Statistics Difference Between MailWizz & ESPs Dashboard

Osjtya

Active Member
Hello All,

Hope all fine! From the last few weeks I observed an error/bug in MailWizz, where there is a difference (sometimes huge) between statistics (Opens, Clicks, etc.) of MailWizz & ESPs (Email Service Providers) dashboard while using SMTP connection. What do you guys think, why there could be any difference? Any valid reason?

For some reason, can't use API in this case. So have to find solutions in SMTP protocol to transmit emails to ESPs.

Hope we all together can find some valid reasons here ;)
 

twisted1919

Administrator
Staff member
From the last few weeks I observed an error/bug in MailWizz,
MailWizz tracks all actions, and it does so correctly.
If you get smaller number in MailWizz than at the ESP level, then you should think that your server couldn't handle all the clicks and opens that happened after your newsletter was sent, thus it missed some of the opens/clicks while your ESP didn't, because they're better equipped to handle these situations. That's just an idea.
 

Osjtya

Active Member
@twisted1919 I don't think it's related to the server where MailWizz is hosted. Because this particular MailWizz instance is only sending few thousand emails from a server that is equipped with following hardware 32GB+ RAM, Multi CPU, 500+ GB SSD, 100% optimised to send over million emails per hour and only hosting MailWizz not even MTA is on that server.

So something else in the tracking system of MailWizz which is not matching with ESP's technology. Can it happen because of some proxy?

This is sort of confusing me :eek:
 

twisted1919

Administrator
Staff member
It's very simple to test this. You send a campaign to yourself, and open it several times and click the links several times.
Do you get each action tracked?
 

Osjtya

Active Member
@twisted1919 if in case we decide to use SendGrid API rather than SMTP connection, then still MailWizz uses own tracking data or it gets fetched from SendGrid?

Also see the screenshots for what I'm talking about, it's huge difference in tracking technologies.
 

Attachments

  • 2020-11-11_05-20-55.png
    2020-11-11_05-20-55.png
    148.4 KB · Views: 18
  • 2020-11-11_05-27-10.png
    2020-11-11_05-27-10.png
    1.1 MB · Views: 18
Last edited:

twisted1919

Administrator
Staff member
then still MailWizz uses own tracking
We always use our tracking data, since as i said, it is very accurate.

The fact you enable tracking from inside sendgrid might actually be an issue, if they somehow rewrite our tracking pixel, then that would be a problem, so i'd say you should try to disable the tracking from sg and see if that makes any difference in mailwizz.
 

Osjtya

Active Member
@twisted1919 thanks for the suggestion. So if I've understood correctly, it doesn't matter if I use SMTP Connection or API Connection with SendGrid, MailWizz will use it's own tracking system to collect open, clicks, etc.
 

nadworks

Active Member
It's very simple to test this. You send a campaign to yourself, and open it several times and click the links several times.
Do you get each action tracked?
That was a great idea - I just did that test and unfortunately, the result shows that something is definitely wrong with MW tracking.
I created a campaign and only sent it to my own address.
I am based in the UK (not using VPN or anything strange).
I opened the email on my Gmail app on iOS.
Here are the stats I got from that one single open:

screencapture-e-nad-works-customer-index-php-campaigns-ao335t7qhq250-overview-2021-02-11-10_48...png

I have no idea how to read that. But it's not what happened. It shows as if I opened the email
  • 2x
  • on a Windows desktop application
  • using Firefox
  • and that I'm based in the USA.
Couldn't be more wrong. Sorry.
 
That was a great idea - I just did that test and unfortunately, the result shows that something is definitely wrong with MW tracking.
I created a campaign and only sent it to my own address.
I am based in the UK (not using VPN or anything strange).
I opened the email on my Gmail app on iOS.
Here are the stats I got from that one single open:

View attachment 12548

I have no idea how to read that. But it's not what happened. It shows as if I opened the email
  • 2x
  • on a Windows desktop application
  • using Firefox
  • and that I'm based in the USA.
Couldn't be more wrong. Sorry.

If your email is gmail or yahoo.

They both most of times uses proxy to open your email.

Read more about gmail proxy for example at https://sendloop.com/articles/the-effect-of-gmail-image-proxy-to-email-marketers/

Thanks
 

nadworks

Active Member
If your email is gmail or yahoo.

They both most of times uses proxy to open your email.

Read more about gmail proxy for example at https://sendloop.com/articles/the-effect-of-gmail-image-proxy-to-email-marketers/

Thanks
Exactly. I'm completely aware of that, and that's exactly my point. It has been the case since 2014 as far as I remember. So nothing new. However other ESPs have overcome this issue.

It looks like it's a matter of excluding the false opens coming from the Google proxy server. This is what other ESPs have started doing:
 
Last edited:

twisted1919

Administrator
Staff member
We have two ways to exclude this:
1. Based on IP range, basically exclude any IP coming from google.
2. Based on user agent. Simply exclude anything that is a bot.
Suggestions?
 

twisted1919

Administrator
Staff member
@Andy Whitlow - I tend to like #1 as well, problem is i'd have to just find the ip ranges for google/yahoo/comcast/version/etc, and this looks like it's just too much to maintain in the long run.
I started with the user agent for now by adding a library to match the user agents against known crawlers. Let's see how this goes, if the results are not optimal, we'll use ip ranges.
 
@Andy Whitlow - I tend to like #1 as well, problem is i'd have to just find the ip ranges for google/yahoo/comcast/version/etc, and this looks like it's just too much to maintain in the long run.
I started with the user agent for now by adding a library to match the user agents against known crawlers. Let's see how this goes, if the results are not optimal, we'll use ip ranges.
Got that.

Let's see how it rolls
 

twisted1919

Administrator
Staff member
Just an update here, after a bunch of tests, we decided to also add the IP Ranges in the mix.
For now, we start with the IPs from google, we fetch them once at 24 hours to make sure we always have the latest ones and we don't block IPs that we shouldn't.
This check is enabled by default (maybe we should keep it disabled?) but you can disable (or enable) it from Backend > Settings > Campaigns > Exclude IPs from tracking:
Screenshot 2021-02-15 at 10.38.35.png
And the Ip ranges:
Screenshot 2021-02-15 at 10.39.07.png
So you know what exactly you will block.
 
Top