How to Use Feedback Loops
For the uninitiated, FBL stands for feedback loop, the mechanism ISPs like Hotmail and Yahoo! use to report spam complaints to senders.
I know this will be totally shocking because, of course, your customers love you, but your messaging programs generate spam complaints if you send commercial email. And while there is a direct correlation between solid deliverability and a healthy response to complaints, I’m not going to drone on about it for another 500 words.
Instead, I’m going to show you how the newly released feedback loops module inside 250ok’s Reputation Informant helps you overcome the historical problems preventing senders from getting the most out of FBLs. It enables you to quickly understand complaint trends, pinpoint problem spots, and take immediate corrective action.
The key component of thinking about how you respond to complaints is time. Most feedback loops operate in near real-time, meaning while there can be a delay of days or weeks between the time you send the message and when it is reported as spam, the FBL notification arrives very soon after the recipient files the report.
The only safe way to interpret a spam complaint is to take it as an indication that the recipient wants to opt-
out of that mail stream. Therefore, your responsibility is to do so, immediately.
Repeatedly sending messages to recipients who have opted-out of those messages is the fastest way to tank your reputation. You can wind up in the spam folder, or worse, be fired by your email service provider for not playing by the rules.
Reputation Informant’s feedback loops module gives insight into several things. It shows your complaint reports over time, the current and historical rates at which you receive complaints, the largest sources by reporting domain, and top sources of complaints broken down by sending IP, subject line, and the sending and receiving addresses.
This visibility enables you to:
- Understand the campaigns generating the highest complaint rates.
- Identify the recipients hitting report as spam most often.
- Predict which of your IP addresses are likely to be affected and at which ISPs.
You’re also able to drill down and examine the fine details of each complaint event.
Most importantly, the feedback loop module makes it incredibly easy to process opt-out for complaints by allowing you to export a list of their email addresses. It’s essentially the magic-unicorn functionality you’ve been seeking in managing your complaints.
The module’s IP and date-based filters, along with hour-level granularity, allow you to examine complaint activity preceding a drop in deliverability at a specific ISP. You can also explore the correlation of your heaviest complaint times with your sending patterns.
Our goal at 250ok is to make every sender look great by empowering them with data and tools that simplify email. By partnering with SparkPost to bring the FBL module to every SparkPost user, we’ve made it easy to understand and effectively respond to spam complaints.
Okay, true. It was also an excellent excuse to create pretty graphs and play with a lot of data. But we swear the main motivation was building great tools for our joint customers. Ahem.
Paul Midgen believes that inside every ruthless revenue-driven sender lurks a recipient-centered Jedi master, and he’s dedicated to setting them free one sender at a time. Before joining 250ok as VP of Engineering, Paul was CEO of Message Bus, ran Inbound Delivery & Anti-Spam at Hotmail, and has spent a lifetime working on things that allow machines to talk to each other for the benefit of humans.
Two new feedback loops (FBLs) were released today by Return Path:
These new Feedback Loops will follow the same format as their existing hosted FBLs: IP-based, ARF reports. For those not familiar with ARF (Abuse Reporting Format) reports, this format obfuscates the address of the user who complained to protect their Personally Identifiable Information (PII). More detail can be found here.
It’s important to keep in mind that simply applying does not guarantee your subscription will be accepted. The goal of offering a feedback loop for these providers is to help responsible senders to actively manage their subscriber experience by removing users who complain about their mail and using the data to adjust their targeting, in order to reduce user complaints moving forward.
In addition to these two new FBLs, there is a new resource on the horizon for senders, which will be available via the M3AAWG website. The M3AAWG Feedback Loop Best Common Practices group, led by myself and Kate Nowrouzi (also of SparkPost), has developed the M3AAWG Feedback Loop Resource Page which will be available to the public through the M3AAWG website in the next couple weeks. The page will contain:
- A definition of the term “feedback loop” for those who are not familiar with them
- A description of the various reporting formats
- A list of all currently available feedback loops and links to their applications
- Additional feedback loop related resources
If you are a SparkPost customer, rest assured that you are subscribed to all available feedback loops. We also provide guidance on program improvement to keep your recipients happy and complaints low. Find out more here.
If you liked this post, you may also like:
- Yahoo! dropping Return Path Feedback Loop
- 4 Things You Need To Know About Handling Feedback Loops
- All You Need to Know About Gmail’s Feedback Loop Offering
Yahoo has just announced that they will be dropping Return Path from managing their Complaint Feedback Loop (CFL). On June 29, 2015 Yahoo will transition the Yahoo CFL administration to their own Yahoo Customer Care. Yahoo does plan to port over all configurations as-is, stating no actions should be required and they will continue sending spam reports during the transition. However, senders should save existing CFL information as it will not be available after the transition.
As a customer of SparkPost Elite or SparkPost.com, no further action will be necessary. The Deliverability Services team will manage all SparkPost Elite CFL snapshots. SparkPost.com is double-signing DKIM for their customers so that only a snapshot of the SparkPost domain is needed and therefore managed by the Compliance team.
As a Message Systems On-Prem Customer that manages your own FBLs, you will want to visit the Yahoo site to take a snapshot of your current CFL configuration prior to June 29, 2015 as it will not be available once the transition is complete.
To save the existing CFL information:
- Go to http://feedbackloop.yahoo.net
- Sign-in with the email address you used for registration
- Go to ‘Manage Existing CFLs’ section (3rd tab at the top)
- Select all the information for existing domains
- Copy and paste the information to a file for future reference
Gmail introduces a pilot feedback loop offering
Back in February, Gmail announced their Feedback Loop (FBL) pilot offering to ESPs to help them with identifying bad actors and spammers on their network. For anyone who is interested in learning more, the enrollment form can be found here.
For those of you who are unfamiliar with the process, here’s a brief recap. Feedback loops are essentially reports that ISPs provide to large volume senders about the number of recipients who mark their mails as spam. It’s a really important service that allows businesses to monitor their sender reputation with the ISPs and quickly take action for damage control if large numbers of recipients are marking their emails as spam. As you can already tell, this is pretty crucial for businesses that are dependent on email marketing as a main revenue stream.
Gmail’s feedback loop, however, differs from other ISPs
Gmail’s FBL is not in ARF format. In order to protect user privacy, their FBL is offered in the form of aggregated spam statistics per customer or per campaign, which cannot be traced back to the email address of the recipient who marked the mail as spam. This daily report provides a percentage of user spam complaint rate per customer and/or campaign of an ESP, and will be sent to the designated email address provided to Gmail by the ESP (eg: GmailFBL@example.com). The service is not designed for list management or delivery evaluation, and if there is a sizable percentage of spam in your traffic, you will receive the aggregated report the next day.
ESPs are highly encouraged to comply with Gmail Bulk Senders guidelines. Gmail requires all senders to use a consistent IP address to send mail, valid RDNS record for all sending IP addresses, and the same address in the ‘From:’ header on every bulk mail. As far as authentication, they highly recommend signing with DKIM, publishing an SPF record and adhering to the DMARC policy.
How does Gmail’s feedback loop work?
ESPs will first need to insert a feedback identifier header “Feedback-ID”. This header will identify the customer and or campaigns, mailings, and mail types. FBL reports will be generated based on these identifiers.
The “Feedback-ID” header will have a maximum of 4 fields, 3 are optional.
- “Feedback-ID” is the name of the header
- “a:b:c” are the optional 3 fields which can be anything the ESP chooses (ie: campaign, mailing, traffic type)
- “ESPid” is the only required field. This ID corresponds to an ESP customer and must be unique and persistent to that customer.
Gmail will aggregate data for the last 4 fields starting from the right side and ignore any extra fields. The data returned in the feedback report will be aggregated by the tag seen in the Feedback-ID header. Every tag will be included in the report and there are no limits on the number of total tags specified. However, Gmail will ignore tags with too few messages to prevent abuse.
FBL data will be aggregated by way of each identifier independently and not grouped across identifiers. Spam percentages will be reported across all the mails containing a given identifier, irrespective of the position in the identifier header. The FBL report will be sent in the form of a CSV attachment and contains data received by Gmail on the previous day by the ESP. This report is intended for gmail.com users and doesn’t support Google-Apps or Google hosted domains.
What are the DKIM requirements?
To prevent spoofing of the “Feedback-ID” header the ESP must strip any instances of this header first before inserting it and then DKIM sign it with the ESP’s domain key. This is in addition to any existing signature and is a practice commonly known as “double signing”.
There may only be up to 10 unique DKIM “d=” signing domains used to sign these headers but subdomains may be used as an alternative.
As far as DKIM key length is concerned, Gmail requires a minimum 1024-bit long key. According to the Gmail postmaster site, Gmail has been treating all emails signed with less than 1024-bit keys as unsigned since January 2013. They recommend affected senders with short keys to switch to RSA keys that are at least 1024-bits long.
How can we correctly implement Gmail’s FBL requirements in Momentum?
Multiple signatures are supported in the Momentum platform using Lua policy. OpenDKIM is now the preferred signing module in Momentum versions 3.6.0 and newer. More information about the Lua libraries for signing can be found in our documentation.
We’ve also created a simple OpenDKIM signing Lua policy that easily configures a second signing domain. You can find more info about that here.
Which Other ISPs Provide Feedback Loops?
If you’re looking for more information on how feedback loops operate, do refer to the “Complaint Feedback Loop Operational Recommendations” RFC 6449 document. And while we’re talking feedback loops, it would make sense to share links to the other trusted FBL providers out there, in case you didn’t have them already.
- Bluetie / Excite
- Earthlink (email only): firstname.lastname@example.org
- Gmail (beta, for select ESPs only, sends aggregate reports for privacy reasons (not ARF)
- OpenSRS / Tucows
- IBM Smart Cloud (email only) email@example.com
- Rackspace (formerly Mailtrust)
- RoadRunner / Time Warner Cable
- United Online / Juno / Netzero
- Yahoo! (requires DomainKeys or DKIM and its is the only Domain-Based FBL provider)
Now, to sum up: it’s great news that Google now provides an FBL. It has some quirks and senders will need to do some configuration to get it to work for them. But for Momentum users, that will be a straightforward process. If any readers would like to share their experience with the new Google FBL, we’d love to hear your story. Feel free to leave feedback in the comments.
VP, Email Delivery Operations
The Message Systems Training Division is proud to announce a new training offering to customers: Quarterly 3-Day Momentum Training Courses.
Once a quarter, the Message Systems training staff will conduct a 3-day Momentum training course at one of our offices throughout the world.
The first of these offerings is scheduled for Columbia on April 1-3, 2014.
The topics covered will include the following:
- Deliverability Best Practices
- Adaptive Delivery® Basic Overview
- Momentum for Sending Overview
- Getting Started: Installation, Starting Services and Basic Configuration
- Momentum: Sending Messages
- Hands-on Message Systems Adaptive Delivery
- Momentum Configuration Management
- DKIM, FBL’s, and Seedlists
- Logging with Momentum
- Momentum Bounce Handling
- Failover: Clustering
- Momentum Policy
- Web Based User Interface
- Lua Introduction, Fundamental API’s, and Policy Scripting
Attendees include professionals from different organizations and industries, each with their own unique sending issues. Gain insight into the challenges and solutions faced by other businesses or network and learn from industry peers. If you’d like to register for the training course, please email firstname.lastname@example.org.
Can’t wait for training to start? Here’s something to keep your occupied in the meantime. Check out our white paper on adding Momentum to your business, Overcoming the Challenges of High Volume Sending.