Hey guys - been mailing GI for a little while now and can't seem to step it up to the next level where I'd like to be or get IPs to deliver for more than a week before moving to another server. Any tips on how to treat IPs with good GI data to get them to last a little longer? Let's say with 60 IPs. I'm sure I'll get some bad comments here, but just looking to stir up a little GI mailing convo. Thanks
Hey Mike - first off don't worry about any bad comments - if they come - because there will always be haters on GI. With that said, I'll try to help a little. I'm guessing by "next level" you're referring to revenue, correct? 1st - Welcome to GI. You're going to burn IPs and go through servers. It's the nature of the GI beast. Knowing this will help you if you do a couple of things to kind of "ease the pain" of losing them from time to time. Start lining servers/IP ranges up so you have immediate rollover capability. Being prepared for the loss will help you stay consistent as far as mail going out. If you get with a solid ISP (one comes to mind) that you can set up servers with that allows for rotation of IPs, it will help, but with these guys you have to follow some guidelines which I'll touch on here in a bit. 2nd - Set up a click thru server if your platform allows for it. What I mean is set up a dedi box with 2-3 IPs on it that you'll NEVER mail from. Use the IP on this server for your click thru's so WHEN you lose a server (main block of mailing IPs) you won't lose your residual clicks. The other option is to constantly monitor your IP space and have this server on standby so when your server goes down you do a quick zone file change so you don't lose your clicks. This however can raise some serious redflags at registrars so you may not want to do this if at all possible. 3rd - Seriously CLEAN YOUR DATA! Use ImpressionWise and Quickie. It might cost you more up front, but in the long run it'll pay off and pave the way for you to get to the next level. Why both? Because each offers something different. With IW you're going to get a very clean file back, itemized and completly mailable. BUT, and there's always a but, they don't do anything about hard bounces. That's where Quickie comes in. They claim (I say this as my disclaimer but there are plenty here that vcan testify to this I'm sure) to be able to remove 70-90% of hard bounces. And to run through two services is just simply a good CYA (cover your ass) move for GI mailers. As you're finding out, you can burn up IPs quick and hard bounces certainly don't help your cause. And when dealing with ISP's that allow for rotation (as I mentioned above) these added precautionary steps go a LONG way in building up their trust and confidence you aren't trying to clean data on their IP space but rather you want to make a serious run at masking money with GI. 4th - Rotate domains frequently. I have always swapped out domains in around 3-5 hours. So do it quicker, others not so much. The way I learned the GI ropes was churning domains. Now I did find out that I could let them go much longer, but I erred on the side of caution and just settled into a routine of rotating them. As I mailed with .info's it was cheap so I figured wtf. 5th - And like all mailing/data, do your best to keep your list fresh (work in openers/clickers/fresh clean data), keep offers rotating so it doesn't get stale on you. Realize you're mailing a volitile segment of the email marketing world. Most would agree that around 98-99% of all traps are in GI. Getting SBL'd is almost a given. As is losing servers/IP and domains. So keep your shelves stocked up and settle in for the storm. GI can be great money if you know what you're getting into. It can also bite you square in the ass and leave you on the side of the road crying for your mommy. These are my recommendations and advice. Everyone that mails GI has there own techniques and methodology and I hope a few chime in with some solid advice for you. If you have any more questions, by all means ask. That's what we're here for. Share your successes as well as your failures with us and we'll do our best to help you along. Good luck! :afro:
Wow that was probably the best response I've ever had on a forum. Thanks a lot for that. So to go down the list: 1 - Yup, I'm def used to burning IPs/Servers with GI so it's not coming as a shock. I always have a backup server ready to go, but believe it or not I have not been getting shut down and have only had a few complaints. I am just switching servers on the fly when delivery is suddenly awful. 2- In good news, my software already incorporates a click through server for that reason exactly. I never mail off it and it is maintained indefinitely. If a server goes down, no residual clicks are lost. 3- I use impressionwise and quickie already, but there are still tons of hard bounces so I don't think the hard bounce cleaning works too well. 4- Rotating domains - this is something I have only done a little. I actually am very interested in focusing on this as this is something I have overlooked. Any tips on how often to change out domains? Also, how many IPs per domain? Overall, it looks like I'm on the right track. Currently it seems that with GI atleast I need to constantly be sending different offers which is obviously quite challenging since some of them destroy, while some flop ruining that day's revenue. So now that I know I'm on the right track, I have a few questions: 1 - How many IPs per domain is best? 2 - How frequently is it recommended to replace domains? Rotate or Replace? 3 - Mail off same IPs everyday? 4 - This is probably personal preference and some may not want to share, but what is best time for morning drops if you plan on 2 for example? I've been doing 6am EST and 9am EST for example based on best results from TLDs 5 - How long of a wait before you guys hit same offer again to same list? I know these are specific questions and some may not want to share (ha) but thought I'd ask. Doing decent now, but should be making a good amount more than I am. thanks again for the great response.
for a stable list you're done cleaning you really only need to deliver from one ip at a time, after you've delivered to everyone you can on that one ip, take all the 421 users such as "too many delivered this hour" or "youre on xyz domain or temporary blacklist" and send them on their own ip don't use .infos for GI most likely you're on URIBL within minutes of your campaign starting, so every time you start a new campaign or switch ips until you get cloudmarked and/or put into the various spam assassin oriented blacklists you should also carefully track where your clicks are coming from, if you have cables mixed in with your gi or other large isps, maybe your response drops so dramatically because you get blocked on those domains quickly while the rest of your gi list remains stable you can only figure this out yourself by testing.. split test, half the campaign at 6am, half the campaign at 11am, etc don't believe the myths, split test. if youve got an offer you want to mail again but you just mailed it yesterday or 3 hours ago, mail half the list to that offer again, half the list to another offer you want to send, see which does better
Thanks again for a great response. So you said don't use .infos for GI. 2 things - 1) I've split test .infos and .coms and noticed almost no difference, so is there another reason for not using .info's aside from spamhaus? 2) Obviously .coms are 5 times as expensive, so this is why I am also asking how many IPs per domain and new domain frequency. Currently I throw a .com on 4 IPs (a box with 100 IPs for example I would have 25 domains - 4 IPs to each domain) and just run that box like that with whatever # of IPs I have until the delivery sucks. I'm guessing this isn't ideal, so what is recommended? From your recommendations it sounds like you're not spreading a big send out over a lot of IPs. I've been told 2 ways and both people swear by their preferred way - Spread out decent sized drops over tons of IPs, while others say with GI send tons off of a few IPs then move on to next IPs. Sounds like you're the 2nd option? As far as cables/isps, my lists are as GI as you can get so I'm good there.
.info domains can give you penalties in spam assassin and other spam scoring systems, but if you're sure it's not affecting your revenue then i wouldn't worry about it i subscribe to the second method yes, a lot of people swear by things they've never really put any solid testing methodology behind and/or because other people have told them that and/or because it's just what they've always done and assume it must be the best for that reason
Ok...so by the second method, are you talking like 100k off 1 IP and then move on or are you talking like 1mm? How many IPs per domain are you doing with your method? 1? I want to test both ways and get this thing delivering!
When doing GI, after bounces, spamcops, etc have been cleaned from the list, I mail 1 offer to the entire list from one ip, one domain, even if its a 1mm+ list. but, i also can deliver fast. if it takes you 3 hours to get the list out maybe its not such a great idea.
gotcha - totally different method. I get it though. Seems to burn IPs but if you can get a lot out fast you can use minimal IPs and have a fresh IP/domain each send. I'll have to give this a shot.
I never get a box with a lot of IPs simply because I know I'm going to lose it eventually...hopefully later than sooner but still, it's gonna die. So I buy boxes with a /28 or /27 because you can pick those up for around $100-$150 and then it's all about pounding out as much volume as you possibly can to maximize your ROI on it + your domain cost. I use .info's exclusively. I don't see a point to burning up .com's on GI and certainly not at the rate I go through them. So what I do is this: 1. set up rDNS on the /27 using a nice old .com, I'll never be mailing from this domain however. 2. set up my mailing domains across the /27 like this : 34webz.mailingdomain.info ( I do this on every mailing IP with ONE domain. But I set the entire range up with 20-30 domains as I will run mail out for around 2-4 hours before I swap out the domain for the next one in line). It really depends on how much I'm getting out, but it's safe to assume 900k per hour on my platform. I usually see more like 1.1-1.3MM but I like to schedule it so I know basically how many times my active lists will get hit per domain. 3. keep this up for as long as the server is up and running and I'm still making $. I can (and have) kept a $100 server up and mailing for a month with zero problems from the ISP and I was sending on avg. 20MM a day. I killed that server! 4. As far as when to drop - 24/7! Why would you not? I don't give two shits where they are geographically, what time it is, when I "think" is the best time to send them a coupon for free McDonalds...I let them decide when it's best for them to check their mail. I just simply provide them material to paruse at their leisure. My job is to get the mail out, not try to figure out when my recipients want to look at mail. 5. Hitting the same list - pound that sumbitch! I do a lot of creative rotations. I will test an offer with anywhere from 3-5 creatives if they're available. It gives me the sense of not over delivering one particular offer to my list, but also quickly shows me which version will work the best, then I can concentrate more on that one and send it to my better lists for the $ drops. If you have big lists up, you can pound it with decent frequency because every drop, every new IP/domain will deliver to different addresses. So over time you're going to get the most out of your data (opener/clicker wise) if you keep it active. Never be scared to test, test and after you test...test some more. HTML vs. text. Froms, subjects, content, image size, format...these are all things that you can play with on any one offer. If your platform automatically rotates froms/subjects then create small sets of them and test each one. If you find a solid from/subject combination that gets good opens, try text only to see if it improves your CTR. Just don't fall in a rut where you start thinking "I GOT this shit now!" because you don't. You might have a good piece of it...for a while, but wait...it'll change. So stay on your toes and you'll see that you can really scale it up the way you want to.
Great info! So just for clarification, you're using one .com with no subdomains across the whole /27 for RDNS, but you don't mail off that domain? Hate to say it, but I didn't realize that was doable. Also, when you set up the mailing domains across the /27 with "34webz.mailingdomain.info" for example, you're saying atleast for the first 2 or 3 hours, every IP on the /27 is assigned to 34webz.mailingdomain.info, with the .com as your rdns? THEN you keep same IPs after 2 or 3 hours and just a new .info assigned to the whole /27 and keep the same rdns? If so, got it. It would seem that a drop of 2-3mm would be fine before a .info swap across a whole /27. So overall, you're not exactly giving a daily limit per IP volumewise it looks like. Just rotate domains and keep blasting off same /27 all day? Contradicts a lot of what I've been told, but I'm more than open to experiment to get good delivery!
yup. try it. Again, yup. I never change rDNS, I only swap out the mailing domains every 2-4 hours and I use the same IPs over, and over...and over....and over. Nope, I impose no limits on my IPs. The only thing that keeps me from doing more volume on them is the platform I'm using. But then again, I don't consider 20MM a day to be very "limited" :top:
Well looks like I got what you're saying! Well I'd have to say this is the third completely different method I've heard for GI, but seems to make most sense. So the 3 methods I've been told by GI guys are: 1) Your Way 2) Using 100 IPs for example, assign 4 IPs per domain and spread out say 3 million over the entire 100 IPs, get good delivery and repeat but only once daily (the idea is to keep very low volume off each IP daily so they will last). 3) Send a ton off 1 IP with a fresh domain, then next send move to next IP and fresh domain and repeat. To me, 1 and 3 seem to make more sense than # 2 since domains get listed so quickly. Is there any benefit to #2? It seems that #1 would allow the same /27 to be used with decent delivery for a longer period of time due to domain rotation.
with #2 you would have to ensure each mx (not each domain) was only going to be hit from the same ip, otherwise the entire range will go on whatever their backlist is immediately. My guess would be people who think they're gaining by doing this are just trying to overcome limitations in bad MTA software.
So it looks like bottom line is, domains need to be constantly changing? You can't keep the same domains for a month on a set of IPs and just keep mailing off of that setup daily?
Isn't it just assumed that you'll be on URIBL pretty quickly? I checked and every domain I've purchased is on URIBL.
Ok so to follow up, I've tried both of your methods and they seem to be working. So that's great news. It seems to be primarily because of the constant domain swap. I'm noticing that I am on URIBL within an hour or so. That's pretty crazy but it's how it is. Anyway to avoid this or is it just going to happen? I like pushsend's method of changing the domain on the entire /27 everytime but unfortunately on my platform it is VERY tedious to "un-assign" domains/IPs as you need to delete each assignment one at a time and it is not fast. That being said, if I use mx10's method, will my delivery drop? I'd probably use the same # of domains per day so it doesn't bother me either way. Pushsend - My only question about your method involves the RDNS. My platform shows you whether rdns is successfully setup after you setup rdns with your host. When I use a .com as the "permanent" rdns, it always shows on my platform that the rdns is not properly setup when mailing with a .info. I'm assuming this is because it's looking for the rdns to be mail.domain.info (example), but it sees the rdns as domain.com. So even though my platform doesn't show it is right, as far as mail servers receiving the mail, do they see it as having rdns? Just asking so delivery is as good as it can be. When I check rdns using random dns sites, it shows the .com so it's working, it's just that I was always under the impression that it had to match the mailing domain since my platform gets pissed if it doesn't.
Ok, so this sounds like strictly a GI thing. I'm just debating whether it is more effective to go 1 IP at a time like mx10 said to do. Definitely easier of course, I'm just wondering how that IP would perform the second time you used it a week later for example. Any info on that?
