ZeroPark visits from domain that don't even exist?!? (17)
I have a question for those who have experience with ZeroPark.
So, i started a popup RON campaign and I checked the targets that brought me visits, just wanted to check them see what's it about, and in order to see the URL of a target, you need to create a Target campaign, add the target code name there and shows its URL.
Surprise, surprise, I checked 2 targets that sent me the most "visits" and their domain names didn't even resolve the DNS. How come I can get traffic from domains that don't even load?
Found another domain which isn't even registered.
Got a reply from ZP support, saying that this pop traffic could be from adware (toolbars): "When a user types in a domain and hits enter a pop can open."
Ok, I doubt someone will type a domain that's not registered, like "cdbkmro5.com"
I'm waiting for another reply from them. Something seems fishy.
Another question, from my stats looks like I get a lot of traffic from onclickads.net which, if I'm not mistaken, belongs to propeller ads media. This means ZP is brokering pop traffic from multiple networks? I didn't know that, I thought they're an independent platform, if I may say that.
Originally Posted by mc2882
Originally Posted by mc2882
Currently my problem is I have very low CTR (0.15% on a sweeps offer, with a typical lander, intro&exit popup, etc). So I suspect some bot traffic here. I enabled logging for my cloudfront distribution and I'll check them for "suspect behavior"
Update:
Wow... I exported the logs from AWS and filtered everything in excel. From 1600 hits from ads (this excludes hits for my own resources), 600 are from duplicated IPs. Ok, might be the same user seeing the ad from different ad networks as @compound said before, but even so, how can you explain this:
40 requests from a single IP
all coming from onclickads.net
all unique zoneIDs
all of them in an interval of 2 minutes or less
Wow, that must have been a very angry user with fast-finger refreshing the page... BUT WAIT, I set a daily cap of 1/24h, so...it's not so easy to explain it now, right?
And this pattern is similar for many IPs, probably not 40 duplicated visits, but easily 10-20 visits from one IP.
I'll put together a nice report and I can't wait to talk to ZP tomorrow about this.
PS: I hope they don't read this thread meanwhile, so they can prepare a nice speech for when I contact them, lol 
Hey mc2882,
We appreciate your attention to your campaigns. If you have any targets you suspect to be of low quality or bot, please send us a report by target. This is one of the cases why tracking is mandatory on our platform, to allow us the opportunity to investigate any claims. We also have a number of our own security checks put in place to identify low quality traffic in which, thereafter are banned from our system and all advertisers are immediately refunded the spend from those sources. It's in our best interest as well as our publishers to single out any suspicious traffic and find the source.
Thanks for your reply.
It's a bit hard to match the cloudfront access logs to target names, but I'm working out on figuring how to do this. I wanted to contact you today about this, but realized it's almost weekend, you're probably getting ready for ASW anyway and I wouldn't like to waste my time writing a long email and risking getting ignored for a few days. So I'll let more data gather through the weekend and we'll talk about it next week.
Anyway, pretty disappointed for the moment, seeing about at least 1/3 of the "visits" are fake.
By the way, don't you have any algorithm for the auctions that checks for duplicate requests from same IP at the same time (I mean, I have 32 different auctions "won" in a matter of 6 seconds) and most are from duplicate zoneIDs, but also different zoneIDs.
How do you explain that? How can a user generate in a second 3 popups for my site, coming from different zoneIDs? And this patter is all over my stats.
mc2882 : there is some % of bot traffic in EVERY ad network on the planet, period! Some do it on purpose to earn more money by scamming the advertisers, thats true. But even the most transparent and whitehat network will have some % of bots. According to the latest reports I have read, between 30% - 55% (depends on the source) of all internet traffic are bots - there is no way to stay bot free, no matter how hard you try. Im not trying to defend zeropark here, just felt the need to bring this to your attention as people often forget about it.
Zeropark sells adware traffic, pops, traffic from dropped domains etc ... this isnt exactly the cream of the crop when it comes to traffic quality and thats the reason why you can get clicks there for the fraction of a penny in many cases. This source sells cheap traffic in large volumes, nothing more nothing less Its crucial to act fast with networks like this, block whats not working, bidding higher where it makes sense, babysitting the traffic and catching each bot click isnt gonna make you money, trust me
Ended up adding 600+ IPs in the ZP's blacklist settings.
After the optimizations from yesterday (banning the targets that sent about 50% of my total traffic, completely bots/fake), I got from
CTR 0.11% to CTR 0.46% (4 times higher)
ROI -62% to 142% (1st green day, yesss...)
Though, we're talking about $7.5 profit, so I won't afford a beach-house yet from this campaign, but at least I managed to filter out most of the fake/bot placements. (FeelsGoodMan)
Now let's work on the lander, 0.5% seems low for a sweeps lander, right?
Today banned a few more IPs and placements. It's surprising how much bot traffic there's in ZP. I can't wait to discuss this with them next week.
Are you running mobile traffic? If so be careful with IP blocking since there is a lot of pooling of users to a single gateway/IP by the carrier.
If the domain has no DNS records then its definitely dodgy - the only way it could send you traffic is if browser adware intercepts it and sends it somewhere (i.e. capitalising on a user making a typo).
Some combinations that look retarded can be real, e.g. if someone goes to type a common domain and their fingers are a key unit left or up from where they think they are.
E.g. fiifkw,xin if you were to right google.com all to the left by one key. Only adware will pick up that junk.
As was said earlier... if it has positive ROI, keep it. If not, block it. If it has zero conversions, a lot of traffic, and is entirely suspicious, report it.
Wow... the traffic quality decreases with the increase of bid :)
Joke of the day:
Let's see...Quick update on this. I finally had a chat on skype with someone from ZP, honestly didn't help me at all. Sent him all the logs, said "looks like these are bot/face visits", but I'll have to wait for the manager to return from ASW to look into this and have a talk with the guys from propeller (again, as I said, the traffic comes from some of their publishers).
Then, since I got the chance to talk to a ZP representative on skype, I thought will be nice to ask him a few things about bidding. Basically, I have a RON campaign where I have a bid of $0.001 and I'm on average 2nd BP. I'm 1st BP for most targets, and for some higher traffic targets I'm 2nd, 3rd, etc, so the average would come to 2nd.
The day before I already created a duplicate campaign and set the bid to $0.002, I wanted more traffic and as I've read in other threads, ZP suggested to other guys as well to raise the bid, even though they were shown as 1st BP, because of the traffic being brokered to other sources, etc. Also, ZP said that "theoretically" a higher bid will get me access to higher quality traffic targets. As expected, this campaign got just a few more visits, but lowered the number of visits I got on my initial campaign. I said it's no problem, I expected that, but at least I was getting higher quality traffic, right? We'll see about that in a sec...
Then after my discussion with that ZP representative, he confirmed me that "theoretically" with even higher bid, even though I'm 1st, I can get access to higher quality traffic and placements. So I created a 3rd campaign, duplicated settings, but with bid $0.003. In a few hours, this campaign already bought more views that the 2nd campaign running all day, and at the end of the day, it was time to check the campaign stats and the "higher quality traffic" I get from higher bid campaigns.
- C1 - initial campaign, with bid $0.001
- C2 - 2nd campaign, with bid $0.002
- C3 - 3rd campaign with bid $0.003
Here are the stats for yesterday:
C1 spent $1.76, 1784 visits, 57 clicks, 1 conversions, CTR 3.2%
C2 spent $11.95, 6013 visits, 139 clicks, 0 conversions, CTR 2.31%
C3 spent $18.39, 6185 visits, 113 clicks, 1 conversion, CTR 1.83%
I know these stats aren't very statistically significant, but one thing is obvious quite clear and statistically: C1, with the lowest bid, has 100% chances at being the best in terms of CTR. In terms of CR, can't really say now.
So, the ZP's theory of getting higher quality traffic by raising the bid makes no sense. Not to mention the fact that the new campaign still had thousands of bot visits, even though I already have like 800 IPs blocked in ZP.
As a conclusion, I'll wait for the ZP's manager or whoever is in charge for this to return from ASW and have a chat with him about this. We'll see how it goes.
This test is not at all convincing unless the placements are identical.
You're running RoN with a few thousand visits and probably thousands of placements.
It would only take a small shift in the actual placement distribution for large changes in CTR, CVR, cost, etc. for these results to occur and at face value we can't see what the placements are doing.
Theoretically, bidding higher == higher quality traffic - this makes sense. Why? Because the bids are based on a marketplace where peoples bids are related to their revenues. No one bids to run at -75% ROI. If people are making a lot of money on a placement, they will bid higher to take more traffic and as a result the highest CPM placements tend to have the highest EPVs (note: only within certain verticals/campaigns, not across the board).
On the other hand - on a per-placement basis, taking the first bidding spot usually equates to higher conversion rates on traffic but not necessarily higher ROIs.
I would test something like $0.001, $0.005 and $0.01 as staggered bids on RoN and be mindful of the placements that are actually sending the traffic - in the lower bid ones there will be a lot of junk and a lot of placements in the higher bid campaigns that you couldn't dream of getting much traffic from in the low bid campaigns.
The majority of traffic came from same targets, but for campaigns with higher bid I got more visits than the campaign with lower bid.
I know that theoretically higher bid = higher quality traffic. Practically, not really. And you know why? Because propeller (brokered through ZP) has a sh*t ton of fake traffic and I found pretty much the same amount of bot traffic for all campaigns, no matter the bid level, even after having 800 IPs blocked. You might think 800 IPs is a lot, and anyway ZP allows only 1000 entries, so I guess I'll have to block C-class IP groups now, because I have already over 1200 IPs that sent me bot traffic over the past 4-5 days.
And the problem is, this 30-50% bot traffic appears on all my campaigns, doesn't matter the bid. So, to do a simple math here:
Campaign 1: $0.001 bid, I buy traffic worth of $100, 50% bot => $50 wasted cash
Campaign 2: $0.003 bid, I buy traffic worth of $300 (to have the same number of clicks), 50% bot => $150 wasted cash.
So to buy the same amount of clicks, I would have to waste 3 times more money on bot traffic. Not really worth it.
Plus, now I'm testing the offer, landers, etc. Of course for someone that found the winner offer/lander/target combo can afford to bid 10x the amount I bid, but he knows he gets the most out of that traffic. Now, since I'm testing, why bid like crazy since I don't have any winner combos yet?
Anyway, I'll probably stay away from ZP, seems like a very bad traffic source, at least for testing. For scaling, might work, when you know for sure your lander converts, so you can factor in the % lost on bot traffic. But for testing, it's eating a lot of money and you don't even end up with some clear stats because of their fake traffic screwing everything up.
PS: if you really think I'm paranoid about this bot traffic, I can send you by PM an excel with some stats so you can see it with your eyes: literally, thousands of visits from IPs with up to 60 visits in a few seconds. And you know what's funny? They tried to fake this by using different zoneIDs and different targets (and of course, different User Agent String). I have no idea how these bots work, seem to be pretty smart, but I don't intend to bother anymore with them. I banned the targets, banned the IPs, still everyday I see 30-50% bot traffic, from other IPs. Why bother fighting with something when I can avoid it? There are plenty of other sources out there.
Well, I think this test is a bit crippled by the bot traffic - not really fair to make conclusions about legit traffic quality if half of it is bot traffic that biases everything - worse if it does so across most placements (maybe a broad bot/spy tool messing with things?).
As for Zeropark, its not a bad source (at least last time I tested it). I have no idea what you're doing or how you seem to be encountering so much bot traffic. It's not the summary of the entire source - if it was very few people would be using it, most certainly not true.
You don't need an optimised lander/offer combo to have success, I mean, no one is getting 10x EPCs just because they split-test a few landers and offers. Everything is out in the open, there are no secrets that give you 10x the conversion rates and bidding power.
Btw, is this mobile traffic? Or desktop pops? Why not try desktop direct nav traffic instead?
Yes, it's just mobile pops. I'm not sure if I had bad luck or not to get so much bot traffic on my first campaigns, but the stats don't lie. Anyway, after excluding the bot traffic, the campaign is doing pretty well, it's on green, but too little traffic (we're talking here about single-digit profit/day, lol).
Anyway, it was a good learning experience, probably not the best one, but every lesson helps.
I'm seeing good conversions and far less bot traffic from popads and popcash, so I might consider using them for now. I might try ZP in other geos, who know, maybe only a few targets in some geos send this huge bot traffic.
Hmm with mobile I would be careful with using IPs as a way of flagging bot traffic.
Carriers pool IPs... so there may be 10,000 smartphones all using the same IP at the same time, so on your end you may see 100 near simultaneous visits from the same IP with different placements, but it may be real.
It depends on the geo, the carriers, etc. In the US you might find an average of 10-20 visits from the same IP on the same day with device-based frequency capping. In Indonesia you may see 500, I can't say I have the specific experience to advise there.
If it was desktop traffic it would definitely be suspect.
Since it is mobile pops... I would move on and take it in stride. A mobile popup is quite disruptive (vs direct nav) and I am not sure whether these tend to be terrible quality because of that.
Home > The Newbie Zone > Questions and Answers