My network shaved 30% of conv. How to prove it?

[Void]

Just got my weekly report from one of the mid-tier networks. Conv rates dropped 30% from last week. No changes on my end, traffic source says performance same. Feeling like they're playing with the numbers. ICYMI I always run a small % of traffic to a tracker with its own pixel just to compare. That tracker shows conversions the network never reported. Checked the usual stuff - duplicate clicks, timezone mismatches, maybe cap issues. Everything lines up. My AM just says "algorithm adjustments" which sounds like bs. Anyone else run into this? What's your go-to method for catching shaving besides just running a second tracker? It's like you need a full-time data analyst just to get paid what you're owed. Still tryna figure it out. FWIW my da tool shows their site health dropped last month, maybe they're struggling and need to keep more cash. Catch the shavers, not the bag

 

[Bounty]

I ran into this myself and found that setting up a separate, independent tracking domain helped me catch shaving. It's kinda like having a secret sinker test that the network can't mess with. Once I saw discrepancies between my tracker and theirs, I knew where the shavings were coming from. Just make sure that tracker is not connected to your main flow, so it stays unbiased lol.

 

[Amplify]

Quick thought: do u think switching to a completely different server or ISP for ur tracker might make it even harder for networks to tamper with the data? just curious if that adds any extra layer of security.

 

[Mirage]

Been doing this 13 years and found that running a server-side pixel with a unique user-agent string helps spot shaving better than client-side alone because it's harder for networks to tamper with that data without breaking the user experience

 

[Gleam]

Haha, sounds like my last update when I thought I fixed tracking but turns out I just broke it more. I once had a network dip that I couldn't quite prove but found out later it was a sneaky bot filtering out my conversions. Took me ages to dig through logs and compare time frames but finally got some good proof.

 

[Blitz]

I think you might be mixing up the cause. My experience says when convs drop, it's usually tracking or attribution issues, not the network itself. Proving it means digging into your logs, not just claiming it.

 

[Hone]

Different angle: bot filtering could be part of it but smh, how sure are you it's not a tracking issue? Without solid logs or server data, proving it is tricky, no?

 

[Rhythm]

Use post-click analytics and compare user behavior pre and post the change, especially on landing pages. Are you tracking the same KPIs? Smh, might be the attribution window or tracking gaps causing this

 

[Tundra]

you gotta check if your post-click tracking stayed consistent. I had a similar issue and found that some of my conversion pixels got moved or stopped firing. Also, compare the attribution windows and make sure there aren't any attribution shifts happening.

 

[Boulder]

can confirm.

 

[Logic]

man 30% is huge but if you wanna really prove it, look into your ad-level data and check for any significant drops in click quality or engagement metrics before conversions tanked. Sometimes the traffic looks the same but the audience is different, which kills your EPV and conversions.

 

[Lintel]

bruh, you sure your tracking stayed solid? or maybe your traffic got shittier?

 

[Quanta]

check your attribution window, maybe it changed or your last conversions are counting different. also, run a post-click analysis to see where drop-offs happen. smh, tracking errors are the usual suspects.

 

[Void]

Thanks for the tips guys, I'll check post-click analytics and ensure my tracking pixels are firing correctly. Will also compare attribution windows to see if that's skewing the data. Appreciate the help, always trying to get more precise. Catch you later!