Hi all,
A bit of a technical issue here, for a few weeks now, our team has been seeing drastic differences in the page view reports provided by Marketo and GA.
Marketo Support has not been able to find a a solution (yet), so hopefully there are some smart minds here who can help figure out the problem.
This is our situation:
With Google Tag Manager (GTM) we execute the Munchkin code and the Google Universal Analytics code, on all pages.
However, in GTM we made an exception rule, so that the Munchkin code is not fired on any Marketo landing page. (see next 2 screenshots)
The GTM script is placed in the body of the landing page template. The exception rule should ensure that the munchkin code is not fired twice.
The Munchkin code is also placed separately in the body of the HTML in the landing page template.
Our problem:
We have landing pages that supposedly have a lot of views according to Marketo, but in GA, these numbers are much lower. One landing page, which is reporting contradicting values is this one: Use E-Commerce to Keep Your B2B Company in Top Shape - Sana Commerce (it's in Dutch by the way, if you may wonder)
It's an old landing page, and it is hardly used, so it is very odd that Marketo reports high page visits.
This is what Marketo tells us:
Note the time of activity. Starting at 0:56 o'clock on August 8th, and ending on August 14th at 23:55 o'clock. We see this happening for this page for the past month already.
And this is what Google tells us:
So 187 compared to 6 ... That's a huge difference.
In GA, we exclude certain IP addresses which are not relevant to track. However, this could not explain the big differences we see in page views.
Would anyone be able to tell us if we are doing something wrong, or guide us to a place to look for a possible solution?
Appreciate all the help!
GA uses only JS-based tracking.
IIRC, Marketo LPs count pure HTTP GETs (not just the Munchkin JS tracking layer). Hence any bot traffic that isn't detected and discarded will count as an impression.
Hi Sanford,
My colleague who's responsible for Google Analytics tells me that for this specific page in this time frame, the unfiltered GA report shows 14 page views.
That is still drastically lower than what Marketo tells us.
Would you have any further explanation?
As I'm not too experienced with JS and Munchkin terminology (and the exact way how it all works), can you break it down in comprehensible steps?
Thanks for your help!
What do you mean by "unfiltered"? If you use only JS-based tracking, it won't show any non-JS impressions (which is quite deliberate, the idea being that only a negligible number of people turn off JS, and those that do turn off JS probably turn off all communications to known tracking services, so you would miss them anyway).
To put it visually, if my LP looks like this:
and then I open it in a non-JS browser (emulating a bot) it'll look like this:
GA will not log this hit.
I might not even be able to use the page, since parts of it may require JS to even display or register clicks. It might be ugly as sin if JS is used to lazy-load CSS styles. But it's still a view in Marketo.
The same distinction is there if you run your own web server and look at the raw server logs. They'll show every single attempt to view the page, even those that didn't result in anything a human could read. In contrast, JS-based logging like Munchkin and GA will only run if the client viewing the page is capable of running JS scripts (that client could well be a bot with higher-end capability, but smarter JS-capable bots will specifically exclude tracking scripts even if they could run them).
Can what you're saying explain a difference of 187 views in Marketo compared to only 6 in GA?
Sure, it could. Have to see more of the distribution of User-Agents to draw an absolute conclusion.
A couple points:
The Munchkin code is also placed separately in the body of the HTML in the landing page template.
This isn't necessary. Marketo will automatically add Munchkin to all Marketo LPs.
Marketo LPs count pure HTTP GETs (not just the Munchkin JS tracking layer)
If that's the case, then this makes Marketo's reporting of page views completely unrealistic and shouldn't be used. Can someone from the Marketo product team comment here?
Bear in mind that simple Marketo LPs don't require JS: that's why forms on native LPs have a NOSCRIPT version burned into the page.
Whether it's necessary to provide such compatibility -- especially given that without JS, leads will never follow a tracked link -- is a good question. But as long as the NoScript compatibility is purposeful, you have to track NoScript hits.
Hello All,
Just ran into this exact issue and didnt realise the magnitude of the problem.
If this is the case do you disregard the landing page views and therefore your conversions rates which are actually much higher than reported?
thanks
Mark
As for me, I don't consider the Filled Out Form/View ratio to be useful. Views are still useful if you see zero or only a handful (i.e. indicating there was a typo when you linked to the page).