Hello Badsender ! I made a mistake... What should I do?

The one who has never made a mistake(s) when programming a campaign throws the first stone at me (not too loud please!)... It is often said that mistakes are human and I know what I am talking about, during my very first e-mail campaign (in 2008 anyway), I targeted the entire database of a client! In my defense, the "deduplication" box was not activated by default - there were 99% of duplicates - I let you imagine the result... some email addresses received more than 100x the same email... 🙁

Anyway, I'm not writing this article in order to make my redemption (though - sorry again Christophe D.)! A few months ago, one of our clients contacted me in a panic to alert us that they had made a targeting error for one of their campaigns. The campaign went to the entire base, almost 3 million email addresses.

As I often say, a good reputation can be lost in less than 5 minutes, but it will take several weeks or even months to regain it ! Therefore, it is always better to be sure about 100% before clicking on the button "SEND"...

The idea of this article is to list what we did to manage this crisis situation and how we managed to get him out of the water! As life is not really a long quiet river!

Measure the impact on reputation!

When our client alerted me to his deliverability problem, unfortunately the campaign was already over. If he had realized it before, he could have stopped it during the sending process and thus limited the impact on his reputation!

As the damage was already done, the first action to take was to measure the impact of the campaign on the current reputation. It is important to know that our client has an excellent reputation on all ISPs/Webmails, the only sensitive point that made me tremble a little bit was the IP strategy. Its IPs are shared between all its brands, which means that if the IPs are blocked or delivered in spam, all the brands will suffer the impact (yes I know, it's not cool but it's the risk of the mutualization!). Positive point, each brand has its own sender domain even if the technical domain (return-path) is mutualized :'

Come on, no more talk, let's get to work:

' Audit of internal indicators

What I call "internal indicators" are the campaign statistics dashboards provided by the router. If you have a good router, you will quickly see if you have an impact on deliverability thanks to the deliverability rate vs. bounce rate, the number of complaints generated and by calculating the dissatisfaction rate !

For our client, the statistics spoke for themselves... And they weren't very encouraging! A little more than a quarter of the e-mails were not delivered...

  • Deliverability rate : 73.1%
  • Bounce rate : 26.9% (more than 790 000 bounces generated)
  • Total unsubscribes : 14 319
  • Total complaints * : 7 581
  • Total Signal-Spam complaints ** : 10 050
  • Dissatisfaction rate *** : 81%

* Complaints reported via standard feedback loops: Outlook, AOL/Yahoo, La Poste.
** Non-name complaints from Orange, SFR, La Poste, Signal-Spam provided by the router over 2 days, 77% of the complaints come from Orange.
*** Number of unique clicks on the unsubscribe link vs. total unique clicks.

Small reminder on the rate of deliverability : https://www.badsender.com/2020/11/10/quest-ce-qui-se-cache-derriere-votre-taux-de-delivrabilite/

A little reminder about the dissatisfaction rate : https://www.badsender.com/2017/07/25/delivrabilite-surveiller-taux-insatisfaction/

Now that we know there was an impact, again if you have a good router that gives you access to bounces by destination, you will know the ISPs/Webmails impacted since they will generate "Block" type bounces.

For our client, there were many "Block" bounces, as well as other soft bounces (which are also blocks)...

  • Other soft bounces : 296 577
  • Block bounces : 286 851
  • Hard bounces : 166 558
  • User bounces : 33 415
  • Technical bounces : 10 909

By looking more precisely at the destinations of the bounces, I was able to list the ISPs/Webmails having generated a block, i.e. Microsoft, AOL/Yahoo, Orange, Free, La Poste, SFR...

Small reminder on the stats by destination : https://www.badsender.com/2020/11/25/marketeuses-marketeurs-veillez-toujours-a-analyser-vos-statistiques-de-campagnes-par-destination/

' Verification of external indicators

After analyzing the internal indicators, I moved on to the external indicators. These are simply all the monitoring tools made available by ISPs/Webmails.

Here is a list of the tools I use to monitor the reputation of an advertiser:

  • Postmaster Tools by Gmail allows you to check your domain / ip reputation at Gmail (requires a setting on the domain name)
  • Microsoft SNDS IP reputation : allows you to check your IP reputation at Outlook (requires validation from the router or IP owner in some cases)
  • Free Postmaster This tool allows you to check if your IP is blocked on Free's servers! And as a bonus, you will even know the reason for the blocking.
  • Anti-Spam Filter Lookup Check if your IP (or your domain) is listed on an Anti-Spam filter. Some companies offer this possibility: Cisco Talos, Cyren, Proofpoint, Sophos, Symantec, Trend Micro, ...
  • Third party reputation score : Senderscore allows you to have an external reputation score from data compiled from ISPs, Webmails, Anti-Spam filters, ...

In the end, the sender domain and all the IPs lost reputation at Gmail, only one of the IPs shows up as blocked in the Outlook tool (phew), the rest are clear!

' Conducting seedlist tests

To complete this measurement, it is essential to go through a seedlist test. This test will allow you to measure your reputation (inbox, spam, missing) at a given moment on different international webmails, hosts or Anti-Spam filters.

Need help?

Reading content isn't everything. The best way is to talk to us.


At Badsender, we use the seedlist ofEverestIt allows us to reach more than 800 addresses on US, French, American (North & South), Asian, European webmails, hosting companies or major Anti-Spam filters (if you want the complete list, PM me :p).

Test 1 was done as a famous bike racer would say "unwittingly"! And yes, since the client did a full base, he targeted Everest addresses 🙂

Test #1TrendInboxSpamMissing
AOLSPAM5%90%5%
GmailSPAM0%100%0%
HotmailINBOX90%5%5%
Yahoo USSPAM10%85%5%
FreeINBOX71%0%29%
La PosteINBOX100%0%0%
OrangeINBOX57%0%43%
SFRSPAM44%56%0%
Results of test 1 on the main webmails of our client.

Test 2 corresponds to the test that our client did following my request, 15 days after the impact.

Test #2TrendInboxSpamMissing
AOLINBOX100%0%0%
GmailSPAM0%100%0%
HotmailINBOX85%0%15%
Yahoo USINBOX100%0%0%
FreeINBOX100%0%0%
La PosteINBOX100%0%0%
OrangeMISSING0%0%100%
SFRINBOX100%0%0%
Results of test 2 still on the main webmails of our client.

Even though there was a 15 day interval between the 2 tests, three trends emerged:

  • AOL, Yahoo, Free and SFR have all regained their reputation.
  • Loss of reputation at Gmail for all IPs and the sending domain.
  • Blocking of one of the IPs at Microsoft, blocking of the sending domain by Orange (a test with their other domain was done on my Orange address and no blocking had been detected on the IPs, Orange will confirm us later the blocking of the domain).

Alert whoever is entitled!

This action is interesting and can potentially save your life or at least part of your life reputation! Orange is a very important ISP of our customer and very touchy, my priority was to warn Orange as soon as possible that there were going to be mass complaints... I could also have alerted Free, La Poste, SFR or even Vade but I didn't see for the moment the interest of contacting them without knowing the real impact (but I kept this idea anyway just in case).

As for the US webmails, I didn't do anything special, it's not like they were open to discussion. I preferred to define an action plan with the client after the reputation analysis, even if it means communicating this plan when I contact Microsoft!

Take corrective action!

Once we have all the analysis, list of impacted ISPs/Webmails, it is time to define an action plan to regain reputation!

3 big actions have been implemented:

  • Implementation of a double verification / validation of targets This will avoid errors (some routers offer this type of system).
  • Reduced targeting at Gmail Targeting assets (clickers, trackable openers - I like this name from Sendinblue) will boost the reputation quickly because the people we target will multiply the positive behaviors (openings, clicks, switching from spam to inbox, ...) and that the US webmails ... THEY KIFFENT.
  • Remove blocked domains As long as a blocking is not lifted, we might as well discard the impacted domains. Unfortunately, we could not remove the blocked IP at Microsoft (router limitation on IP management by ISP/Webmails) but the Orange domains were temporarily put aside.

Make a commitment to get a release!

Now that the corrective measures have been defined, it is time to contact the appropriate person to unblock the situation!

I prefer to remind you that you should not contact an ISP/Webmail if you have not done a minimum of investigation and/or if you have not set up an action plan to avoid the same problem ! ISPs/Webmails can be accommodating but you should not abuse their time and kindness!

In the context of the problems encountered by our client, I have only contacted a US Webmail (Microsoft) and a French ISP (Orange):

' Outlook Support!

Microsoft provides a contact form where you are asked for various information. Once validated, you will receive an "automatic" answer (which will tell you that NO, we don't unlock you)... No panic! You answer this e-mail by introducing yourself, describing the problem and the measures taken by the customer... A few hours later, the Microsoft support (mainly based in India) answers you and if you are lucky, grants you what they call a "mitigation" (We have implemented mitigation for your IP and this process may take 24 - 48 hours to replicate completely throughout our system), basically we unblock the IP but it remains under surveillance.

Unfortunately, for our client, I got a refusal from Microsoft (which breaks my 4 unblockings in a row!) because this IP had already been blocked a few months before. Despite my excellent explanations, they didn't want to hear anything! I proposed two options to our client: either put the IP aside for all the ISPs/Webmails, or keep it and generate some blockings! Option 2 was chosen by the client.

Spoil I tried a new unlocking request 1 month later, this one was validated by Microsoft ::applause::

Small reminder about Microsoft SNDS : https://www.badsender.com/2021/01/20/configurer-outil-snds-pour-monitorer-votre-reputation-chez-microsoft/

' The Orange Abuse Desk!

When you are blocked at Orange, you will remain so until you contact them! And here again, it is better to come with all the necessary information (date sent, sender, IP, error code, ...). The response time is variable but generally remains one day.

Unfortunately, during the first contact, Orange made it clear that they did not want to unblock the domain. After giving more information and asking the customer to commit to be careful in the future (we can call it a last chance) to not make this type of mistake again, Orange unblocked the domain for us! (thanks to them)

A last word for the end...

It was hard and long - almost like an obstacle course - but all our efforts paid off! After multiple analyses and exchanges, we were able to unblock our client's domain and improve the reputation of its IPs:

  • Microsoft IP unlocked after 47 days.
  • Gmail ' Reputation recovery of the domain after 2 months, reputation recovery of the IP after 30 days.
  • Orange ' Unlocking of the domain after 64 days.

If you also encounter difficulties, do not hesitate to copy / paste the different steps of this article or to contact us if you need help!

Share
The author

One Response

  1. Who hasn't made a mistake when programming a campaign? Badsender tells you what to do in case of reputation impact...

Laisser un commentaire

Your email address will not be published. Les champs obligatoires sont indiqués avec *