A friend recently sent me a URL via a Forward-to-Friend.com which is a service of MailChimp. While I really love the guys at MailChimp their URLs for their Forward-to-Friend.com are simply awful. There days of social media well designed URLs are finally being recognized by many as being extremely important, but not everyone gets it yet nor does everyone know best practices for designing URLs.
Make ’em Short and Sweet
One of the traits of a well designed URL is that they can be grokked with a quick visual scan. They should also be no longer than really necessary because one of the more common link sharing sites (Twitter) shortens long URLs automatically. There are many other traits of a well designed URL, some of which are specific to context but if it’s too long and you can’t understand something about the URL by looking at it something is really wrong. And anything that impedes sharing of links is a foolish addition. So I bitched about this URL on Twitter that a friend of mine sent me in email (let’s call her "Jane Smith" and @BenChestnut asked me to clarify. Here’s the URL:
What’s wrong with this URL?
So what’s really wrong with this URL? Let me count the ways:
This subdomain seems to imply that its specific to the US which I’m lukewarm on having a subdomain in this context it adds unnecessary characters. And what’s with the "1?" Is there a ".us2?" Is this just a server convenience? C’mon guys, hide that crap the user; they don’t want to know.
Okay, so it’s a cool domain, but you really couldn’t you come us with something shorter than 21 characters?!?
Uh, one word: "Why?!?"
Do I really need to say anything about this? I mean, it’s waaaay too long and how does any of this mean anything to anybody? The only thing is does it make the programmer’s life a tad easier to uniquely identify the user but only on the day it was implemented.
Another too long and non-meaningful computer number. The "id=" identifies the URL being forwarded. But does it mean anything?
What would be better?
So here’s a better hypothetical URL with analysis to follow:
The "fwd2.net" domain is owned by a squatter. Why not pay them a few bucks and pick it up? (or get something similar and short?)
Not super short but much like Twitter’s screen name it identifies the links shared by the user who picked the name "janesmith" (i.e it replaces "?u=0fea6c2e08126550f4c318d4b.")
Again not short, but as this would be selected by the user before sharing it would be as short as the user wanted it to be. So the user could have picked just "coyle-fernbank" or "fernbank-oct2" or similar. But what is really important is that it is meaningful!
And another benefit?
With this format you also get this URL:
At that URL you could have all the links "janesmith" shared when she is logged in, and she could set those shared links to be private or public, or later once more functionality is added the links could be made selectively available to different groups of friends.
Further there could be groups of URLs shared such as anything with a trailing slash could be tagged links, i.e. in this case "jazz":
Hopefully you can see a tremendous amount can be done with URL design but sadly there are still too few people who pay attention to it. Maybe that’s because there’s no book of best practices. Hmm, might be an opportunity there…
Still think it is unimportant?
And for you skeptics out there who really think that "users don’t look at URLs" take a look at the apps that are succeeding lately, Twitter being a main one. Most of them are designing their URLs well. Coincidence? I don’t think so.
Thanks for asking
Anyway Ben, thanks for asking. Hoping you see the value in it, make the suggested changes, and find that it’s made a positive difference.
8 Replies to “What’s Wrong with Forward-to-Friend.com URLs?”
Good points all around. Over the years, people have asked me variations of this same question, and I’ve been meaning to write a blog post about it. So I’m going to be really detailed in my answer below then just link here from now on from our knowledge base. :-)
Prepare for waaaaaaaay TMI.
A while back, we had to choose what URL to use for the FTF service.
We could’ve used the mailchimp.com domain, but that was risky. If people used/abused the FTF tool too much, then that would tarnish the reputation of mailchimp. See:
So it had to be separate.
BTW, this “abuse risk” is also why we severely limit the number of emails one can forward from this tool.
Another issue is that a bunch of our users are designers, who send campaigns on behalf of clients. They needed a domain where it’s not obvious that they’re using “MailChimp” to send the emails. We could’ve picked something really generic back then, like “f2f.com” or “ftf.com.” But for one, the cool ones were already squatted. And back then, we didn’t have a few bucks to pay squatters. :-)
Incidentally, this is also why the FTF landing pages were so “generic” looking for a while there. It had to fit a wide variety of users’ brands. We only recently wrapped that feature into our form designer tools so that users could customize it to match their L&F.
Secondly, we felt that the domain had to be very human-readable, b/c suddenly introducing a fresh new domain into every email coming from our system was sure to trigger spam filters: “What’s this newly registered domain that I’m suddenly seeing in millions of emails globally? I’m going to alert the abuse desk.”
Yeah, that sounds like a conspiracy theory, but spammers use bots to create fresh domains constantly, so networks like this:
are setup to keep track and communicate their findings all over the globe. Introducing a fresh new domain into our system is something we don’t like to do often, because it can take months to build up a good “reputation score.”
We learned this the hard way when we introduced our mcsv1,2,3…domains. Took us months to convince reclusive admins that we weren’t evil. Those were the days. There wasn’t nearly as much transparency with ISP and spam filter abuse desks as there is now.
Anyway, we figured that if our then-new FTF tool raised any red flags with abuse desk admins, we wanted the domain to be so friggin’ descriptive that it defused the situation immediately. It’s also why we chose list-manage.com for managing profiles and lists, and campaign-archive.com to host copies of your emails.
You make a good point that humans do read domain names, and they really matter. Even more so before clicking a link in their email. Another reason our domains tend to be on the long, overly-descriptive side.
Okay. That was all a long time ago (2006-ish).
Since then, we’ve brought additional data centers online, so yep — there’s a us2, and soon a us3, 4, 5… as well as a uk1 (warming up now), and soon a uk2, and perhaps au1, au2, 3…in the near future.
I can also say that every once in a while, we talk about upgrading the FTF tool. We’ve had lots of cool ideas for it, albeit none as cool as you recommend above. It’s just that every time we look at the FTF usage stats, we realize it’s *rarely* ever used. Receivers just hit the FWD button in their email apps, no matter how nicely we ask them to use the trackable FTF link. Sigh.
I absolutely *love* your idea of letting senders pick their bookmark names, and then having access to it later. But since few people use this tool, we can’t justify taking the time. Might be more realistic to just integrate with someone else who specializes in FTF, but again — domain reputation is key. 3rd parties have proposed integrating with them for this very thing, but if we partner with someone with a domain that’s also being used by bad senders somewhere, it can ruin our network for over 170k users.
You asked about whether the u= and id= stuff was necessary. It’s ugly, but it’s in there for tracking. Not for personally identifying forwardees (that would be evil), but aggregate data like, “this is how often your subscribers FWD’d your campaign.” We *had* some cool plans to visually represent all this virality, but meh—FTF is so rarely used, it’s sad. Poor little feature. :-(
Lately, we’re seeing more and more email marketers shifting to “SWYN” (share with your network) vs. “FTF. so I’ll admit most of our focus these days is on supporting that trend.
We did recently create our own URL shortener, eepurl.
We originally built it for twitter sharing and tracking, but the plans were to eventually expand its usage. It’s possible we’ll use it for the FTF functionality too. We’ve just had to slowly build up eepurl’s reputation among all the spam filters around the globe before we make it a permanent fixture in all outgoing user campaigns.
Sorry you asked?
Wow, Now that’s a long response! But no, not sorry I asked. Must better to get a reply than to shout in an empty forest. Thank you.
I hear you about the domain name, and of all my suggestions that was the least important in my mind. What you chose is actually not bad. URL design is a matter of juggling competing priorities so there will always be tradeoffs. Trading length and introducing readability is a fair trade for overcoming the issues you mention.
As for the “almost nobody uses”, I hear you. I will posit a rhetorical question though: “If the tool had more compelling features (i.e. a better URL structure) maybe people would use it more?” I don’t know that they would in this case but I do believe that well design URLs are a significant contributor to success for sites that use URLs correctly. For example, do you really think Twitter would be as successful if my Twitter URL instead of this:
So yes my comments were about your Forward-to-Friend.com but they were almost to trigger thought and discussion regarding URL design in general.
Awesome post Mike! I’ve been on the web as long as anybody, and I’ve read and re-read so many articles on designing and using urls and uris that I almost didn’t read on.. Really enjoyed this one, its always nice to read something that’s worth reading, which is becoming more difficult all the time..
Disregarding the main topic, the post is actually insightful into a trend I have been watching, that of developing and creating with the end-user in mind.
So I agree 100% with your approach of improvements based on the end user, and IMO that is the surest way to having a successful product.
Sidenote: I really enjoyed bens input, Behind the scenes breakdowns are fascinating.
I’m extremely honored you found this post useful. Your blog is definitely a cut above all the rest (as you can tell by my current blogroll!)
Is there a problem with a number in a short domain name (ie sss5.com)?
@Glenn – Hi! Numbers in domains are fine if they mean something to the user.
“Mailchimp not evil”?! Mailchimp and its redirects are blocked on my machine, and I report each new variant to blacklists as they sprout.
because of the many times I’ve been autoenrolled in a Mailchump campaign “per my opt-in” despite always leaving merchants the request:
“PRIVACY REQUEST: Please do NOT contact except regarding the specifics of this order. Please do NOT share my data with anyone for any reason”
If the mailing list industry were to start sending opt-in confirmation email at the start of an auto-enrollment, and then NOT continue if a response isn’t returned, then I’d take your anti-spam disclaimers seriously.
But you’ll never do that because you profit from spam. May you die young, lonely, and soon.
I think you got it all wrong. In a see of players with questionable ethics (i.e. other ESPs), MailChimp is one of the good guys.
I know the people at MailChimp personally. Their president spoke an an Atlanta Web Entrepreneurs group I hosted in 2008. I didn’t know him before that but after that I became a huge fan of him and MailChimp:
(Full Disclosure, last week MailChimp sponsored our WordPress conference: http://www.thebusinessof.net/wordpress/ But I believed in MailChimp before they sponsored my conference.)
What you have is not a problem with MailChimp but a problem with some of their customers. Just as it’s very hard for an ISP to police its webmasters, it’s also very hard for an ESP to police it’s emailers although I do know for a fact that MailChimp tries very hard to do just that.
Want some specifics? Here’s MailChimp’s President Ben Chestnut tweeting about someone criticizing MailChimp for not allowing spammers to run wild:
Here’s a whole group of people who love MailChimp (admittedly they are all email senders) but there is one who points out how they won’t service a class of email senders who are know to send lots of spam:
“Mailchimp have more features than most but their terms list affiliate marketing content as prohibited content. This is why I haven’t used them in the past.”
And from their “Unacceptable Use & Content” terms:
MailChimp was built to help businesses send email marketing to their customer lists. Simple as that. You may NOT use MailChimp to send the following types of content:
— Pornographic or potentially offensive material (even if it’s totally legit and double opt-in—it’ll just get us in trouble)
— Multi-level marketing campaigns (even if it’s just to a circle of friends)
— Deliberately false or misleading information
— Emails to people who are not customers, and who did not specifically request to receive emails from you or your client.
— Racist, hateful, and nasty emails meant to harass others
You get the picture. And just because it’s not on this list, it doesn’t mean it’s okay to send. If you’ve got something inappropriate, don’t be a jerk—keep it off MailChimp.
So all in all, I really think you are off-base in criticizing MailChimp. The old saw “Don’t shoot the messenger” applies here albeit in an unexpected way.
I have a suggestion; the next time you have a problem with someone spaming you via MailChimp, contact their support or better tweet @MailChimp and/or @BenChestnut on Twitter. I’d be surprised if MailChimp didn’t ensure that that particular company never harassed you with spam again.