Posts filed under 'Web Analytics'
Google recently announced a new addition to their Google Trends tools called “Google Trends for Websites”. In short, you can now use Google Trends to see traffic estimates, related sites, and related searches for website URLs you enter into the tool by regions and over different time periods.
According to Google, the data they use for this service is from Google.com searches, the Google toolbar, and opted-in Google Analytics data. This would lead one to believe that if you use Google Analytics as a publisher, you should choose to opt-in your data if you want to see the most accurate results possible in the trends tool.
As an overall web analytics fan, I’m always happy to see new sources of data, especially in the public website traffic field where the tools out there are all trying hard but still don’t quite deliver the goods. The players in the field are Comscore, Hitwise, Alexa, Compete, Quantcast, and possibly some others I’m forgetting about at the moment. They all have some qualities I really like, but it’s hard to totally trust any of them yet.
So will Google Trends for Websites break away from the pack and finally be a tool we can trust? How does it stack up against the competition? Well, in my first few minutes of playing with the tool I came away with mixed feelings.
Positives
1. It’s another source of data, and different data from the competition. It’s the only one of the competitors in this field that can use the true Google search, analytics, and toolbar data.
2. Viewing trends over regions is cool and helpful. Andrew Chen wrote a cool post that compared Myspace to Facebook by country and mapped it against the largest ad markets to come to some interesting conclusions.
3. Seeing related sites is helpful to know what kind of neighborhood Google thinks a site is in, and it could be used to find sites in the link building process for your own site.
4. Seeing related searches is cool for your own site but also helpful to know what keywords your competitors are getting traffic on from Google.
Negatives
1. As expected, it appears that Google relies heavily on data from Google Analytics. If you’re a publisher using that tool you have more accurate data in the trend tool. If you’re not using it, the data gets poor. As an example here’s a graph of the traffic to Right Media’s website:
As you can see, the data essentially becomes non-existent in June 2007. I happen to know this is the exact time we stopped using Google Analytics as our web analytics tool because we were acquired by Yahoo!. It seems that Google no longer had enough data to track RightMedia.com. While not the highest traffic site in the world, Right Media gets enough traffic and is a well enough known company that you’d think through Google search and toolbar usage there would be some data to go on.
2. No data for smaller sites. For one, this blog shows no data in the Google Trends tool. I definitely know it gets traffic from Google, and I’m running Google Analytics. Wakeboarder.com, a site I used to own before selling a couple of years ago gets on the order of 3 million page views a month and uses Google Analytics and shows no data in Google Trends.
Hopefully they will continue to move down the tail and find a way to have data for sites like these. When the competitive tools do have data, it will mean Google Trends is useless unless you’re looking at huge sites.
3. Lack of accuracy. As Fred Wilson pointed out in a comparison of Comscore with Google Trends, he says that Comscore jives with the actual traffic statistics of his portfolio companies while Google Trends isn’t directionally correct. Ouch.
4. Google’s sites show no data. Mike Arrington theorizes this is because Google toolbar data would really skew these numbers, and he’s probably right. Although Google gave an excuse about it being as if they were giving interim financial guidance, which since it’s an estimate would not be true.
What is the answer?
If even the mighty Google can’t get it right using their massive amount of data, how can people get reliable traffic measurements and comparisons?
For the time being, I’d recommend using all the available services to provide a general view. If all of them show the same thing, it probably means it’s generally correct. If one service is off, take a look at how that service tracks data and see if that would explain it.
For the future, I like the sound of the stealth Mozilla data project that is rumored to use the Firefox browsing data to show traffic statistics. This would be a huge chunk of data, but also would have its own bias. It would give a boost to sites that skew towards Firefox users, and would cut out all Internet Explorer users (still the largest browser market), Opera users, and iPhone and Mac users using Safari.
Perhaps Microsoft could launch a service using Internet Explorer data? It’d be interesting to compare that to Firefox data.
I still like the idea that Quantcast has which is to use numerous data sources to estimate while also allowing publishers to get “certified” and put a Quantcast tracking tag on their actual website. This insures they will get the full set of traffic across browsers and users that hit that site. However, the big problem they have is that publishers have to know about this, agree to do it, and then suffer through the fact that anytime you add more javascript code to your site it has a performance hit on the speed.
Overall, we don’t have the answer yet, but at least Google is providing one more data source that we can all scratch our heads over until a true solution is discovered.
June 22nd, 2008
Lo and behold, screenshots of Microsoft’s upcoming web analytics application Gatineau are surfacing on the web along with an interview with Microsoft’s Ian Thomas about the application.
Following on the footsteps of Google Analytics, we have another large player with what looks to be a powerful and free web analytics application. It looks like it will have a heavy focus on ecommerce and demographic segmentation which is important since the goals of these free analytics apps are to get the AdCenter and Adwords advertisers to spend more money on their ad platforms. Like Google Analytics, Gatineau can be used on it’s own without being an AdCenter advertiser.
The beta invites are supposedly going out next month, sign up here if you’d like to get in line.
September 18th, 2007
Read/Write Web does a nice roundup of 10 free and innovative web analytics tools. I’ve used most of them but I’m excited that they brought up a few new ones to try. My thoughts on the tools that I have used they mention:
Clicky
My favorite simple web analytics application of the moment. Some really nice features, it’s easy to use, and great for the blogs. The Spy feature and RSS feed control are highlights for me.
CrazyEgg
A really cool way to generate heatmaps of your site. Heat maps can be used for general information about where people are clicking, but can really be helpful for navigation changes and ad placement.
103Bees
103Bees is a cool way to see more about your natural search results and identify long tail phrases you may be able to expand more on to get more traffic, but it kind of feels like it’s just a subset of a web analytics application. There’s just not enough there that I want to use that much to keep me coming back to it.
MeasureMap
Like Clicky one of my favorite of the simpler analytics applications, but it never came out of alpha as Google bought it and ported a lot of the functonality and the team over to Google Analytics. I’m not sure if it should really be included on lists like this anymore since I don’t think they’re really giving out new accounts or continuing to develop it.
FeedBurner
The most comprehensive RSS analytics application that also provides numerous other features relating to feed management and monetization. It’s nice they also bought BlogBeat a while back and added in some nice basic web analytics to make it a “full” basic solution.
ClickTale
Like CrazyEgg a provider of heat maps, but different ones and with the recording of user sessions using your site. ClickTale has heat maps that show you how people scroll, and how they hover on links and for how long. They also allow you to record user sessions so you can actually watch users scroll and click around your site. Very interesting data.
MyBlogLog
There is a lot of functionality here beyond analytics, but MyBlogLog is also a good example of it being handy to use one piece of code to get good widget functionality, community, and analytics in one package.
August 30th, 2007
I’ve said it before and I’ll say it again, I like what Quantcast is doing in trying to measure the web. Everyone complains about the inaccuracies of Alexa and other similar ranking tools, and they have good reason to complain.
The reason I like Quantcast is because they actually give publishers code to put on their site so you can actually send them real traffic data in which to rank your site accordingly and report interesting demographics and other information. As a publisher not only will you be ranked more accurately, but you can use the information to help pitch your site to advertisers.
Unfortunately the downfall of their system so far is that it requires publishers to actually take that action to install the code, and I think it’s always going to be an uphill battle to get enough of the web to do that to make their ranking system as accurate as it could be.
Either way, they’ve announced today that they’ve added the ability to add their code to Flash videos and widgets so that data is added to the mix as well. Instead of summarizing it all in my own words, here’s the highlights from their press release:
—Tabbed Quantified Publisher Profiles: Quantcast profiles will
always feature the popular summary view for each online property. But
now they include an enhanced traffic tab that provides greater depth
and insight on key audience reach, engagement and frequency metrics.
Key metrics, including unique visitors, visits and page-views, are
presented in dynamic daily, seven-day and trailing 30-day charts. In
addition, a traffic summary automatically calculates your average
visits per unique visitor and page views per visit. Coming soon, this
tab will include time-on-site metrics.
—The Quantified Publisher Network View: This new interface –
presently in beta – enables you to track your portfolio of properties
(websites, videos, widgets, etc.) in a single dashboard. It also gives
you one-click access and a single point of control for the Quantcast
profiles on all of your sites and services.
The Quantcast Network View features the new traffic tab with the same
easy-to-use dynamic charts to provide a rich overview of network
metrics. Unique visitors (“uniquesâ€) are reported as unduplicated
network reach, and the other network metrics are shown in the
aggregate. It’s created when you quantify a second site or add a media
element. As always, you can adjust your settings to control which
elements the world can view.
—Secure Quantcast Measurement Tags: Paired with conventional
Quantcast measurement tags for content pages, new secure Quantcast
tags enable you to measure traffic to your transactional ‘https’ pages
with no interruption to check-out, and no affect on the shopping
experience.
August 15th, 2007
Centernetworks has a nice background interview with Sean Hammonds who is one of the founders of analytics solution Clicky.
I’ve mentioned it before, but Clicky is currently my favorite blog analytics solution. It isn’t as high-powered as a lot of solutions out there, but it has a very usable interface and some nice features not seen in other applications. What’s maybe the most impressive thing though is that Clicky is self-funded and built by just Sean as the programmer, and his partner who handles business matters. Great applications can be made by a small team!
July 23rd, 2007
I love web analytics and traffic rankings, but unfortunately the web still suffers from a lack of reliable and accurate comparison metrics between sites. Nielsen announced last Tuesday that they are ditching the page view in favor of the “time spent” metric. The question though is if this is a good thing?
Many people have criticized the page view over the last few years as sites like Myspace that are very page view intense rise higher in the rankings than sites that may just do a better job of not making users click around so many pages. Time spent as the metric is a chance to rectify that by making the amount of time a user spends on the site the most important thing. That seems logical, but it has a lot of flaws as well.
Just like the page view metric provided incentive for sites to actually make the user experience worse by creating more pages needing to be viewed to get a task accomplished, time spent also provides incentive to make the user experience take longer than necessary. If pages load slower it can help, if things take a long time to download it can help, or just keeping all those extra pages around to make people take longer to accomplish tasks can help your time spent metric.
It also gives a huge advantage to video sites where a user naturally stays engaged longer to watch videos that need to download and then require watching time. Expect YouTube to rocket in the Nielsen rankings, while a site like Google.com itself which gets the user to it’s goal quickly and off of Google.com will suffer.
If you leave your Yahoo Mail or Gmail open all day, that will add a huge boost to those companies time spent rankings, but does it really mean they should be listed higher because of that? Maybe, but I’m not so sure. It’s also been said that AOL Instant Messenger usage will count for AOL.com, giving it a huge boost as so many people leave that application on 24/7. In no way does running AIM constitute “website usage” to me, but maybe I’m crazy.
So how can we truly measure a user’s engagement with a site? What matters most? How much time users spend? How many pages they view? How many uniques just view a site in general?
I’d argue that we need a formula of some type that factors in multiple things to truly rate sites. Perhaps a combination of time spent, page views, and unique visitors could somehow get us the most accurate picture. I don’t pretend to have the answer, but I know so far nobody else does either.
July 11th, 2007
People have been complaining as of late that all the innovative companies are getting snapped up by the big internet players which ends up killing innovation. I’m not sure whether or not that argument has any truth to it, but one thing that can be said that’s good when innovative companies get acquired, is that sometimes their services become free. Feedburner’s TotalStats package is now free after they were acquired by Google.
July 3rd, 2007
Over the past year or so the frustration with web statistics tracking services along the lines of Alexa and Comscore has grown. As companies look for validation, VCs look to measure potential investments, and bloggers find interesting things to talk about, accurate stat services are something everyone wants.
Yet people are frustrated with the current offerings. It has lead to the launch (or at least growth in prominence) of Compete.com and Quantcast. I like seeing more players in the field, and I especially like how Quantcast actually allows publishers to put a tracking tag on your site to give them accurate data. Now, the problem is that although my data is accurate since I choose to send data to Quantcast, it’s only a small minority of sites out there participating. When you put the effort on the publisher’s side without providing them obvious benefits, it’s hard to expect much of an uptake.
We know the “panel approach” or “sampling” of web visitors has its flaws. We know the Alexa toolbar approach has its flaws. We know the Quantcast tag approach will struggle to get enough publishers using it. So how do we get reliable data?
One thought I’ve had is if there was some sort of Firefox/browser plugin that people could use that reported your browsing data to a central service that then made rankings based on that. The problem is it’d be skewed towards the Firefox audience, and what is in it for the average web user to want that plugin?
So what if Firefox just did this by default? Maybe it was a privacy option they allowed you to turn off. Of course it would be anonymous, but Firefox itself could become a ranking service based on data it collects from users. Perhaps Microsoft could do this with IE, but I think people would be less trusting of their use of data compared to Firefox.
I’m sure there’s reasons the browsers can’t/won’t do this I’m not really thinking of, but what’s stopping Firefox from also ranking sites and providing us the closest thing we could get to accurate traffic statistics?
May 13th, 2007
The Google Analytics Blog has announced the first major upgrade to Google Analytics since they initially launched it after acquiring Urchin.
It was announced at the Emetrics Summit today, and it was interesting to note that Jeff Veen of MeasureMap (a Google acquisition) was involved in the announcement. I’d been wondering for a while what the status of MeasureMap was with Google, and if Veen and team were working on it or Google Analytics. While the answer still may not be clear. it’s obvious from this screenshot that the MeasureMap team had a hand in the redesign as there are definitely some similarities to the MeasureMap interface.
They’ll be rolling over accounts over the next few weeks to the new interface, and more can be found in their Google Analytics FAQ.
May 8th, 2007
Matt has announced that Wordpress has released a stats plugin for self-hosted Wordpress uses that integrates your blog analytics into your dashboard.
I’m a huge fan of integration of analytics and making it easier for publishers to do as much as possible in one place. Since everyone who uses Wordpress has to login to the dashboard to post, it’s going to be easy to check your stats quickly as well. I can save time to not have to go login an analytics application to get some quick blog stats, which means I have more time to create real value.
What does this mean for analytics companies? Probably not a ton, although it could hurt the blog-focused analytics packages a bit since Wordpress integration will be easier and quicker for people to use, although I’m guessing less powerful.
I’m going to install it shortly and will give my thoughts on the stats it offers.
UPDATE:
It won’t accept my Wordpress.com API key. I’m using the same one I’m using for Akismet, and the Wordpress.com Stats page that asks for it is suggesting I use that one and it won’t accept it.
UPDATE:
Andy Skelton who worked on creating the plugin quickly emailed me and offered to help me out (great service!) and then eventually concluded that I need to upgrade Wordpress. Doing so now…
May 6th, 2007
Previous Posts
