How does email tracking work?
June 28, 2012 3:57 AM Subscribe
We're trying to understand e-mail tracking (Mailchimp etc). It's confusing.
Help!
I'd like to track whether people are opening the newsletters we send out, and I understand that this is possible (expected?) nowadays, but the ins-and-outs seem mystifying. We've tried readnotify, but it seems that gmail users, at least, are only counted as having read the email if they click '*Images are not displayed.* Display images below' . And they surely can't be the only email client that doesn't display images by default. Indeed, it seems like messages which include an invisible image full of script would be prime candidates for immediate spamfiltering!
Are there mefites who are happy with their solutions in this area? And what are those solutions?
I'd like to track whether people are opening the newsletters we send out, and I understand that this is possible (expected?) nowadays, but the ins-and-outs seem mystifying. We've tried readnotify, but it seems that gmail users, at least, are only counted as having read the email if they click '*Images are not displayed.* Display images below' . And they surely can't be the only email client that doesn't display images by default. Indeed, it seems like messages which include an invisible image full of script would be prime candidates for immediate spamfiltering!
Are there mefites who are happy with their solutions in this area? And what are those solutions?
I'll second that "In short, there's no way to know if someone has read an email." Anyone who tells you any different is lying to get your money. You'll need to track success in another way, e.g. click throughs or sales or something.
posted by Blake at 4:22 AM on June 28, 2012
posted by Blake at 4:22 AM on June 28, 2012
Apologies if offtopic, but isn't this snooping against netiquette?
posted by gijsvs at 4:27 AM on June 28, 2012
posted by gijsvs at 4:27 AM on June 28, 2012
Tracking whether people have read your mail is an "opt-in" thing, because it's an invasion of privacy for you to be able to see what I'm doing on my own PC when I don't even necessarily know you.
This is why people with mailing lists started using image tracking - to get around people who don't use read receipts. And this in turn is one of the major reasons why email providers disable images on emails from unknown senders!
So there is no good way to do what you want, it's not polite to do it without consent, and if you did find a way to do it, so would all the spammers, and mail clients would pretty quickly catch up and prevent it from working.
However, if you can make the images in your email actually useful to your core supporters (people who at least initially, want to read the stuff you write), then they may be more likely to enable images voluntarily.
posted by emilyw at 4:33 AM on June 28, 2012
This is why people with mailing lists started using image tracking - to get around people who don't use read receipts. And this in turn is one of the major reasons why email providers disable images on emails from unknown senders!
So there is no good way to do what you want, it's not polite to do it without consent, and if you did find a way to do it, so would all the spammers, and mail clients would pretty quickly catch up and prevent it from working.
However, if you can make the images in your email actually useful to your core supporters (people who at least initially, want to read the stuff you write), then they may be more likely to enable images voluntarily.
posted by emilyw at 4:33 AM on June 28, 2012
There are companies that sell services that claim to work, but it's all very suspect and never very well explained.
The technical details of how most of it works, especially for things like identifying mail client, are here; most of it comes down to image bugs.
Without images (or scripts, though that just doesn't work), there's no way email makes a request to a remote server, so there's no way tracking can happen.
posted by 23 at 4:35 AM on June 28, 2012
The technical details of how most of it works, especially for things like identifying mail client, are here; most of it comes down to image bugs.
Without images (or scripts, though that just doesn't work), there's no way email makes a request to a remote server, so there's no way tracking can happen.
posted by 23 at 4:35 AM on June 28, 2012
Think of sending an email like sending an actual letter. You write it and give it to your postman, who gives it to a mail truck, who sends it to a depot, and so on and so forth until it arrives. Once it's there, you can't tell whether someone has opened it or not!
Of course, you could include a postage-paid envelope inside your letter and ask people to confirm. These days people will tend to throw out such envelopes though.
This analogy isn't half bad if I do say so myself! The letter is a bunch of packets, the depots are routers between you and the recipient, and a postage-paid envelope is a "callback" image.
posted by katrielalex at 5:05 AM on June 28, 2012
Of course, you could include a postage-paid envelope inside your letter and ask people to confirm. These days people will tend to throw out such envelopes though.
This analogy isn't half bad if I do say so myself! The letter is a bunch of packets, the depots are routers between you and the recipient, and a postage-paid envelope is a "callback" image.
posted by katrielalex at 5:05 AM on June 28, 2012
I read much of my mail in mutt which displays it as plain text. I only download an image if I need it for some reason (usually never) so no one knows if I read their mail to me.
posted by Obscure Reference at 5:52 AM on June 28, 2012 [1 favorite]
posted by Obscure Reference at 5:52 AM on June 28, 2012 [1 favorite]
an invisible image full of script would be prime candidates for immediate spamfiltering!
All that needs to be embedded in the mail is an image with a unique ID number in its URL; websites can tell what URLs have been requested, so if they see that ID in the request log then they know that email was read.
This is exactly why most email programs do not load images by default anymore.
posted by ook at 6:18 AM on June 28, 2012
All that needs to be embedded in the mail is an image with a unique ID number in its URL; websites can tell what URLs have been requested, so if they see that ID in the request log then they know that email was read.
This is exactly why most email programs do not load images by default anymore.
posted by ook at 6:18 AM on June 28, 2012
The protocol used to deliver emails, Simple Mail Transfer Protocol, includes no guaranteed method for determining if a recipient of an email has opened the email.
That being said you can use lots of delightful hacks an HTML emails to surreptitiously figure out if someone is reading an email, because many webmail platforms expose parts of HTML that a browser can be "helpless" to. For a quite thorough treatment of this check out:
https://emailprivacytester.com
Send yourself a test email, and see what your webmail provider leaks! In the case of Google Mail nothing gets through if you don't show images.
posted by asymptotic at 6:31 AM on June 28, 2012
That being said you can use lots of delightful hacks an HTML emails to surreptitiously figure out if someone is reading an email, because many webmail platforms expose parts of HTML that a browser can be "helpless" to. For a quite thorough treatment of this check out:
https://emailprivacytester.com
Send yourself a test email, and see what your webmail provider leaks! In the case of Google Mail nothing gets through if you don't show images.
posted by asymptotic at 6:31 AM on June 28, 2012
Mailchimp provides us with the number of opens for the mail we're sending out, and Blue Hornet etc do the same thing. I have no idea how this works, but everybody else in the whole wide world is satisfied with the data on open rates MailChimp, Blue Hornet and other providers dish up.
In terms of images, not everyone has them disabled - that's an opt-in feature for Gmail and Yahoo mail services. And we try to account for images being disabled (this has nothing to do with tracking) by ensuring all images have some sort of alt text that displays when images are disabled, and ensuring the size of the header image up top is relatively compact.
We also tend to link to the content on our clients' websites.
As well, we really are executing permission-based email marketing - emails are sent out after a double opt-in process, so our lists are pretty small - 20,000 subscribers at the most.
But I think it's fair to say you can trust the data on open rates that MailChimp is providing. Everyone else trusts them.
posted by KokuRyu at 6:32 AM on June 28, 2012
In terms of images, not everyone has them disabled - that's an opt-in feature for Gmail and Yahoo mail services. And we try to account for images being disabled (this has nothing to do with tracking) by ensuring all images have some sort of alt text that displays when images are disabled, and ensuring the size of the header image up top is relatively compact.
We also tend to link to the content on our clients' websites.
As well, we really are executing permission-based email marketing - emails are sent out after a double opt-in process, so our lists are pretty small - 20,000 subscribers at the most.
But I think it's fair to say you can trust the data on open rates that MailChimp is providing. Everyone else trusts them.
posted by KokuRyu at 6:32 AM on June 28, 2012
I am a web marketer and I use Campaign Monitor for all of my clients' email marketing, without exception. I am so incredibly happy with the stats CM provides that I am quoted raving about them on the CM front page. I am not affiliated with them in any way except as a happy user.
Open rate tracking happens in exactly the way ook describes. "I'd like to track whether people are opening the newsletters we send out" is the smallest piece of data you can get, and actually only minimally instructive on its own. The value of email tracking is in the data you get on what people do when they do open the email, with click tracking and social sharing tracking. While we are only going to be able to gather data for a subset of subscribers, these things can absolutely be counted and tracked right down to the individual email subscriber, and provide enormously valuable data.
I think your option is to restructure your newsletters so they reside on a webserver and the emails you send out only contain unique URLs pointing to those newsletters.
This is a terrible idea. Your drop off rate will be spectacular. Do not do this.
posted by DarlingBri at 6:38 AM on June 28, 2012
Open rate tracking happens in exactly the way ook describes. "I'd like to track whether people are opening the newsletters we send out" is the smallest piece of data you can get, and actually only minimally instructive on its own. The value of email tracking is in the data you get on what people do when they do open the email, with click tracking and social sharing tracking. While we are only going to be able to gather data for a subset of subscribers, these things can absolutely be counted and tracked right down to the individual email subscriber, and provide enormously valuable data.
I think your option is to restructure your newsletters so they reside on a webserver and the emails you send out only contain unique URLs pointing to those newsletters.
This is a terrible idea. Your drop off rate will be spectacular. Do not do this.
posted by DarlingBri at 6:38 AM on June 28, 2012
But I think it's fair to say you can trust the data on open rates that MailChimp is providing.
Only in that there exists no more accurate method to track open rates than the (probably very) inaccurate method they use.
Image bugs can give you a hard minimum, an "at least this many users opened the mail" value. If it were known what percentage of mail users have image loading disabled in their mail readers, it would be possible to get a statistical prediction of the real number from there -- but I doubt that percentage is known or even calculable; it would probably vary wildly depending on your audience.
As darlingbri points out, read receipts, even if they existed, wouldn't be a terribly useful number; what you really want to know is if the reader follows up on the message by clicking a link in it. This can be tracked accurately (by putting an ID on the link URL the user clicks instead of on an image the user's mail program may or may not have loaded.)
posted by ook at 7:11 AM on June 28, 2012
Only in that there exists no more accurate method to track open rates than the (probably very) inaccurate method they use.
Image bugs can give you a hard minimum, an "at least this many users opened the mail" value. If it were known what percentage of mail users have image loading disabled in their mail readers, it would be possible to get a statistical prediction of the real number from there -- but I doubt that percentage is known or even calculable; it would probably vary wildly depending on your audience.
As darlingbri points out, read receipts, even if they existed, wouldn't be a terribly useful number; what you really want to know is if the reader follows up on the message by clicking a link in it. This can be tracked accurately (by putting an ID on the link URL the user clicks instead of on an image the user's mail program may or may not have loaded.)
posted by ook at 7:11 AM on June 28, 2012
I guess the point about trusting MailChimp is, even if the information is unreliable, everyone else's information is unreliable too. And you can only measure your success by comparing your performance with industry standards.
posted by KokuRyu at 8:06 AM on June 28, 2012
posted by KokuRyu at 8:06 AM on June 28, 2012
There is an option in MailChimp to add Google Analytics tracking to all URLs -- you'll see it on the setup page. So if you have a unique URL for each image, that should help with tracking.
posted by vickyverky at 10:09 AM on June 28, 2012 [1 favorite]
posted by vickyverky at 10:09 AM on June 28, 2012 [1 favorite]
Response by poster: Well, thanks to everyone for your replies. We've gone with Mailchimp for the moment, and it does seem as powerful as you can get.
In response to several posters - Read receipts are useful because they allow us to gauge how effective our subject lines are at persuading people to open the email in the first place. e.g. Low read rate, low click through = our subject lines are terrible; high read rate, low click through = our message content is terrible.
I'm pretty sure that "uninspiring subject line" is one of the biggest impediments to getting people to take whatever action you want them to take from a mailshot ...
Thanks again to all.
posted by Ricardus at 10:18 AM on July 2, 2012
In response to several posters - Read receipts are useful because they allow us to gauge how effective our subject lines are at persuading people to open the email in the first place. e.g. Low read rate, low click through = our subject lines are terrible; high read rate, low click through = our message content is terrible.
I'm pretty sure that "uninspiring subject line" is one of the biggest impediments to getting people to take whatever action you want them to take from a mailshot ...
Thanks again to all.
posted by Ricardus at 10:18 AM on July 2, 2012
I have no idea if Mail Chimp will let you do this but FYI, Campaign monitor will let you A/B test subject lines before sending the bulk of your mailing.
posted by DarlingBri at 11:08 AM on July 2, 2012
posted by DarlingBri at 11:08 AM on July 2, 2012
« Older Southwest Spirit Magazine Surfing Article | Please recommend movie scores and soundtracks that... Newer »
This thread is closed to new comments.
Yes, there are various hacks to send the message out in HTML and embed an image with a unique URL that gets logged whenever it's accessed (thereby identifying that a user has "read" the message), but as you've discovered, they only work under certain circumstances because they rely on the client to act in a convenient matter.
I think your option is to restructure your newsletters so they reside on a webserver and the emails you send out only contain unique URLs pointing to those newsletters. Or if you're just interested in knowing how many people read the newsletter (why do you need to know if particular people read the newsletter?) just keep it on the webserver and send everyone the same URL and count how unique hosts have accessed it.
posted by RonButNotStupid at 4:11 AM on June 28, 2012