Self hosted tracker/Binom questions (11)
So I had the following question on
1) Do you guys use Magic Checker to detect/filter bots on pop traffic when using it with
2) This one can be applied to self-hosted trackers in general. Do you guys install one tracker instance for each continent you run traffic from or is a EU instance enough for at least US and EU?
You have damn good questions. I'm interested in any replies. I know Amazon AWS is suggested for serving landers due to their speed and geographic distribution, I'm not sure if that would help with this purpose though.
AWS S3 buckets can't host a tracker, so if you're looking to host a tracker you'll have to get a VPS from companies like Linode, Vultr, DigitalOcean, Hetzner and then install the tracker on the server either yourself or with the help of tracker's support team.
Originally Posted by regjoe
1) You can use Magic Checker to filter anything but tha't not really its primary purpose. It has amazingly good implementation on Binom. Takes a few steps to set it up, but once you have it, it's magic.
But honestly, unless you plan on cloaking: you don't need it for 99% of bot stuff. Just use regular rules within Binom to redirect shitty traffic.
2) I have Binom and all my landers hosted on the same dedicated server in DE with Hetzner. Cloudflare CDN/DNS is fast enough for everything I do. No need for separate servers. If you're itching to see the speed difference, you can always spin up a VPS on Digital Ocean for a free trial, and then like $3 a month or so unless you're running crazy high volumes to it.
Originally Posted by jaybot
BTW, does it work well in blocking spy tools, as that can be it's another use?
I too would love to just use a single server in EU, but was wondering about the cost to benefit ratio with higher % of clickloss vs using a local server in the offer geo.
Originally Posted by regjoe
Overthinking it. You're free to run tests if you like. For landers, cloudflare is the largest cdn across the world. After the first page load in any geo, it's cached. And if it's cached on cloudflare, it will be instant.
As for the few ms of extra from your tracker, I wouldn't bother. Unless you're running huge volumes on pops, it's not going to make much of a difference. For any other traffic, like Push, Social, or Native, it's negligible.
I came to the conclusion that as long as you are using a fast server and everything loads fast, you don't even need to use a CDN as the benefits are minimal. These days I'm running EU campaigns on a US server or vice versa and it works just fine. The only exception was when I was running high volumes in Asia, having a server in a local datacenter did help somewhat, but still, the difference was not that big as I thought it would.
Originally Posted by matuloo
Originally Posted by jaybot
To be fair, I was testing it with banner traffic which definitely behaves differently than POPs would, so that's something to keep in mind too.
Just to be clear, I'm not advising against using a CDN, not at all! I was just expecting it to have a massive effect on the performance of my campaigns, but it simple did not
Together with Cloudflare for CDN and DNS I use it for all campaigns worldwide.
Yes, you can see some differences between far away locations but it´s not that huge that I would create a complicated setup for it.
Use DNS service to resolve the domains faster and CDN to provide static files faster.
When the landers are coded good the differences shouldn´t be that huge to make a big difference in performance.
And for sure you could alwayy try to squeeze some faster redirects or so but then we are talking about milliseconds and this is something you don´t need to worry much about.
Important is that you use a good CDN and DNS provider, not important is if your lander loads 10 milliseconds faster on provider A compared to provider B.
1. Use Cloudflare because their DNS is fantastically fast, given their focus and huge edge network.
2. Use Cloudflare to proxy your tracker server, purely to obfuscate your real server IP. This will limit the risk of DDoS'ing and some hacking risk, since people need to go through Cloudflare to get to the server (and CF has its own security + WAF you can use if you really need it). This will also prevent traffic sources and others from banning your server IP directly, which will screw you. They can ban/flag domains but you can change easily, and no one is going to blacklist domains because they use CF nameservers (would impact half of their userbase if they did that).
3. Use Cloudflare to proxy your landing page domains for the above reasons and purely for the CDN benefit of caching static assets. It's a no-brainer.
4. Use Cloudflare because you need to use HTTPS for everything in this day and age, and their SSL negotiation is going to be faster than what your servers do (and their network here is pretty cutting-edge in terms of using every new HTTP standard that improves performance). Then end-to-end time vs going direct to your server may be slower in some cases, but globally its going to out-perform and be more stable, especially when users are far from your origin server.
5. Put your tracking server in EU West if you run generic/global traffic. I did latency tests in the past between 100's of cities and several origins. EU West gave the lowest average. Unless you focus on Asia, in which case put your server in SG.
6. Put your tracking server in EU West if you have any concerns about GDPR and storing user PII data.
7. Overall, your server being in US vs EU West, for most campaigns, is not going to make a losing campaign into a winner or vice versa. Unless you are running high volume arbitrage/pops with slim (like <10%) margins, it is probably not going to have a huge impact. We know load times affect performance and conversion rate. But most of our knowledge here is from optimising landers from 5 seconds load time to 1 second and using CDNs. It's not from testing 400 ms vs 150 ms network connectivity timings on redirects. Humans can barely perceive the difference between such small time differences in regular usage. The more important aspects are "time to glass" i.e. from click to the lander loading properly, since you are battling for attention span.
8. Don't try to install self-hosted tracking servers on multiple continents, use separate domains for them or try to load balance them... or try to geographically replicate databases and so on. Its not worth the effort unless you're highly technical and its out of hobbyist interest. Having all your data in one place and a simple approach is going to provide more value.
9. Regarding bots... my stance is... there's not a huge point in filtering/blocking them from purchased traffic. Blocking stuff that's spamming/stressing your server is another issue. But if you run pops, you are going to get bots. Who cares? At the end of the day, money is what matters. You pay $X for some zone/campaign and it earns $Y. The traffic you bought for $X could have 1% or 50% bots, but it could be profitable in either scenario. Keep in mind if bot % goes down (traffic quality up), CPMs often go up too and the marketplace can level things out. Unless you are highly sensitive to server/network costs, focus on the money side.
10. However, for stuff where you need to generate a blacklist quickly to prevent burning cash, its obvious to block bot-ridden placements early since they are very unlikely to convert and DO bleed you money until you catch them. Blocking these won't help, you need to know what they are to then stop paying for them at the source. For this, you need some sort of optimisation rule -- e.g. if X visits and no clickthroughs (based on known average CTR expectations), kill. Or if CTR is anomalously high, kill. Or if you have some bot filter and it says this placement is >50% bots, kill etc. That's really up to experience, and again you might find some bot heavy placements still convert. It could be that the placement has a lot of successful advertisers so there are always spy tools scraping the site... doesn't necessarily mean the placement will always be unprofitable.
11. As much as I want to recommend a "cloud" tracker like mine, FunnelFlux Pro, if you're running pop traffic, self-hosted trackers have an obvious advantage given the super low CPMs of visitors. Just don't dump too much time/energy into trying to make a self-hosted tracker achieve the benefits that a cloud/SaaS tracker intrinsically has.
Home > Technical & Creative Skills > Tracking Campaigns