>From: "Nick May" <nick@kyushu.com>
>
> I am setting up a few lists to send a smallish/intermediate number of
> emails into the various keitai networks (a few thousand -> 10,000 a day,
> daily).
>
> The lists are fully opt in, obviously....
>
> I have no experience of this. Am I likely to face problems? Are there any
> peculiarities of the various services email servers that I should look out
> for? Are they prejudiced against large numbers of bcc's ? (is it better to
> send "to" everything?) How strictly "correct" does my DNS have to be - do
> they do reverse lookups?
Since April 2000, we've distributed numerous small and large opt-in
(no spam here) mailings (with some being 2,000,000+ recipents per
campaign) for companies like eBay, Wells Fargo, and Bank of America.
These issues you raise are very valid, but will depend on the recipient's
email provider.
In general, this is some of the things that we have found (I recommend
implementing all to be safe):
1) BCC: Do not BCC -- have each email individually addressed. Large
BCC's raise flags, even if they are all opt-in.
2) Reverse DNS Lookup -- make sure this is activated. This is a must.
3) Thundering Herd -- If you are sending large messages quickly then
you may have a thundering herd, will the
receiving
mail server is overwhelmed or triggers the
"bulk
mail" routine and either cans a message or
puts
it into a bulk mail folder. Yahoo will
often place
messages like this right into the bulk
mail folder.
Their algorithm seems to be based not only
on
total number of recipients, but also some
peak
transmissions. I noticed this when we
where
distributing around 20 a second to them,
this would
occur -- there are some ways to mitigate
this, but
a little bit of a hassle.
It's interesting to note, though, that even Yahoo's own properties, like
their eGreetings, gets "bulk foldered" during holidays by its own system.
4) Subject line: Be very careful what you put in there -- even as opt-in,
all these spam filters are picky, terms like "Free", "--- Off", "Save", etc.
all can and many times do trigger spam issues.
> Do any of the services blackhole email senders - if so, under what
> conditions? I know that lots of IP ranges are blackholed - the NTT OCN
> single IP static addresses for example
5) MAPS - http://mail-abuse.org
It depends, generally based on customer/recipient complaints of spam.
MAPS will now lately give you some due process before this happens.
I talked to Peter Popovich Director Operations at MAPS about this
before - he was very reasonable.
Also be sure to keep a database of the recipients, opt-in process,
ip address and date and time stamp -- this is valuable for MAPS.
To mitigate this, be sure to have the alias mailto:abuse@yourname.com,
and monitor this -- respond immediately to any spam complaints with a
small email and take them off. You should have a very small spam complaint
list with opt-in. In many cases, we received 1 out of every 1,200,000 on the
low end to around 200 out of every 1,000,000. So it verys, and depends
on the mailing, how it is done and how the recipient responds.
Also, we create and maintain seed email accounts, we seed the house
list with our email addresses and monitor these for delivery accuracy
(formatting, errors, etc.), performance and see if they are counted
as "spam". For instance, you can set them up on docomo or whatever.
Most of our work has been on the desktop, so we maintain, AOL,
Yahoo, Hotmail, WebTV/MSNTV and then a bunch more that
are OS, and email client related -- such as Eudora version x.x,
MS Outlook, MS Outlook Express, whatever.
You can check this for certain handsets, too.
> I stress - this is all above aboard email. My email address dates from
> 1986 and I get horrid amounts of spam, so I am personally rather sensitive
> about it.
No need to apologize if it's opt-in. Couple words of caution: if mailing
lists are allowed to grow stale, some people forget they subbed and call
that spam or need to be reminded. Also, email addresses, to change hands
over time, so you want to keep them getting exercised to keep the problems
out. Better to have 5 spam complaints a week or 10 weeks, than 500
complaints in one day -- could be the same reasons why, but 500 makes
you look like a spammer.
We also usually sample a new "opt-in" list to check -- we look at metrics
like bounces (hard and soft) and unsubbs and monitor abuse. Actually
the abuse complaints are so low, but so important and urgent that we
route via SMS to our cell phones to reply. It's very effective.
> How do you handle email address changes that lead to bouncing? Is it worth
> putting bounced mail addresses into a database and cutting the account if
> it hits an arbitrary number of bounces, for example...? (The BBC appears
> to do this, or something similar. It is probably fairly trivial to do.)
It depends. Many issues here. There are hard bounces (i.e. address doesn't
exist, no host name) and soft bounces (temporarily didn't go through,
mail box full, host down, whatever).
Sometimes it's hard to resolve all soft bounces and hard bounces, but you
can set up rules, too. One or two hard bounces you are out, or only
one. In addition, soft bounces may be good for so many.
You'll have to calculate the ROI on what's it cost to send a bounced email
versus what it's worth if it finally goes through.
In my experience, a list that has aged for a few months, can easily
have 25% bounces. That depends, though, on your company, product
offering and recipient audience.
By the way, what we do, is hold bounces and set different business
rules for each client on cleaning them.
Also, make it easy for people to unsub -- either by replying and other
clicking an easy link.
Anyway, there's a lot of possibilities, issues, solutions, etc. here,
so a lot will depend on your needs and use cases, but hope this
provides a little framework and food for thought.
Good luck!
James Santagata
A U D I E N C E T R A X
http://www.audiencetrax.com
Received on Fri Aug 23 21:39:47 2002