What's wrong? Massive bot traffic on Zeropark? Can't be, probably me (15)
What's wrong? Massive bot traffic on Zeropark? Can't be, probably me
Ok, so I ran ~500 zeropark pop impressions to a landing page, just a typical one you see floating around. I found it odd that I had zero (0!) clicks to the offer page. I tried direct linking to the offer and it worked fine. Also, when I had done domain redirects in the past, I did not have this problem. This is my first experience running pops.
So I set up a basic bot script and tested it on myself. Javascript is set to fire after 300ms. It fired when I ran it, I tested on another device, worked there too. So then I ran on like 200 or so zeropark impressions.
My logs show like a full 2/3 of impressions don't fire the 300ms script at all. Do you think this is caused by popup blockers or what? I expect some click loss but this seems excessive.
I used a completely blank landing page, that all it does is fire an ajax request when loaded:
<!DOCTYPE html>
<html>
<head></head>
<body>
<script src="https://ajax.googleapis.com/ajax/libs/jquery/1.11.3/jquery.min.js"></script>
<script>
function bot() {
$.ajax({
url: '<TRACKING LINK>',
dataType: 'jsonp',
complete: function(dataWeGotViaJsonp){
}
});
}
bot();
</script>
</body>
</html>
what ISPs are associated with the traffic. are you running desktop or mobile
Not sure of the isp, I don't have that info. All mobile. I can post my logs, it got a little bit better but still 50% or so. What's the most efficient way to post like 600 rows of results on here? I have user agent, ip address, os, and zeropark source
My last experience with Zeropark had lots of bot traffic. Can't be bothered posting screenshot though.
There will always be bot traffic when buying pops. It's just the nature of the game, and you'll have to monitor and block accordingly for every campaign.
If you check by ISP and it resolves to a datacenter (typical of bots and crawlers on a server), block that target.
Originally Posted by dotcom
Yeah but is 45% or so reasonable? They're getting through to my server just fine; they're just not loading the javascript...
Originally Posted by mrpayne
Bots are an integral part of the game.
Most of the sites will have a ratio of bot traffic to it. Your ROI for the placements will help you determine whether this is worth keeping or cutting.
These are the rules I usually follow:
1. 3x payout cuts usually already account for the bots.
2. 1-1.5x payout cuts for higher payouts
3. I may cut a placement earlier on if the CTR is way way below the average CTR of the whole campaign (ie 0.1% vs 6% CTR)
Zeropark is quite okay, just too pricey.
Alright, I ran some more tests.
First I ran directly to the offer page. Only 55% of the bought impressions made it to the offer page. I'm not sure if my network tracks with javascript or img pixel or if that's just raw click loss.
Then I ran through to my tracker, and then immediately redirected to the offer page. Only 69% of bought impressions even made it to my tracker. From there, 70% of those clicks were registered at the offer page of the affiliate network, leaving a total of 48% of the original impressions I purchased ending up at the offer.
Then I ran through to my tracker and did a javascript bot ping test, and then immediately did a javascript redirect to the offer page. Only 67% of the bought impressions made it to my tracker. On my tracker, the javascript test reported that 51% of all visits were redirected, and of those that were redirected, 95% were reported on the affiliate network's offer page.
This tells me the aff network likely isn't shaving clicks because when I track js users, nearly all of them show up on the aff network's stats.
This tells me that I am experiencing about 30-ish% click loss before anyone even gets to my server, unless zeropark is screwing with me.
This tells me that of the people who even make it to my server, only half of them have javascript enabled. Mind you, this is AU traffic so I would think they have top quality browsers.
Thus, the results of these tests tell me either my code is so buggy that it messes up a lot, or that there really is a ton of fraudulent/no-JS/bot traffic on the placements I selected. I tend to suspect the latter, because when I ran directly to the aff network, they showed the same levels of fake/nonreported traffic that I did. And my assumption is that I can trust them because they reported nearly 100% of the people I knew I redirected (in the 3rd test)
Two things I may want to try: 1) run with a pixel-firing bot check instead of js. If results are similar, that would make me even more confident bot traffic is really high. 2) Get a server in AU and see what happens. Some of the original click loss could be due to higher latency. But that wouldn't really have anything to do with the bot problem- I've already demonstrated that the biggest drop off is among people who already did load my site, meaning latency didn't matter in their case.
Can someone who is an expert verify that my thinking is correct here?
Alright, so I spent some time sticking everything on Amazon Cloudfront. My reasoning was that perhaps there was a latency issue.
On the old setup, ping tests showed that Australian traffic was taking ~300ms to hit my server. After moving to Amazon, traffic was taking ~17ms. A huge win.
However, the actual bot stats/traffic loss seems to be comparable, and perhaps even a bit worse.
I purchased ~350 zeropark AU pops. My landing page only loaded JS on about 110 of them, giving me nearly the same ratio as before. About 1/3 of pops I pay for ever actually load js on the page. WTF.
What is going on?
are you saying 350 pops? this is nothing.
Don't think or worry about bot traffic in your early stage.
Cut placements after 1000 impressions without conversion or 250 without click. If you run high payouts, double or tripple the first number
Get revenue, get traction, make money, then worry about advanced optimisation.
And as already said, 500 pops is nothing. Come back after 100,000 impressions.
Great, I won't worry for a while then.
Hi natedl98,
We do our best to filter out bots, but, as I'm sure you can imagine, there are plenty of publishers out there looking for ways to make an extra buck in not exactly the most legitimate of ways.
Our clients are almost exclusively performance marketers, so it's purely in our best interest to provide traffic that converts.
If you can send us a list of targets that you believe sent you bot traffic, we can analyze them on our end, and if the traffic indeed proves to come from bots, we'll block them and send you a refund.
Regards,
Zeropark
Home > Paid Traffic Sources > Mobile