Affiliate clicks vs Google Analytics data

[phoenixmac]

I run an online business and ran an affiliate campaign with a company where I would pay them based on clicks they generated from featuring my business.
I had to pay them for 58 clicks, my GA says there were 72 page views (which they said is the closest match in GA to compare), however these 58 clicks I paid for came from only 16 unique users. Is this a normal discrepancy to see, is this possible?

 

[Graybeard]

WELCOME ABOARD ...

━━━●──────────────
The agreement should be specific to either:

• Unique referrals per day
• Unique referrals per time period --like; one calendar month or year quarter

Google-Analytics is not that accurate. A year ago I sent ad network traffic to one domain, recorded the clicks out to my second domain in my own tracking logs, there was a loss at domain 2 this test is about 2 years old

I tested (for my own benefit too) -- accessing a GA tacked page on (one of) my servers, with both my own ISP and VPN (in Canada) with both Firefox and Chrome. My Firefox blocks G-A and other 3rd party trackers. 2024-01-29 09:10~ today **added

Only the Chrome was tracked by a cookie ISP or VPN did not matter --not conclusive.

 

[Graybeard]

What is a *click* a visit or a unique referral?

If you are going to use Google-Analytics is the final arbitrator you need to agree on terms.
This is why PPC and CPC costs are never as the advertiser says because of variables like these.

If its 90% or better in agreement with your data fact set you are doing well IMHO.

 

[phoenixmac]

Thank you very much for your insights @Graybeard! I appreciate it. That is what I thought also, but the discrepancy is over 25% which seems abnormal. My main concern is also around how there were only 16 unique users yet we had to pay for 58 clicks - how can I trust that their advertiser wasn't just clicking on our link over and over in order to obtain more money from us (since we paid per link click)? What do you think, is this a lost cause and we should give up the battle, or is there a way we can challenge them to prove they were driving up revenue with scam clicks?

 

[Graybeard]

1. and how do they come up with 58?
2. can the prove that 58 with any real data on each click --what data source? where is the log?
3. what is the purpose of paying per click and is there some immediate conversion revenue possible for you?
4. are you paying a premium over network DSP type traffic you could buy?

 

[phoenixmac]

1. and how do they come up with 58?
2. can the prove that 58 with any real data on each click --what data source? where is the log?
3. what is the purpose of paying per click and is there some immediate conversion revenue possible for you?
4. are you paying a premium over network DSP type traffic you could buy?

1. Via their own click tracking software (I am unsure of the name of the one they use, but you may know the company this was through as they are fairly well known - named Linkby)
2. The log is just published in a day-by-day breakdown on their performance/analytics tab of the campaign on their website, on our dashboard
3. The purpose is they work with some of the biggest publisher names (Vice, Buzzfeed, Marie Claire, etc) so there is a hope and assumption these clicks would drive conversion revenue, however unfortunately this wasn't the case for us
4. We were paying £1.50 per click (about $2 USD)

 

[Graybeard]

4. We were paying £1.50 per click (about $2 USD)

I would guestimate after 200 to 500 clicks you would have a reliable conversion ratio. under 100 is not statistically meaningful.

however these 58 clicks I paid for came from only 16 unique users

Were they legitimate users or clickbots?
If they were repeat visits then it might be a good thing.

What would Bing Ads or Google Ads Cost (SEM)?
Would that be competitive?
Could those SEM ads be used as a benchmark (in some way)?

All products/services are not equal in conversion costs or rates either.

 

[phoenixmac]

I would guestimate after 200 to 500 clicks you would have a reliable conversion ratio. under 100 is not statistically meaningful.

Were they legitimate users or clickbots?
If they were repeat visits then it might be a good thing.

What would Bing Ads or Google Ads Cost (SEM)?
Would that be competitive?
Could those SEM ads be used as a benchmark (in some way)?

All products/services are not equal in conversion costs or rates either.

Were they legitimate users or clickbots?
That's what I'm trying to determine — there's no way currently for me to verify that this wasn't the publisher (and potentially their employees) clicking on our link several times over to drive up the clicks. The 16 unique users to 58 clicks ratio drives up my suspicion. Currently the only thing I have to go off is their dashboard and GA.
What I'm trying to work out is if there's a way I can dispute this with them to get them to give me evidence this wasn't their publishers abusing the system collecting their own clicks. Is there even a way for me to do that, or do I just need to count this as a loss and not take the risk with this platform again?

About your other questions, I haven't run Google Ads or Bing Ads in a long time so it's difficult for me to compare, but it's a good question.

However the CPC when I was (years ago), was $0.28. Given that Google Ads don't give data on unique sessions/users and only clicks, I'm unable to compare this discrepancy.

 

[Graybeard]

What I'm trying to work out is if there's a way I can dispute this with them to get them to give me evidence this wasn't their publishers abusing the system collecting their own clicks. Is there even a way for me to do that, or do I just need to count this as a loss and not take the risk with this platform again?

Match the IP addresses in your server logs, of the clicks that you are paying for, and check the locations of the IP address "endpoints" that are repetitive --and see if there is a real pattern.

The burden of proof is on you in a dispute like this. You are the one paying ...

 

[phoenixmac]

Match the IP addresses in your server logs, of the clicks that you are paying for, and check the locations of the IP address "endpoints" that are repetitive --and see if there is a real pattern.

The burden of proof is on you in a dispute like this. You are the one paying ...

Thanks, that's good advice. I've just spent some time trawling through GA seeing if there was a way to narrow down a filter which would identify a single or two users with an abnormal amount of clicks, but realised there's no way in GA to then see if those users were returning over different dates. I'm unsure how to access IP addresses in our server logs, or if this is possible. Is this something the platform should be able to disclose to me? Any other tips appreciated!

 

[Graybeard]

Every (most servers) have (keep --record) log files.

example default locations:

they are located in /var/log/<server_name>/ as: access.log
or where the path is given in the server configuration files for each virtual domain


In order to track with server logs you need to have a logical character string to match in grep

All links to be tracked need to have identifiers.

usually that is done using query string e.g.; //url/?r=1234&s=12 code for `r` is referrer and `s` is site numbers that only you use

Toolish way with GA4 (current Google-Analytics):
GA4 will track utm_ values by default or rather easily
article: A Beginners Guide to UTM Parameters (And How to Use Them)

e.g.; //url/?utm_campaign=1234&utm_source=12 or whatever makes sense to you in your own situation.