeply after that with a Team plan costing $125 per month for 10 sites and the Agency plan is $595 for 50 sites.

Running A Test

To run a test you follow a 4 step process:

Configure Site – just input your website address and tick if you want to see the advanced options.

Mogotest is easy to get started

Create Test Group – here is where you add paths for each URL that you want to test specifically. If you have lots of URLs to test then this could be a bit more time consuming but nothing like how long it would take to test each URL manually.

It is also possible to schedule tests to run every day, every 2 days, etc., which is a great feature for sites that are undergoing regular updates. Within this step you can also set whether the test should wait for any Ajax to load in before starting.

For my test I added 3 URLs, all for the WebDepend website including the home page, one of the service pages within the Website Testing section and the Blog main page.

Once you’re happy you save your test group and proceed to the next step.

Configure Authentication – you can set whether your site requires authentication, useful for being able to test password protected sites that are still in development or for testing pages within a members section.

Prepare Test Run – final step shows all the items you have configured and allows you to go back to edit settings before clicking on Run Test!

Test Results

Whilst the test is running you can view a Test In Progress page that shows how your test is coming along and the estimated time to complete (900 seconds) although my test took far less than that.

Once the test has run your screen shows 3 tabs:

Easily see which browsers your test failed in

Browser Compatibility Report – from the screenshot you can see that the WebDepend site’s blog page and home page did not pass for Firefox 3.5 on Linux (not a combination I regularly test for it must be said) but that all other tests passed successfully.

Clicking on the items that didn’t pass will display the browser rendering results where it will show the browser you have clicked on side by side with your reference browsers (mine is Firefox 3.5 on Windows) so you can easily pick out the differences.

Any issues will be displayed underneath the side by side view, in this case 3 elements for my Blog main page were incorrectly positioned in Firefox 3.5 on Linux when compared with Firefox 3.5 on Windows.

As it turned out, the issues that Mogotest had found were very minor problems but you can see how useful it is being able to find these problems by running a test rather than manually trying to pick out browser related problems.

Site Problems – shows any HTML or CSS validation errors, broken links, server errors, empty content warnings or redirects. Mogotest found 5 validation errors and warnings that were not picked up when I manually ran each page through W3C’s validator a couple of weeks ago.

Validation Reports – displays a list of the URLs you tested and shows which of those have problems. Clicking on a problem pages takes you to a view of the HTML for that page with row numbers against each line of code (very helpful) plus a list of the issues found underneath. Rolling your mouse over an offending line with also display the issues, which is very helpful when investigating the problems that need attention.

Once you have fixed any issues found, you can go back into Mogotest and click Retest in order to carry out the same tests again, which will hopefully now pass for each browser.

Conclusion

Overall, Mogotest is a really useful service. By automating a lot of the browser checking process it really does make life easier and will pick up lots of layout problems between browsers that could go unnoticed or take time to find.

However, some manual testing is still needed in my view, it is important to browse and use the website in each browser to notice issues that Mogotest is not yet capable of, such as Ajax related content or JavaScript menus.

The browser support could be extended, as Mogotest is running tests in Chrome 5 whereas we’re up to Chrome 9 now and there is currently no support for Mac browsers at all.

If you run tests regularly, across several sites, you could find yourself running out of available tests pretty quickly and needing to upgrade. Also, there doesn’t appear to be anywhere that I can see how many tests I’ve got left or if I could purchase a block of tests without having to upgrade to the next plan up.

When running tests, I was lucky in that there were a small number of issues found. I wonder for projects in early stages of development what happens when there are large amounts of problems to wade through and how easy it is to see the most important ones that need fixing first.

A good addition to any website tester’s, web developer’s or project manager’s list of testing tools to make life easier, Mogotest does do what it sets out to and can only get better from here.

Loop11 - User Experience Tool

Web-based usability testing tools are starting to become more prevalent these days with the likes of Clixpy, Loop11 and UserTesting all appearing recently.

We shall have a look at Loop11 first of all, which has been in private beta for around 4 months now but is soon to be released to the general public.

According to their web site – Loop11 is a web-based user-experience testing tool, allowing companies to conduct online, unmoderated user testing on any kind of digital interface. Loop11 is not a survey or web analytics tool, but a user experience tool… helping you to understand user behaviour.

Sounds good so far, so how do you get started?

First of all, once you log in, click on the big ‘Create New User Test’ button in bright orange so you can’t miss it. You will be presented with a form with the following fields:

Public Title – this will be seen by your test participants
Working Title – used to identify your test for you only
Copy User Test From – so you can duplicate an existing test or just start with a blank one.

Loop11 - Create User Test

Once you have filled those fields in you enter into a 5 step process to complete the creation of your user test.

Step 1 – User Test Details
This includes the ability to set your language and Loop11 supports over 40 languages.

Step 2 – Tasks & Questions
You have the ability to add tasks or questions to your user test and Loop11 provides some help if you are unsure what to do in this respect. A task would be something that you would like the user to perform such as find a particular piece of information on your web site and to add one you give it a name, a scenario that tells the user what it is you want them to do, the start URL where they commence the task and the success URL, which the user has to navigate to in order to complete the task.

I created a task to ask the user to find step 6 from the article ‘How To Test Web Forms in 7 Steps’. When the user is carrying out this task they can either abandon the task if they could not accomplish it or mark it as complete.

To add a question, you choose the question type from a list including multiple choice, rating scale and open ended. So you could ask ‘How Usable In This Web Site?’ and ask the user to rate on a scale from 1 to 10 or using words such as Poor, Average, Good, Excellent.

I entered the question, ‘How easy is it to use the search facility on this web site?’ and gave the possible answers as being Poor, Fair, Good and Excellent (I copied them from Loop11′s help text) plus I made the question mandatory.

If you want the user to type in an answer such as correctly answering how much one of your services costs then choose one of the open ended questions.

Each time you had a task or a question you are presented with the Step 2 – Task & Questions screen helpfully showing all the items you have added with the ability to either edit, delete, duplicate, preview or move them up or down in the list. Once you are happy with what you have set you click ‘save and continue’ to move onto step 3.

Loop11 - Tasks & Questions confirmation

Step 3 – User Test Options
This is where you set the number of participants that you would like to complete the user test, from 1 to 999 and specify the thank you text the user sees on completion of the site. There are some other useful settings here such as being able to block or allow specific IP addresses.

Step 4 – Invite Participants
So now you get to invite the lucky participants to take part in your user test. Here you have 3 options:

i) Create a link to your user test – Loop11 generates a link for you that you can distribute any way you like.
ii) Create a popup invitation that can reside on your own web site – Loop11 gives you the code for a popup window to be placed on your web site.
iii) Purchase participants – Loop11 gives you the names of 6 companies that you can potentially purchase participants from with a link to direct users to your user test. This seems to essentially be the same as choosing item 1 although you get the details of 6 companies that you can buy users from.

Step 5 – Launch
Once you have sorted out how you are going to invite your participants you end up on the final step, to launch your user test. This gives you a summary of what you have selected and allows you to preview your test before you launch it. You should be careful because once you have launched your test you are not able to go back and edit it unfortunately.

If, like me, you realise that the questions you have set make little sense to someone seeing your web site for the first time then you should go back and edit them now before you finalise your user test.

Preview – when I clicked on the Preview button when still on step 1 I got a message telling me there was an internal server error and that someone had been notified of it. Once I moved onto step 2 the Preview button did seem to work fine and showed me the first couple of pages that a user will see when participating in the web site evaluation.

One of the tasks I had prepared was for a visitor to submit an enquiry form because I wanted to test that visitors could use the enquiry form including field validation. When previewing the test, I submitted an enquiry but my enquiry form kept telling me that I had not input the validation code that I have to prevent automated software from filling it in (even though I had input the code). This is possibly due to Loop11 conflicting with the validation method of my enquiry form so I removed that task from my test. I will raise this separately with Loop11 because many tests could require this type of input and so the preview does need to be able to handle that aspect.

Launch – once you click on Launch you will be presented with a summary page showing the details of your test plus the URL that will link to the test that you can distribute. If you elected to choose that you wanted the invitation for the test to appear in a popup window then the details of the JavaScript that you need to add to your web site will be given to you.

Loop11 - Launch screen

My Projects

Once you have launched your test you can view the results from within the My Projects tab. For each project you can see the date it was launched, how many responses received, its current status and then you can view the design of the project (what tasks, questions and other settings you selected when creating the project) and its report.

View Report

Once your test starts gaining participants you can view your report, which gives you access to the following screens:

Dashboard – first of all this shows the average task completion rate for your test presented as a pie chart. As I only had one task I can see that the average completion rate was about two thirds whilst one third abandoned the task. Underneath is presented the task results overview which shows the average page views and average time taken to reach a resolution for each task. You also get a bar chart showing the task completion rate for each task.

You are able to export the results of the report at any time in either CSV, Excel XML or PDF formats, which I think is pretty useful for any presentations that you are putting together and/or any followup actions required as a result of this testing.

Loop11 - Dashboard screen

Tasks – this page shows more detail for each task including the completion rate taken from the dashboard but also further aspects such as page views to complete task, time to complete task, most common success page, most common abandon page, most common first click and most common navigation path.

There is also a function called Participant path analysis, which allows you to go through each participant and what they clicked on when completing the task.

These reports give you invaluable information for how users approach each task that you have set them, what they click on first (possibly giving you the most eye opening results), how long it takes them to complete the task and more.

Questions – this report goes through each of the questions that you posed for your users and the results or answers that they gave. For the question in my test, ‘How easy is it to use the search facility on this web site?’, two thirds of respondents felt it was good and one third said it was excellent. However, for my open ended second question, ‘How much does the monthly link check service cost?’ I got a variety of answers where I think respondents did not really understand the question I was asking.

Not only are you testing how users experience your web site, you are also testing your ability to set good tasks and questions that your users will understand.

Participants – the participants section gives you detailed information on each person completing the test including the time they spent on the test, average time taken per task, average page views per task, their IP address, their user agent or browser and the date they completed the test.

Loop11 - Participants screen

Overall I think the results that you get are pretty comprehensive and will allow some very worthwhile followup actions to be completed to improve the user experience of your web site.

The pricing of Loop11 has still to be announced and, if the pricing is not too steep and indications that I have received from Loop11 are that it will be a reasonable price per test, my belief is that you will get a lot of great feedback that will easily pay for the cost of the test as long as you think carefully about the tasks and questions that you set for any participants.

Update (26th August) – I have been informed that pricing for Loop11 will be set at $350 USD per project, which includes unlimited tasks and questions and up to 1,000 participants. This pricing will be in place once the private beta concludes on 1st September.

LinkAlarm - Link Checking software

LinkAlarm is a web-based link checking application that has been around for over 10 years and scans your web site checking each and every link on each web page. You can have LinkAlarm check your links once a week, once a month or you can run the scan manually whenever you like.

So the principle of LinkAlarm is straightforward enough, let’s see how it performs.

Signing Up

You can register for a free account, which is a 14 day trial that gives you enough credit (more on credit later) to check links on up to 100 pages.

Registering is easy enough, a few details about you (your name, email address and how you found LinkAlarm) and your web site (the URL, how often you would like to check – once only, weekly, every 2 weeks or monthly and the category of your web site).

Once you have registered you are presented with a screen with a welcome message, your site details and a further message saying that your site will be checked in 5 minutes and notification of the report will be emailed once finished.

LinkAlarm - Welcome Message

I waited 5 minutes and nothing happened, no email. So I went back to the screen and read it again. It appears I had ignored some text that said ‘Please continue by confirming the entry of your site’ but there was no confirm site button and I had got confused. I clicked on the ‘Site Control’ link so see what that did and was presented with a page where it said ‘Checking’ in flashing green text. Good enough for me, I’ll leave it a little while and see what happens.

LinkAlarm - Checking Site

So, whilst that is running let me explain how the pricing structure of LinkAlarm works.

You pay for credits to be added to your account and then every time you run a report (either manually or automatically) LinkAlarm deducts 1 cent for each page of your web site. So if you have a 20 page web site you will pay 20 cents every time a report runs.

In my book that is pretty good value but is obviously more expensive for larger web sites and how often you want LinkAlarm to scan your site.

About half an hour later I received an email from LinkAlarm with the results and links to access more detailed reports online.

The Results

Within the email that LinkAlarm sends to you there is a section displaying your overall link check results. So my Testing Web Sites blog has 18 pages (still pretty small but growing) with a score of 64 and a failure of 3.7% (oh dear, really?). I had spotted that the category average for my web site is 2.5% so a failure percentage of 3.7% is a bit embarrassing and worrying. Let’s have a look at that in some more detail.

The email also gives you a number of statements (I won’t list them all here), which give you a quick impression of how well your web site performs in terms of internal links, external links and an overall score.

So, here are a couple of statements from the scan completed for www.testing-web-sites.co.uk, the comments are my comments relating to each statement:

Of the 18 pages checked, 18 (or 100%) were found to have one or more alarms.

Comment – so there is something broken on every page of my web site, great.

LinkAlarm checked 53 internal links and found 1 (or 2%) need attention.
Internal links are URLs that point to files inside your site.

Comment – there is 1 link that has a problem and it is probably on multiple pages, which is why I am getting such a high failure percentage.

LinkAlarm also checked 29 external links and found 2 (or 7%) need attention.
External links are URLs that point to files outside your site.

Comment – again, just 2 links that need attention but a high percentage because of the relatively low number of external links I have in place at present.

The LinkAlarm Score for this site (64) is an average of the page and alarm
ratios above. A score of 100 indicates all links checked reported no alarms.

Comment – I am guessing at the moment that a score of 64 is not great.

The link failure rate for this site (3.7%) is worse than
the benchmark link failure rate of 2.5% for the category – Computers & Internet.

Comment – I had spotted that too so will look into those failing links now to see what is wrong.

What the email from LinkAlarm also gives you is an understanding of what it would cost to run this report regularly on your web site plus links to their order form to order some credits. The costs are based on the number of pages and how often the report is run, for example, once each week. As I explained above the costs are based on 1 cent per page so if my web site stays at 18 pages then it would cost 18 cents to run each report or just under $10 each week for a year, not exactly big bucks.

It is helpful to know how much it is likely to cost to keep LinkAlarm checking my site but the only problem I have with this costing model is that I am aiming to grow my web site a lot over the next few months and so my costs to run LinkAlarm will continue to grow as the number of pages in my site increases.

So let’s have a look at the report now.

The Report

A screenshot of the LinkAlarm report summary

As mentioned in the email, the report gives you a summary first and the average for your category.

Then comes an Alarm Summary that shows the type of link failure for internal links, external links and total links. As you can see one of my internal links has a 405 error, which is Method Not Allowed. If you don’t know what this is then there is some handy help text available to give you an explanation.

405 Method Not Allowed means:
Common Alarm. Most commonly seen in ACTION links on FORMs where the server has not been configured to allow POST operations.

What you can do: Enable the ACTION for the correct location in your server configuration.

Clicking on the number of links shows me the pages that LinkAlarm found with the Method Not Allowed issue. This problem is being picked up on a PHP script relating to the posting of comments and is found on 7 of my 18 pages. At first glance this looks like an issue to do with my WordPress blog or perhaps the theme that I am using, which will need further investigation and is unfortunately outside the scope of this review.

Now let’s have a look at the external links, the 2 errors are showing as 403 Forbidden. Again, some help text gives us a description of what that type of error is.

403 Forbidden means:
Common Alarm. The server will not provide the requested file. Commonly seen where content from an intranet has been made publicy available on the public internet or when your link points to a directory on a server and that server is configured to not provide directory listsings.

What you can do: Remove protection or provide and index file in the directory if necessary.

Both issues relate to links in the footer of each page of my blog, linking to the theme’s author. Clicking on those links manually does not result in a 403 Forbidden message but obviously LinkAlarm was getting that response.

So how does this happen? My view is that my theme’s author’s web site has some checking in place that works out if an automated bot is trying to access its pages and maybe responds with a different response code (403 in this case) than if my browser were trying to access the page.

This gives me a bit of a problem in trying to tidy up my LinkAlarm report because the software is going to think there are issues every time the report runs whereas in actual fact there does not appear to be any issues with the quality of my external links. This problem then is also not going to be restricted to just LinkAlarm but will extend to other link checking software too.

LinkAlarm, however, allows you to ignore certain pages or URLs and so you can maintain a list of any links such as the ones above. After ignoring the 2 external links that were producing incorrect 403 errors I ran the report again. This time I got a much more satisfactory result of 1.3% failure rate and an overall score of 86. There is now just that 1 internal link to investigate and deal with.

Other Features

It is worth detailing some of the other main features that you get as part of your LinkAlarm account, as follows:

History – shows you the history of the reports run with a graphic and summary information. Useful to see the improvement or worsening of your link check reports over time.

Download – you can download reports as zip files so you can keep copies of reports for later use.

Ignore URLS – as mentioned above.

Checking Limits – you can limit the number of pages that LinkAlarm checks if you want to. May be useful for larger sites that could stretch into hundreds or even thousands of pages.

Password Access – LinkAlarm can check password protected sites so useful for Intranets or web sites still in development.

Sites – the ability to add multiple sites to LinkAlarm so you can have it regularly checking all of your web sites using the same overall account. A good feature for agencies or larger companies who maintain a number of web sites.

LinkAlarm screenshot showing sites feature

Conclusion

Regularly link checking is a good thing and means that you are able to pick up on any link quality or linkrot problems at an early stage. Just the report run today on Testing Web Sites has raised some interesting items.

There is, however, only so much that any automated link checking software can do and this highlights the point about manually checking each and every link when you first complete an update to your web site or add a new page or blog post.

LinkAlarm is easy to set up and its free account enables you to start checking links very quickly. The ongoing costs are not great for smaller web sites but I wonder how LinkAlarm would perform for large, dynamic web sites such as an ecommerce site where they are potentially thousands of individual pages (although you can limit the number of pages or check specific pages if you want to).

The additional features that you get mean that LinkAlarm is worth considering for a wide variety of different companies and I will be continuing to use it unless I can find a better alternative.

Overall, I would recommend LinkAlarm for static web sites and possibly some further testing would be needed for larger and database driven web sites.