Seeking Help Big difference between purchased traffic and Analytics in PopAds

[Fatto]

Hello!

I've been buying traffic on PopAds for some time now and I've noticed that about 70% of the traffic is lost along the way. If PopAds says that they have sent me 1,000,000 visits, in Analytics they show around 300,000. What can this be due to?

Thanks!

 

[Graybeard]

1. Massive bounces
   Pop up blockers whose preloads are being counted.
   Known bots and proxies that are not validating with GA4?
2. Traffic fraud (possible but not not likely?)
3. Browsers that are blocking GA cookies and not running the GA scripts (likely)
4. You server is too slow and the popup is not loading in a few seconds so it gets closed as a nuisance (possible)
5. Combination of all of the above.

70% is a pretty bad loss, all traffic is not equal, you would need to do server forensics and study the patterns --not just take GA as the 'actual' . I have found GA4 to error as much as 30% to 40% as compared to my server2server logging in an experiment when GA4 became mandatory.

 

[Fatto]

Hello!

Thank you for replying!

I answer point by point to rule out:
1. Bounce rate: It is not relevant in my case.
Pop up blockers: Maybe, is there anything that can be done about it?
Known bots: It is my initial idea, although I don't use GA4 but Universal. I assumed that Google was detecting fake traffic or something like that, but I find it strange.
2. What exactly do you mean by traffic fraud?
3. I don't think this is it for several reasons (among them, that other scripts and cookies do run and that 70% is too much for this case), but any easy way to check it?
4. The server does not indicate any high resource usage (for what I have contracted) or downtime. In addition, if I make a visit I check that the load time is what it should be.

Regarding the last point, I don't take only Analytics, but, comparing, I can say that the Analytics data is representative. In the campaigns I run, I also receive as number of clicks a number similar to the one shown in Analytics.

 

[Fatto]

Thanks again!

Maybe this is a very newbie question but.... What good does it do me to know the IPs of the bots?

I can't block visits at source based on IPs, so even if I identify the IPs of the bots (which I think I am able to do), I have no way to prevent their purchase.

I'm not working with CPA or anything similar, I'm doing other work. I'm not looking to refer those visits somewhere outside of a campaign and keep only the good visits. I am looking, first of all, to identify the reason why there is that discrepancy between what PopAds and Analytics indicate and, in case it is bot traffic (my initial hypothesis and what all the tests I have done point to), how to avoid their purchase at source.

Right now I'm working on whitelisting and blacklisting, but any other idea is welcome.

 

[Graybeard]

Take a random sample of 2,000 IP addresses and the files shown as served
from your server's access.log file

$ cut -d'-' -f1 access.log |uniq|shuf -n 2000 >ip-test.csv
#that's assuming the 111.xxx.xx.xx -- is the first field of your log --that's the usual default format

Write a bash script [.sh] and loop the access log and grep the 2000 ip sample and look at the full request and analyze the requests. A lot depends on how you are logging (what MIME types).

You can try adscore.com and redirect the requests to your server through their platform and observe the requests that are filtered. Tool solution but that will not tell you why ...

 

[Fatto]

Hello again.

Sorry I haven't replied, I've been away for a couple of weeks, although I haven't stopped thinking about this.

From your reply there is one thing that stands out to me because it would solve the whole thing for me (I think). It is this:

• each IP from the ad network had a site_id={SITEID} in the query right?

Where/How can this information be obtained? In the hosting logs there is nothing like this

 

[@annabuying]

Hello!

I've been buying traffic on PopAds for some time now and I've noticed that about 70% of the traffic is lost along the way. If PopAds says that they have sent me 1,000,000 visits, in Analytics they show around 300,000. What can this be due to?

Thanks!

Do you use a tracker?

 

[Graybeard]

• each IP from the ad network had a site_id={SITEID} in the query right?
• so you can grep the log matching site ID
• if these IPs do not get image files they are not real users. Be aware Selenium/headless chrome based bots will request images --so that is not fool proof.
• The logic is to identify the site to blacklist based on the % of (IP) suspects --not on some vague clickid s2s postback or the network's stats.

You may need to make the sample 20,000 (widen the scope) to identify the *bad-site-id's* However, I would want to see 65% or so bad before I would blacklist --there may be conversions in that traffic too.

You are asking simple answers for very complex problems. I would buy 100CPM from another ad network and compare the results.
Same loss ±10% you are the problem --if the loss is 30% or less --then popads is the problem. That is probably the best thing to do at this point.

I use my own logs; traffic-in.log and click-out.log (to the money page from the landing page) and compare those numbers respectively, ad network and the target page.

Another thing will be the GEO and your bid. US traffic is the most expensive and the bots earn the most

You can also get the hostname for the sample IP addresses. Almost all (2)SERVERFAIL -and- (3)NXDOMAIN replies are datacenter servers and not residential IP. However, VPN's using free app backdoors are commonly used for click-bots when the price of the traffic is right. So, in the end *residential IP is not meaningful in all cases.

Also know you user-agents only the current ones (last two versions) are *probably* human traffic. Caveat: it's super easy to forge the user-agent header --so that may not be so indicative.

 

[Graybeard]

1.If you use an ad network add the token for site id on the URL

2. access logs are usually found in these (default locations)

root@DS-11:/# find . -name "access.log"

./var/log/apache2/access.log
./var/log/nginx/access.log

You can read the logs as a user account (sometimes)
I don't know what *hosting* means in the way you are using the term --contact the host support. I use VPS servers and Standalone or dedicated servers and have root access. However, as I recall: on even shared hosting that I used years ago when just beginning, I had access to my server logs.