Mobile Click Loss (30)

Hey Guys,

I recently shifted my focus from Web to Mobile. I have always used CPVLab for tracking web campaigns, and it has always served me well. I started testing some mobile offers, and one of the first things I am noticing was that my traffic source / CPVlab click stats seem to match up, however, at the CPA network, those click stats are off by 60-70%. For example, yesterday on one of my campaigns I was testing, I see 550 clicks at the traffic source, and about 540 on CPVLab so that is fine. But at the CPA network, I only see 130 clicks registered. Now keep in mind this is a Direct Link campaign, so there is no CTR from LP to Offer to consider. Is the redirect loop from the tracker to the offer causing this huge click loss or is there some other explanation here?

That means clicks are not being redirected fast enough.

Two reasons I can think of :

1 - You have cloaking on or SSL that slows down the redirects.
2 - Your server is not fast enough to serve pages. Keep in mind mobile speeds are lower than web.

One more scenario is that of duplicate clicks. I think it would be combination of above 3 points.

Anywhere from 5-20% click loss is considered normal for mobile.

Also If your running a lot of international traffic and your server is located in the US it will cause latency / slow redirects, the best way to solve this is to get a server hosted pretty central to the geos your targeting.

Guys thanks for your response: So per what you guys have said it could be one of 3 things:

1) 1 - You have cloaking on or SSL that slows down the redirects.

Well CPVLab does use a Double Metatag refresh to hide the referrer. But thats seem important to have, and I use it on Web offers without issue. I could understand a bit of abandon but 75%?

2) Your server is not fast enough to serve pages. Keep in mind mobile speeds are lower than web.

I dont know if this could be it,not using any type of shard hosting or anything, acutally the tracking is done on a VPS server, using Beyond Hosting.

3) One more scenario is that of duplicate clicks.

I could understand but it seem hard to see such a huge discpeancy from just that. But who know?

In any case, I'm still a bit stuck as far as the solution to this. Anyone using CPVLab without issue on Mobile? Or any other suggestions?

Originally Posted by vp5005

Guys thanks for your response: So per what you guys have said it could be one of 3 things:

1) 1 - You have cloaking on or SSL that slows down the redirects.

Well CPVLab does use a Double Metatag refresh to hide the referrer. But thats seem important to have, and I use it on Web offers without issue. I could understand a bit of abandon but 75%?

also depends on devices and carriers. Features phones have more lost clicks compared to iOS and Android. Wifi has less lost clicks compared to mobile carriers

Hey, Thanks man for the response. I was considering Voluum, but aside from the fact that I am used to CPVLab, I just dont like that fact that I cant install it on my server, its a hosted tracker. And I didnt know that DMR causing such an abandon, I emailed Beyond Hosting, I am on their Apache VPS platform, they told me that upgrading to Litespeed would improve the redirect speeds by 50% or so. So I may try that. But as far as the redirect, Are you guys comfortable not hiding the referrer on Mobile campaigns?

Any chance you could try and NOT wipe the referrer? You'd see more profit.

Remember that the referrer will be passed to your CPA network which will have an SSL tracker (most cases), and will be wiped in other redirects to the final advertiser. Your network is not going to pass juicy data on. To prevent campaign stealing (if your CPA network reps are budding mobile entrepreneurs), you should include a quick php referrer check on your LPs with logic like, "if referrer is not [admoda pattern; decisive pattern; etc.] then show super compliant page".

Spy on the competish, see who's using S/DMR, and who's going clean.

I gotta tell you, some DMR can take many seconds on a desktop computer to truly wipe the referrer. Mobile browsers are funny sometimes too.

Yeah guys, I got a message back from Beyond Hosting on this issue. This is what they wrote:

I can tell you now the issues are clearly the other domains not on your server. These other items literally take up to 4 seconds to load which is incredibly slow.
The redirect occurs in 307 milliseconds and 284 milliseconds from our VPS - this is incredibly fast. Now as soon as it hits and begins loading that servers content it is in the 1-2 second range which is slow. This is already after your redirect has occured and is out of CPV Labs hands at this point. What all this means is, the server hosting the offer is running really poorly or has really poorly coded content. This is your issue and certainly is going to be really bad for your campaign to get any kind of visitors to complete through all of this waiting.

Do you guys think this is related to this offer only or is that typical running mobile?

If that's the case ( I would double check ) I would just move to another offer.

1. For god sakes get rid of any meta refreshes. I don't know why anyone uses them especially for mobile. Get rid of SSL unless your server is very close to the users. Understand that establishing an SSL connection can afford you referrer cloaking but can double-quadruple latency.

2. The network and advertiser redirects/landing pages are going to account for most of the delay... its pathetic really and I wish something like Cake would push for global performance. Lots of bigger advertisers will have a smarter solution setup but small advertisers might do it all via a US-based server, which isn't great if its an offer being pushed in e.g. Australia.

3. If you're running stuff outside the US, chances are any tracking system/server located closer to the users is going to perform better (within reason), regardless of whether it's a VPS or dedicated server. In this respect, not using e.g. Voluum (which has multiple datacentres) because it's not self-hosted is idiocy. Deploy a tracking system closer to your users, up your game or throw away such ideals. You're literally throwing away money if you avoid a more efficient (read: less click loss) system because you want it to be self-hosted, or you want to cloak everything, and so on.

We should petition Cake Marketing to be transparent about their network structure and expand to offer better global latencies. It's pretty depressing if you can track > redirect > load a landing page in the time it takes Cake to finally pass the offer page URL to some users browser.

We use DMR to protect our IP. we don't want even the network to see our lps. Dmr does have a significant performance hit.

If anyone has a more effective way to blank out info and avoid anyone looking at our lps, we're listening

I would be more inclined to do a normal redirect to an arbitrary domain (e.g. hosted on server in same datacentre), perhaps with SSL, then to the network.

You could jump through multiple domains/server with a normal PHP redirect and just pass querystrings. It will be better than a meta refresh that requires the browser to load the page DOM before redirecting.

Better yet, you could do this with a server running CloudFlare's Railgun or similar to establish SSL 'at the edge', making the performance hit from SSL much much lower.

Ok, for those with more experience using Voluum: Would using the redirect instead of dmr in Voluum wipe out the referrer and make the lp untraceable? Or would you still need to direct traffic to another domain or yours and redirect once again to the network?

Originally Posted by zeno

1. For god sakes get rid of any meta refreshes. I don't know why anyone uses them especially for mobile. Get rid of SSL unless your server is very close to the users. Understand that establishing an SSL connection can afford you referrer cloaking but can double-quadruple latency.

2. The network and advertiser redirects/landing pages are going to account for most of the delay... its pathetic really and I wish something like Cake would push for global performance. Lots of bigger advertisers will have a smarter solution setup but small advertisers might do it all via a US-based server, which isn't great if its an offer being pushed in e.g. Australia.

3. If you're running stuff outside the US, chances are any tracking system/server located closer to the users is going to perform better (within reason), regardless of whether it's a VPS or dedicated server. In this respect, not using e.g. Voluum (which has multiple datacentres) because it's not self-hosted is idiocy. Deploy a tracking system closer to your users, up your game or throw away such ideals. You're literally throwing away money if you avoid a more efficient (read: less click loss) system because you want it to be self-hosted, or you want to cloak everything, and so on.

We should petition Cake Marketing to be transparent about their network structure and expand to offer better global latencies. It's pretty depressing if you can track > redirect > load a landing page in the time it takes Cake to finally pass the offer page URL to some users browser.

I have the same problem.
Direct linking with Voluum.
Brazil offer.
Redirect mode 302

Airpush clicks 5390
Voluum clicks 5341 (3539 different IP's)
Aff network clicks 1244

My guess is the duplicate clicks... we show about 50% of the clicks from Voluum in Aff Network (Voluum: 24k clicks, Cake: 11k clicks) but when clicking on reports » include duplicate clicks... that number goes from 11k up to 21k... and yes we have conversions from clicks marked as "duplicate clicks"... but why does the platform treat clicks from the same ip as duplicate? We should have the option to have it reported that way or not because it means that our EPC's on the network are also incorrect.

With aff network "duplicate click's" enabled its 3454 clicks

Still almost 2000 clicks less than from Voluum or Airpush

Originally Posted by afilia320

Ok, for those with more experience using Voluum: Would using the redirect instead of dmr in Voluum wipe out the referrer and make the lp untraceable? Or would you still need to direct traffic to another domain or yours and redirect once again to the network?

Although you could use https with Voluum to stop the referrer passing.

To do this you just need to use your default xxxxx.trackvoluum.com domain, at least for your CTA link. Your custom (CNAMEd) domains won't work with SSL.

As for the whole duplicate click things in Cake... doing it by IP is the easiest way. Who cares about EPCs? Revenue is more important. EPCs at your tracker or traffic source are usually more useful. They could do duplicates by some sort of 'fingerprints' but as long as duplicate clicks can still convert, i.e. the same IP can convert multiple times, it's not that important.

Adding SSL to the cta does not wipe out the referrer (as the Voluum/click SSL link is a 302 itself). You'll need your Lander to be SSL and then the wipe out works.

Originally Posted by afilia320

Adding SSL to the cta does not wipe out the referrer (as the Voluum/click SSL link is a 302 itself). You'll need your Lander to be SSL and then the wipe out works.

Correct, I did fiddler trace and 302 does not wipe the referrer when the only ssl in the path is 302. Tried this on Chrome and IE. No idea how mobile Browsers would behave but I'd say it's not reliable. ..

One more point to check - is your affiliate network detecting a bunch of those clicks as coming from the wrong country and thus geo-redirecting them?

That can be a real problem with some offer / geo / browser combinations. Opera Mobile is a particularly significant culprit.

FYI : Anyways I did a few things to reduce the click loss and now am down to about 20% which seems to be within the normal range.

1) Changed CPVLab setting from : DMR to Direct Redirect

2) Upgraded Beyond VPS Hosting from Apache >>> Litespeed

It seems that all I can do from my end, The advertisers server is slow which seems to be contrubiting to most of the click loss now.

Since I am Direct Linking this campaign currently, I think adding a properly hosted LP in between, I should be able to reduce the Click Loss even further, hoping to get down to 10% or so, that I can live with.

Hello...

I am curious about this line in the red...

Does it really work?
A simple php redirect like this:

Originally Posted by zeno

I would be more inclined to do a normal redirect to an arbitrary domain (e.g. hosted on server in same datacentre), perhaps with SSL, then to the network.

You could jump through multiple domains/server with a normal PHP redirect and just pass querystrings. It will be better than a meta refresh that requires the browser to load the page DOM before redirecting.

Better yet, you could do this with a server running CloudFlare's Railgun or similar to establish SSL 'at the edge', making the performance hit from SSL much much lower.

It should be the last domain in the chain that was loaded via http.

@zeno

well... that really should be faster than DMR right? because DMR will keep on reloading until the referrer is blanked...

Also... with the php redirect code I have given above.. will that also pass the query strings? from lander to domain1 to domain2 to offer?

like ?s1=campaign1&s2={clickid}

No, it won't - that doesn't happen automatically, so if you redirect to some URL it just goes there and drops everything.

To pass on the parameters, change the code to:

<?php
header( 'Location: http://www.redirect_to.com/new_page.html'.$_SERVER['QUERY_STRING'] ) ;
?>