Secrets about Google Analytics funnels(blog.hublo.co)
blog.hublo.co
Secrets about Google Analytics funnels
http://blog.hublo.co/2014/02/27/5-dirty-little-secrets-about-google-analytics-funnels/
18 comments
I help companies set up their GA tracking[0], so let me offer some tips and responses.
> 1. No events in your funnels.
The simple solution is to use virtual page-views, as mentioned in the article.
If you're using Google Tag Manager (and you should), you do not have to re-code your website tracking. That's the point of GTM: You can make changes to tracking without touching your code.
It's unlikely that this virtual pageview will make up a significant portion of your total pageviews. Your pageviews will not skyrocket. Besides, pageviews is a vanity metric, and you shouldn't be using it to measure your site's effectiveness.
> 2. Funnels are backfilled.
I do wish there was an option to turn this off, but the very simple solution is to look at the Goal Flow instead of the Funnel Visualization.
> 3. Funnels can’t be segmented.
Again, look at the Goal Flow report. You can segment by source, landing page, custom dimensions, and a hundred other things...
> 4. The rule applied to the goal URL applies to all other steps’ URLs.
GA can use some more in-line help text, but this is an example of an incorrect setup, and not an issue with how GA tracks visitors.
> 5. The Goal Flow Visualization is a bit better, and a bit worse.
Ugh, so much FUD here. Yes, this does undermine points #2 and #3 very much, indeed. As for its readability, I never had anyone ask me how to read it. If you're blessed with a lot of traffic, it even has options to hide insignificant paths.
> ...And things start to get awry when you have to explain those discrepancies to your boss, coworkers or client.
That's correct. If you don't know what you're tracking, don't know how to read and interpret the reports, and don't know which metrics to report to your boss, then you're not going to have a happy day. But that has little to do with the tool you're using.
- - -
If you understand the tools you're using, they will be more helpful to you. A learning curve is not a reason to run from the tool.
[0] More info here: http://www.gkogan.co
> 1. No events in your funnels.
The simple solution is to use virtual page-views, as mentioned in the article.
If you're using Google Tag Manager (and you should), you do not have to re-code your website tracking. That's the point of GTM: You can make changes to tracking without touching your code.
It's unlikely that this virtual pageview will make up a significant portion of your total pageviews. Your pageviews will not skyrocket. Besides, pageviews is a vanity metric, and you shouldn't be using it to measure your site's effectiveness.
> 2. Funnels are backfilled.
I do wish there was an option to turn this off, but the very simple solution is to look at the Goal Flow instead of the Funnel Visualization.
> 3. Funnels can’t be segmented.
Again, look at the Goal Flow report. You can segment by source, landing page, custom dimensions, and a hundred other things...
> 4. The rule applied to the goal URL applies to all other steps’ URLs.
GA can use some more in-line help text, but this is an example of an incorrect setup, and not an issue with how GA tracks visitors.
> 5. The Goal Flow Visualization is a bit better, and a bit worse.
Ugh, so much FUD here. Yes, this does undermine points #2 and #3 very much, indeed. As for its readability, I never had anyone ask me how to read it. If you're blessed with a lot of traffic, it even has options to hide insignificant paths.
> ...And things start to get awry when you have to explain those discrepancies to your boss, coworkers or client.
That's correct. If you don't know what you're tracking, don't know how to read and interpret the reports, and don't know which metrics to report to your boss, then you're not going to have a happy day. But that has little to do with the tool you're using.
- - -
If you understand the tools you're using, they will be more helpful to you. A learning curve is not a reason to run from the tool.
[0] More info here: http://www.gkogan.co
FUD would have been to disregard Goal flows entirely. The fact that GA gives you the choice is part of the problem: there are two funnel visualizations, and it confuses most people.
If you believe that web-analytics should be understood and used by just a few, then it's OK, GA is great. But if the discipline wants to reach a broader audience, product choices should be made, but GA will not commit to such changes apparently.
If you believe that web-analytics should be understood and used by just a few, then it's OK, GA is great. But if the discipline wants to reach a broader audience, product choices should be made, but GA will not commit to such changes apparently.
Isn't this an inherent problem of any sophisticated tool? AutoCAD is a nightmare to learn for most people, but that's OK. The ones who just want a simple modeling program can use Google Sketchup. Similarly, anyone who just wants basic metrics (visits, sources) can use GA out of the box or use one of the simpler solutions (many of which have been named in this thread).
Don't get me wrong, there's a lot of room for improvement in GA, and even I continue to learn new things about it all the time. I just don't want to see it watered down to the lowest common denominator.
Don't get me wrong, there's a lot of room for improvement in GA, and even I continue to learn new things about it all the time. I just don't want to see it watered down to the lowest common denominator.
I know that a lot of people use PadiTrack to better understand their GA funnels and it provides segments as far as I know? Plus, it's free. http://paditrack.com
What do you guys use for funnel analysis? I think as a paid solution KISSMetrics covers some of the issues outlined in this post.
What do you guys use for funnel analysis? I think as a paid solution KISSMetrics covers some of the issues outlined in this post.
Well, they are a web analytics company, so I'd wager they use themselves. :)
>It’s a free tool
If by "free" he/she means giving Google the easiest way to track everyone's movement on WWW...
If by "free" he/she means giving Google the easiest way to track everyone's movement on WWW...
What you pay is what you get!
Well, Google Analytics is very generous. It's just that what you save by getting it for free, you have to invest in educating yourself.
I agree. It is hard to use, but if you keep up with it, it is better than Mixpanel in many ways. For instance, it doesn't load your users with massive cookies that slow the site down, particurarly on mobile browsers. The latest GA session cookie is very clean.
I have seen many sites that would be strong and useful in the SERPS, but are trashed because they think they need to collect analytics from 6 different services. I have brought most of my A/B testing server side, because the site loads so fast, the A/B flicker affects the A/B test results.
I have seen many sites that would be strong and useful in the SERPS, but are trashed because they think they need to collect analytics from 6 different services. I have brought most of my A/B testing server side, because the site loads so fast, the A/B flicker affects the A/B test results.
> It is hard to use, but if you keep up with it, it is better than Mixpanel in many ways.
This basically sums up this entire discussion. GA is powerful, and for that you pay the price of a slight learning curve (or paying someone like me to set it up quickly). Oh, and it's free.
This basically sums up this entire discussion. GA is powerful, and for that you pay the price of a slight learning curve (or paying someone like me to set it up quickly). Oh, and it's free.
I agree. This mess is exactly why people pay for KissMetrics, MixPanel and other analytics tools that do funnel reports really well.
This doesn't always hold. You don't pay for 99% of the services you use online (GMail, Facebook, HN, Reddit, etc).
Cool one, thanks.
Then you can create funnels based on dimensions and not just URLs.
Creating ad-hoc funnels that don't have the UI limitations and don't backfill are easy once you transition to the API.
But actually, don't transition to the API. Just start using their Big Query implementation, then you can write SQL against what is effectively marketing aware log analysis.
Or if you can afford it, just use Mix Panel...