Devices statistics

[GildasPro]

Is there a way to know the statistics about the device?

I would like to know my newsletter opening percentage via a smartphone.

 

[twisted1919]

Not for now unfortunately...

 

[nadworks]

So, although this is now available, it does not seem to be working correctly.

I know my audience quite well from GA stats and UTM captured campaign stats, favouring 63%+ mobile devices. However, the device and operating system stats I see here in MW shows 95% Windows / 97% Desktop / 80% Firefox Browser. That's just completely impossible. Yet, it's consistent through all campaigns. It can't be me, because I don't use Firefox.

Am I missing something? Does anyone else have that same problem?

 

[twisted1919]

@nadworks - we extract that info from the user agent string we get from them while opening and clicking campaigns.
It also contains info from proxies which open the emails in behalf of users, or spam filters which do that, so this might be one reason for why the data is off. Remember that you can add ip addresses and ip ranges to block opens and clicks tracking from certain sources. This will give you more accurate data.

 

[nadworks]

This seems to be the Google image proxy issue:
This is my user agent of 94% of opens:
Mozilla/5.0 (Windows NT 5.1; rv:11.0) Gecko Firefox/11.0 (via ggpht.com GoogleImageProxy)

Any way the device stats can be outmanoeuvring Google's cached image? This issue has been around for a while and is not affecting stats of other email platforms as far as I can see. So they must be capturing this insight differently.

Here's more about the known issue:
https://feedback.mailgun.com/forums...88995-fix-googleimageproxy-user-agent-parsing
https://www.gmass.co/blog/false-opens-in-gmail/
https://instiller.co.uk/blog/gmail-image-caching/

 

[nadworks]

Hi @twisted1919 and team - do you have any plans to improve the accuracy of device and OS statistics, which are so very clearly still screwed up?
I'm increasingly asked by clients about their subscribers' behaviour and preference in that area, and it would help me immensely making a case for certain design and build decisions if I was equipped with actual audience's statistics.
I feel the current method is somehow dated and not taking ESP specific methods into account. I.e. the first gmail user's open action will serve all other gmail users that very same cached image. In a nutshell, that first useragent data is subsequently used for ALL other users on gmail, regardless how they open their emails.

 

[nadworks]

Just following up again since this continues to be a challenge when analysing my clients' campaigns.

The device tracking is completely skewed - I assume - by proxy image loading of the main email clients (i.e. Gmail/Google, MS Outlook?). It doesn't seem to overcome the tracking requirements to get real stats - other ESPs have sussed this out long ago.

I'm bringing clients over from MailChimp to Mailwizz. Their MC device stats look typical and make sense. Once sending through MW, 90% of users are suddenly opening on Windows Desktop using Firefox. I mean, come-on. Look at the below, which is what I see for almost all of my various clients' campaigns...




No way is this a reflection of reality. All my campaigns look similar to this.
I don't want to beg to address this. But I guess I kind-of am...

 

[twisted1919]

Take a look in the other thread related to this where we discuss the options we have to filter opens/clicks.
Right now we track those requests as they come, we don't filter them in any way, so if they come from image proxies, we track them just fine. Other platforms filter out requests coming from robots, so i assume we should do it as well.

 

[nadworks]

I remember looking at this when it was first launched a vaguely remember that we managed to exclude the tracking image from being cached (via code). But I'm not able to trace back that info. It wasn't such a massive deal in 2013 as Gmail was still way behind Hotmail then. This has obviously changed since.

Whatever is doable would be appreciated. The way it is right now is making me look bad in front of my clients, especially those coming from MailChimp et al.

Info from out there:

What You Need To Know About Image Caching In Gmail - Litmus

Google has introduced a number of features to Gmail lately. Along with tabs in the new inbox and an updated mobile client, you can add image caching to the list of changes that affect email marketers. Image caching and how it affects email tracking can be a confusing topic. We take a look at...

www.litmus.com

Mailgun Test With Gmail Open Tracking | Mailgun

Gmail is now caching images on their own servers which changes the data given back to email service providers and marketers. Read more –

www.mailgun.com

Google Image Proxy for Gmail - LeadBoxer Knowledge Base

Introduction Starting in 2013, Google introduced a feature for its gmail customers to preload and proxy-load the images that are present in (html) emails. This

docs.leadboxer.com

Google’s New Image Caching: 5 Things You Need to Know - SendGrid

A look at Google's new image caching and how it will affect marketing statistics like multiple opens, geo stats, IP address, browser, and device type.

sendgrid.com

Why Gmail Image Caching Doesn’t Spell Disaster for Email Marketing - FunnelEnvy

Gmail image caching doesn't spell disaster for email marketers. Here's what you need to know.

www.funnelenvy.com

"The trick is actually quite simple: in your email header, your content length coding must be set to zero (Content-length’ (=0). Emails coded this way notify your server each time it’s opened. For a detailed exploration of this approach and a tutorial on changing your tracking pixels, I recommend visiting this helpful post from The Email Expert. It’s important to note that the jury is out on this approach. Some people feel that it violates the spirit of Google’s changes, and expect them to correct for this “loophole.” Others disagree. Only time will tell, but for now if this metric matters to you there is a way to achieve it."

 

[twisted1919]

The trick is actually quite simple: in your email header, your content length coding must be set to zero (Content-length’ (=0). Emails coded this way notify your server each time it’s opened.

This is what we do and obvious it's not the case anymore.

 

[nadworks]

This is what we do and obvious it's not the case anymore.

Yes, it might have worked at the beginning.
I'm a member of the DMA Email Council here in the UK and will reach out to some of our tech/ESP members.

 

[twisted1919]

Just a FYI, i just added code in the app where it will not do tracking for crawlers/robots, so let's see how that will work.

 

[nadworks]

Thanks for the work on this, @twisted1919. Just upgraded to 1.9.24.
The device stats look much better now at first glance.

 

[twisted1919]

@nadworks - unfortunately, we had to revert these changes and released 1.9.25, as they had some side effects, see
https://forum.mailwizz.com/threads/tracking-issues-in-1-9-24-upgrade-to-1-9-25.7402/
We will now take the right time to implement this properly. Sorry for any inconvenience it has caused.

 

[nadworks]

Ah, that's a shame, but your explanation makes total sense. Let's hope you can find a solution that's covering all these holes.