A few days ago I had a client who mentioned seeing something unusual in their Gmail testing. Their messages, which usually end up in the Promotions tab, had been hitting the Primary tab for almost two weeks. They reported clients had seen the same behavior. It’s not unheard of for this to happen at Gmail, but typically it lasts no more than a couple of days.
Turns out, lots of people have been asking the same question to the team at 250ok. Matt Vernhout confirmed via blog post yesterday that the deliverability tools provider is seeing this behavior as well.
There is no official acknowledgement from Google, although yesterday @Gmail replied to a Twitter user’s complaint with instructions on moving mail between tabs:
If you logged into your Google Postmaster Tools dashboard late yesterday or today, you may have been greeted with a troubling sight in the IP reputation graph:
From what we can tell, it appears to be an issue on Google’s end. I’ve checked dozens of IP reputations and all show nothing but red for December 10th data, and many in industry forums have reported the same.
Open rates at Gmail appear to be unaffected in my own data, and Laura at Word to the Wise provided some reassurance via Twitter:
Lots of comments this morning about GPT showing bad IP reputation for EVERYTHING this morning. Checked one of my clients on big shared pool and even those IPs dropped to ‘bad’ over night. Don’t Panic! This is a reporting issue, not a reputation one.— Laura Atkins (@wise_laura) December 12, 2019
No word from Google of course, but since this seems like just a display issue we can all go back to freaking out about other email metrics during this busy holiday season.
Overnight in the US, the AOL Mail and Yahoo Customer Care teams both tweetedupdates indicating services may be down.
No details on scope or ETA were provided, and in my own testing I was able to access my own AOL and Yahoo addresses. Both brands have indicated engineers are working on the issues and they will provide regular updates.
UPDATED 1pm EDT: around 10AM EDT the @AOLMail Twitter account indicated the issue was mostly resolved. My own Yahoo/AOL queues started to clear out around noon EDT, and unofficial word is that the issue is considered to be resolved at this point.
Last week we told you about an outage with Google’s Postmaster Tools data, which was missing data starting on 8/16 (many domains had data for 8/17). According to my own data and reports from others, data was populated for most domains late on 8/21 with data for 8/20 but nothing between 8/17-8/20. For the domains I track, it seems 8/21 data was added for some domains but not others, and no domain has anything past 8/21 at this point.
At the end of last week we assumed the issue was resolved, but it appears data is still missing in many instances. Spam Rate, IP & Domain Reputation, FBL, Authentication, Encryption, and Delivery Errors are all missing data for the same days. Word on the street hints at something new in GPT (but doesn’t it always?), but we’ll just have to keep refreshing that page until Google makes an announcement or something new appears.
Multiple reports have been coming in about Google’s Postmaster Tools data being missing for the past few days. Most reports indicate no data since 8/17, but some of my own domains were last updated on 8/16 even though they sent volume on the 17th.
Generally, GPT data runs a full 1-2 days behind, so the data for 8/17 was likely updated sometime the evening of 8/18. Even so, that leaves us on 3 days without updates. Not terribly uncommon, but many of us are still shook up after the 2-week-plus outage back in June. As usual, there’s been no official acknowledgement from Google.
GPT is a free tool, but it’s the only real insight (directly from Google) we as senders have into Google’s black box of mail filtering. I’m sure it’s not the highest priority project over there, but we sure do miss it when it’s not working properly.
UPDATE 8/22: Sometime last night EST Google appears to have populated data for 8/20 in my own instance and those of multiple others I’ve asked. Data is still missing for 8/18-8/19 (and 8/17 for a couple of my domains). Updates to the data generally happen for me late afternoon EST, so I’ll watch for any new data or backfilling later today.
If you’ve sent mail to Canadian recipients over the past few days, there’s a good chance your engagement rates have been lower than expected. Telus, a major Canadian telecom and mailbox provider, has experienced a multi-day outage of their email system starting last Thursday (August 15th).
Reports have been mixed about the exact impact and scope – Telus claimed on Friday that only 10% of customers were still impacted, then issued an apology video Saturday night in the midst of continued problems. The Telus website carries a banner linking to an Update page. Among the information here, Telus directs users to a temporary webmail account that has been created for them and states that the “old” mailboxes are being rebuilt. No timeline is given for this process.
Will Telus be able to recover all the emails lost since last Thursday? Time will tell, but it’s a safe bet at least some mail will be lost in the process. As a sender, you should evaluate the volume of traffic sent to Telus since 8/15 and whether to halt sending to those addresses until full resolution is received. At a minimum, you’ll have a good explanation when management asks why your engagement rates took a dive.
We’ve talked before about domain and IP reputation, and how mailbox providers use each to determine whether the mail you send reaches the inbox or languishes in the depths of the spam folder. Since so much of your sender reputation is tied to your IP and domain, many senders have the idea that the solution to delivery issues is to simply fire up a new IP, domain, or both, granting yourself the ultimate “do-over” with the mailbox providers. There’s just one problem with that theory…
It’s wrong. Last week, 250ok’s Beth Kittle published a great piece on the fallacy of this “email myth.” (If you haven’t yet, go read her post and commit it to memory.) Beth’s illustration of the incognito party guest is a great analogy for how providers spot your mail even when sent from new infrastructure and how your misdeeds will find you there if you continue to use bad practices.
But what happens when you do see a benefit from switching? Over the past few years I’ve spoken with dozens of marketers who had an example of a situation where their deliverability improved when switching to a new IP or domain. They performed a few tests on the new infrastructure and – lo and behold – they got to the inbox! Because of this, these clients often didn’t believe my warnings that new IPs or domains wouldn’t fix the problem – they were more concerned with temporary gains (and quite a few just assumed they’d talk their way into another IP once the new one started to tank). Don’t miss this point: your deliverability may improve when you switch to a new IP or domain – but only temporarily.
I’ll provide an analogy of my own, gleaned from a former coworker who compared IP and domain reputation to telephone caller ID. If you get unwanted calls from a telemarketer, you learn to stop answering when their number shows up on the caller ID. After days of unanswered calls, if the telemarketer dials you from a new number you just might pick up to find out who’s on the other end. As soon as you hear the voice of the unwanted caller on the other end, you’ll hang up and make sure you don’t answer calls from the new number as well.
This is how mailbox providers and spam filters operate – while it’s possible your first mailing might see better performance, as soon as they identify you’re the same sender they blocked before you’ll see those numbers fall off right away. As Beth notes, their algorithms are very advanced and they will quickly find out who you are. If you want to improve your deliverability, there are no shortcuts – you’ll have to put in the work to improve your practices.
Reports have been trickling in today that senders are seeing spikes in “Bad” reputation in Google’s IP Reputation Postmaster page. In checking my own instance, I found at least 3 major senders whose IPs have been 100% Good for months, then 75% or more in the Bad status for July 3rd.
The results appear to be spotty, but multiple industry folks have mentioned seeing similar behavior in popular forums today. No official word from Google yet, but we’ll keep an eye on this and update as news comes in.
UPDATED 1:45PM EST – After further analysis, deliverability to Gmail does not seem to be affected for clients I’ve checked, and a few others have reported the same. Domain reputation also appears to be unaffected for me, but some others have reported changes in domain reputation that may or may not be related. In any case, these blips certainly lend credence to the loudly-whispered theory that major US holidays wreak havoc on GPT data, with yesterday being Independence Day here in the States.
As of now, data seems to be trickling into my own Google Postmaster Tools display, with some monitored domains showing current data (i.e. 2-3 days ago, as is standard in GPT).
The data is still spotty, though: some domain reputation listings show data points for each day while others show nothing between the 13th and now. IP reputation appears to still be missing for the outage period for the domains I’ve checked that are showing anything since June 13th.
Obviously the issue is not entirely resolved, but it’s nice to see something showing up after 2 weeks of dead silence.
There’s no surer way to have things break than to take a vacation, and it seems that adage again rings true in the deliverability world. During my own time away from the office, it seems Google Postmaster Tools has stopped reporting data for all categories.
In my own experience – and the experience of many others in industry forums – no data has been provided for any tab since June 13th. As usual, there is no official statement from Google but word is that a Google rep has acknowledged they are working on the issue.
While it’s not uncommon for GPT to miss a day or two of data here and there, an outage of this length is fairly rare. The last outage of this magnitude was (if I recall correctly) in April of 2018.
There has been some discussion around how these outages often correlate with major US holidays (Easter, Memorial Day, Father’s Day, and Mother’s Day have all been hit) but there seems to be no confirmation of why. Increased volume causing delays in processing? Resources allocated to higher priority areas? The API upgrade? Whatever the cause, it’s likely we’ll never know for sure.