- The convenience factor is compelling, like auto-complete.
- As a v1 you can see the true magic in a seamless signup.
On the downside:
- It does something unusual / unexpected / creepy.
- It's too slow - noticeable lag vs. say the auto-suggest in search.
- If accuracy is good why bother exposing that to the user? Intercom.io does this all in the backend off of email only registrations and seems to have about a 70% hit rate. The data is useful to business owner but I'd bet the user would feel less enthusiastic knowing that I have their personal fb, twitter, linkedin, etc as soon as they register.
I am sure that svbtle can show the A/B data validating this is worthwhile for THEIR audience.
But I'm not sure for a broader use case this is going to help due to the factors mentioned above.
> If accuracy is good why bother exposing that to the user?
That's a great point. A prior startup I worked at used a nice social referral tool, whose name slips my mind right now, but their backend would asynchronously pull Facebook, Twitter, etc. links for each of your customers. It was really convenient and worked for 80-90% from what I recall.
Best feature suggestion ever. But what would be cool is to do that, but instead of pre-fill in their password, notify them that they are using a password that is known to have been a previously leaked password of theirs.
Presumably, they have some preexisting database that maps email addresses to personal information. Their Privacy page, however, is not transparent how exactly they got this data:
> Broadly speaking, we collect information in three ways: (1) when you provide it directly to us, (2) when we obtain information about you or your company through trusted third parties or indexing systems, and (3) passively through technology such as “cookies”.
This is a disconcerting use of "secret sauce," particularly since email addresses have the same weight nowadays as usernames.
Alex McCaw mentioned in Clearbit's launch post earlier this year that Svbtle is using their API to auto populate their sign up form:
http://blog.alexmaccaw.com/clearbit
My guess is that Magic is using Clearbit's API as a base, then layering on their own scraped data sources as well, which is a pretty common strategy for these types of APIs.
Probably not. That clearbit form tells me that they don't have my email address, but my information was pre-populated into Svbtle when I used that same address.
I got the same thing the first time, but I did it again and it sent me an email to claim my account. Maybe their site got overwhelmed by HN requests? Not sure what happened.
If you don't want to provide clearbit with additional information (e.g. location), it might be useful to remove yourself with this link via a VPN or hide.me
Does anyone know if Clearbit is just a frontend for Fullcontact? I tried both recently, and all I get from Clearbit is basically a subset of Fullcontact (plus some added data eg lat/lng)...
That link doesn't seem to provide opt-out, it just provides a way to request the deletion of specific records. But you have to find the page for the record before you can request it to be deleted.
Why would you use this instead of login with Facebook/Google/Twitter/etc., which only requires two clicks from most users and actually has better privacy implication since the user knows which data you can access?
Or maybe the goal of the service is for web sites to surreptitiously get more data from users who are otherwise unwilling to provide it, but are somehow willing to provide an e-mail address tied to their real identity?
Tried some short gmail handles, can confirm. Creepy and stupid. Cloud-based datastores coming with your browser (aka what Chrome has* and probably its competitors too) have better performance, privacy, and UX; I can't think of any reason why I'd use something like this.
* - AFAIU if you set a sync passphrase you also get some (probably limited) form of "zero knowledge" client-side encryption; this makes me hopeful they implemented it properly i.e. "we literally can't access your shit" https://support.google.com/chrome/answer/1181035?hl=en
It sends you a confirmation email. Inside this email is a link to your profile. You can either add information (linkedin, facebook, angellist, etc.) or delete your profile.
Only if they choose to supply that information to Gravatar in the first place. Magic is not giving people the opt-in option, it's just opting you in whether you like it or not.
On the one hand, yes, I can see how this could improve sign-up rates for those who are concerned about those things, and I count myself among that number from time to time.
Say what you will about Dustin Curtis, but everything he makes is just so nicely designed and enjoyable to use.
I'm curious about the pricing model: $20/mo for 2,000 requests. I wonder if you still pay for the request if it returns a non 200 or if their API can't find a name.
I have a fairly unique last name, which is in my email address.
So far two dead relatives that never really used the internet have been "marketing matched" to my email address. So I get spam for my deceased grandmother and uncle. Its a little jarring. I'm not sure what kind of matching they do but clearly its not working.
Plus isn't the point of using a service being able to decide to match it with twitter, Facebook or just not link to those things.
Because it helps your users save time, Magic increases conversion rates and makes the web easier to use.
Honestly, how much time do you really save by auto completing a couple of fields? This seems like a solution in search of a problem.
Also, while I don't have the data to speak for other users, I personally don't see myself paying for any service that presents my own personal information to me before I even know what the fuck it is.
I wonder if a new way to do signups would be just an emailto: link that you click and send. From there it automatically logs you in and sends you a password.
This is what I was thinking as well, but not necessarily email you a password because it's always some random crazy thing. More so remember the computer you clicked the email link from and either keep you logged in or you can log in with just your email from the computer you clicked the email link from. Sure there could be some security risks with your phone being stolen or a snooping SO, but most sites don't necessarily need to have that worry. For instance, commenting platforms like disqus, reddit, or your favorite news platform, or some site you'll probably only use once. With sites like this it doesn't matter much of other people gain access to it. It's not like your social platforms, LinkedIn, bank portal, government site, etc..
Going out on a limb here, but what if this really a data training layer on top of Clearbit? Maybe they sell the human validated data back to Clearbit or downstream?
I hate how I only get those prompts when the field matches whatever fields Chrome thinks it should, yet many companies don't use these field ID's, so Chrome sits there stupidly.
I would love the ability to teach Chrome that this is a form and yes these are valid fields. Maybe somehow submit it to Google and have them update whatever regex they use. I feel like it fails about 50% of the time.
- The convenience factor is compelling, like auto-complete.
- As a v1 you can see the true magic in a seamless signup.
On the downside:
- It does something unusual / unexpected / creepy.
- It's too slow - noticeable lag vs. say the auto-suggest in search.
- If accuracy is good why bother exposing that to the user? Intercom.io does this all in the backend off of email only registrations and seems to have about a 70% hit rate. The data is useful to business owner but I'd bet the user would feel less enthusiastic knowing that I have their personal fb, twitter, linkedin, etc as soon as they register.
I am sure that svbtle can show the A/B data validating this is worthwhile for THEIR audience.
But I'm not sure for a broader use case this is going to help due to the factors mentioned above.