[registrars] Re: FW: ICANN per-name fee at start of AGP vs at exit.

[Paul Goldstone]

Paul,

I completely agree with your statements, and appreciate the time you 
took to post them to the list.  Jon Nevett recently asked for comments 
on what the thresholds should be, gave a point of reference to ICANN's 
current fee reductions, and a few people have already commented.

I made an initial suggestion of somewhere between 3-5% of new 
registrations for the current (or previous) month, plus the option for 
discretion in special cases where the registrar could prove that they 
were dealing with testing or fraud. This is a model Nominet has in place.

Since this affects all registrars in such a major way, hopefully there 
will be a lot more suggestions.

Rather than wait to see what the ICANN budget looks like, risk having 
to vote against it, then go back and forth, do you think it would be 
beneficial for us all to agree on a reasonable threshold and send a 
statement to ICANN ahead of time, or at least have that available for 
the meeting in Delhi?

To your other point, although the loss from a chargeback for a 
non-refundable fee is clearly a major concern, I wanted to mention 
that my credit card processor does not charge a separate fee for 
chargebacks.  My previous processor charged a whopping $15!  If anyone 
is interested in comparing their rates against what I pay, feel free 
to contact me off list.  Who knows, maybe I'm the one overpaying and 
will end up with a better overall rate. :)

Additionally, we implement many of the fraud detection procedures that 
you mentioned, and I think it would be of great benefit to all 
registrars if some of us were willing to discuss and share that 
information.  I can see it starting out as a discussion and sharing of 
procedures, and perhaps even evolving to sharing known data about 
fraudsters.  I think this suggestion actually came up once or twice 
before.  Needless to say, it would not be appropriate to discuss those 
matters in a public forum, but I wonder if we should get a group of 
people together who want to do so off list.  Any takers?

~Paul
:DomainIt


At 06:15 PM 1/31/2008, Robert F. Connelly wrote:
>At 02:32 PM 1/31/2008 Thursday  -0800, Paul Stahura wrote:
>>I don't see this on the list yet?. Am I having trouble sending to the list again?
>
>Dear Paul:  I'm checking the RC list to see if you are on it with this email address.  I suspect we have your old address.
>
>In the mean time, I've posted your comments to this list.  Regards, BobC
>
>
>> 
>>From: Paul Stahura 
>>Sent: Thursday, January 31, 2008 11:02 AM
>>To: 'registrars at gnso.icann.org'
>>Subject: ICANN per-name fee at start of AGP vs at exit.
>> 
>>We registrars need the AGP.  We all know we do.  But let me explain why in case others who may read this post do not know why we do or how its used.
>>The AGP allows registrars to combat fraud.  
>>When users requests a registration, we registrars do not know if the users intend to 1) pay us or 2) use the name for a purpose which violates our acceptable use policies (such as to spam).  The registration is probably legitimate, though we don?t know for sure, and users demand fast registration service, so we register the name for the user in real-time.
>>Now we (at least eNom, and I?m pretty sure most registrars) do not just wait 6 months to see what happens.  Wait and find out if the credit card was stolen.  If the user spams like crazy.  What we do (again at least eNom but probably most registrars do this) is score each registration.  We use email addresses in the whois, source IP, account, and other metrics in an attempt to figure out the intent of the user.  We automate this as much as possible but it still takes human judgment (our fraud prevention departments) to take all these data points in.  This takes time; though we try to do our checks as fast as possible.  Some we can do fairly quickly, others, takes days.  Five days was (and still is) a reasonable amount of time for us to do our checks.  If we determine that the user will not pay us or is likely a spammer,  we delete the name, and do not charge the card, or refund the fee if it was charged.  Obviously its in our economic interest to charge the cards as soon as possible.  This is the entire purpose, in my opinion, of the AGP.  It?s not to correct misspellings.  It?s to prevent fraud and reduce a high chargeback rate on registrars (because in many cases the credit card is not even charged, so therefore there can be no chargeback).  I know some registrars have had to pay millions of dollars in fines in the past for high chargeback rates (which is why I?m fairly certain we?ve all evolved our own, probably similar, fraud scoring systems).  This is not a trivial amount of money.
>> 
>>OK so ICANN wants to charge the per-name fee on the day the name is registered and not on the day names exit the AGP.  If this happens, then we will be paying ICANN for names which we delete due to a high fraud likelihood.  I propose ICANN charge us only after a reasonable number of names (after a threshold number).   This number can be on an absolute per-cred-per-time (per day, month whatever) basis or on a percentage basis (based on the number of names exiting the grace period).
>>Without a threshold I will be voting against the budget and I urge you to do the same.    The egregious ?tasters? were doing hundreds of thousands to millions of names per day. A threshold which is orders of magnitude under this (and at the same time, orders of magnitude above the number we delete in our fraud-fighting efforts) will eliminate the giant ?tasters? (whether or not you agree on tasting as a legitimate business practice) while allowing us to continue to keep our chargeback rates low and reduce fraud without increasing the dollar burden on us.
>> 
>>I look forward to discussing this with you further on this list and in India.
>> 
>>Paul
>> 
>
>~~~~~~~~~~~~~~~~~~~~~~~
>After all is said and done ---
>A lot more gets said than done;-} 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mm.icann.org/pipermail/registrars/attachments/20080131/46fb0fe5/attachment.html>