Outsmarted by bots?? (13)
I’m following Caurmen’s simple technique to spot bots (using FunnelFlux). If you look at the printscreens below, you’ll see it doesn’t look like any of these placements are particularly bot heavy (the lower the CTR, the higher the bot rate e.g. 90% CTR = 10% bots, more or less) – this seems strange given reading around suggests that MGID is notorious for bot traffic.
Am I missing something in the data?
For context, this is around a week’s worth of traffic ($100/day) for a Tier 2 geo Skin campaign on MGID – my first native campaign 
Spend: c.$600 (I’ve tested bids ranging from 0.07 – 0.14)
Conversions: 1
Revenue: $45
My initial optimisation strategy was based on cutting high bot placements asap (in order to get some decent quality traffic to assess my banners/landers/offer) but I’m $600 deep and unable to cut a single placement based on the data.
So, based on the fact that a lot of sources suggest that MGID is FULL of bot traffic , I’ve come to the conclusion that the bots are outsmarting my filter and after reading through STM have come up with the following potential solutions to turn the tables:
- Test for scroll depth
- Load images
- Testing for cookies
- Looking at user agents
- Keeping track of bad IPs/ISPs
- Use cloaker
- Check the URLs manually assuming the referrer data isn’t blind
- Google analytics > time on site
- Forensiq.com
- Track mouse X and Y co-ordinates
- Check carrier logs
- Add an invisible, clickable link (almost but not quite the same color as the background) – I’m going with this option as it looks like it should be simple and effective
Just wanted to get a sense check from some STMers whether this is a sensible/logical approach? Or whether I'm missing something super obvious??
Thanks!
This is the code for my new bot killer page in case anyone finds it useful - I'm not super techy but it seems to work fine so far
<!DOCTYPE HTML>
<script type='text/javascript'>
function redir(){
window.location.replace('YOURTRACKERURL-ACTION1');
}
</script>
<style>
a { color:#fefefe;}
</style>
</head>
<body onload="setTimeout(redir, 300);">
<p>
<a href="YOURTRACKERURL-ACTION2">insert text</a>
</p>
</body>
<script type='text/javascript'>
function redir(){
window.location.replace('YOURTRACKERURL-ACTION1');
}
</script>
<style>
a { color:#fefefe;}
</style>
</head>
<body onload="setTimeout(redir, 300);">
<p>
<a href="YOURTRACKERURL-ACTION2">insert text</a>
</p>
</body>
You're fighting the windmills.
Forget about bot traffic.
On native two variables make or break your campaign.
First, the offer is super important.
Its tough to find an offer that is converting well on Native.
You need to test several offers that your account managers are recommending or that you see other big affiliates running.
Always ask for offers that are doing damage on Native. Its easier to make an offer fly on FB, so you need to be specific.
Second, the creatives are super important.
The creatives decide which audience go through your funnel.
If your creatives are too much clickbait or don't have commercial intent, you won't make any conversions.
Test at lest 10 different creatives, different pictures and different angels.
Copy the creatives from the key players in your vertical.
Double down on the creatives that show initial traction.
In terms of placements, hire a VA to check out the placements manually if somebody else is running nutra / skin on them. If not block.
Originally Posted by sebastian_r
You’re clearly much more experienced than me but would you really say that bot traffic isn’t a material factor? I agree with you that the offer and creatives are very important (I’m using the top offer on my network and have lifted a number of competitor creatives from Adplexity), but given how easy it is to lift ads/landers using spy tools, shouldn’t having access to converting placements be super important? I know bot placements can still convert, but I feel bots somewhat distort the stats when you’re still trying to test ads/landers/offers.
Thanks for the tip on hiring a VA to check out the placements manually – will look into that ASAP.
Originally Posted by leonardodarwin
Originally Posted by sebastian_r
Respect.
Seems you have everything you need right there: 96% CTR!
Just black any placement with CTR like that, that's ludicrous CTR...
Originally Posted by auditor
Originally Posted by mass_marketer
thank you for confirmation that it works even it is without <head>
Most browsers will load the page even if you don't have valid HTML or all tags like html, head, body included.
They simply try to render the page from what they have.
So that's why it works, but there is nothing wrong to include that head tag (which is better).
Originally Posted by mass_marketer
Placements are super important.
Only so many placements will work well with your funnel.
Those that work have a demographic that is favourable for your angel.
Or in other words, your angel speak to a good portion of the people on the site.
You need to identify placements that work well with your creatives and funnel.
So cutting out non performing placements is important, cutting out bot traffic is important as well.
However, you cannot do anything of this if you don't know if your funnel and creatives are working in general.
So here comes the manual spying in place. You use the work of others to cut bad placements based on the market environment on this placement.
Anyway, you shouldn't work your way through the bullshit yourself.
Always ask a traffic source rep for a basic blacklist or buy a list from another affiliate.
I did it all myself and it took me months to map the networks I'm working with. It's not a smart play.
Originally Posted by sebastian_r
One question I had was, could you possibly clarify what exactly you mean my checking placements manually?
I'm running on MGID at the moment and all of the placements are anonymised via an ID#, so you don't actually know where the traffic is coming from.
I took at a look on Adplexity for publishers, but even then, there doesn't seem to be a way to map the publisher back to the traffic source ID. Am I missing something?
Home > Paid Traffic Sources > Native