Datacenter IP detection: why it matters for affiliate tracking
Datacenter IP detection: why it matters for affiliate tracking
Datacenter IP detection is the practice of flagging clicks that come from hosting-provider IP ranges rather than residential or mobile connections. It matters because most real consumers browse from ISP-assigned addresses, while a large share of automated traffic — scrapers, click farms, headless browsers, bot farms — originates from servers rented by the hour. If your tracker can't tell the two apart, you pay for both. Datacenter IP affiliate fraud is one of the cheapest fraud types to commit and one of the easiest to catch, which makes ignoring it hard to justify. This article covers how the detection actually works, what it catches and misses, and how to wire it into your campaign setup without breaking legitimate traffic.
What is a datacenter IP, and why does it show up in affiliate traffic?
A datacenter IP is an address allocated to a hosting company — AWS, DigitalOcean, Hetzner, OVH, Linode, and thousands of smaller providers — rather than to a consumer ISP or mobile carrier. These ranges are published and maintained by the providers themselves, which is why they can be identified reliably without any guesswork.
Legitimate traffic can come from a datacenter IP in a few narrow cases:
- Corporate VPN users routing through a company gateway hosted in a cloud provider
- Privacy-conscious users on commercial VPNs, many of which terminate in datacenters
- Some mobile carriers and ISP proxies that route traffic through hosted infrastructure
Everything else is a signal. A click from a Hetzner range on a sweepstakes offer at 3am is not a person deciding to enter their email. It's a script, a bot farm, or a click-injection setup. The question isn't whether datacenter traffic exists in your campaigns — it's what share of it is fraudulent and whether you're paying for it.
Why datacenter IP affiliate fraud costs you more than the click
Fraudulent clicks don't just burn CPC. They corrupt the data you use to make decisions. Consider what happens when a bot farm sends 4,000 clicks through your PropellerAds campaign:
- You pay the traffic source for every click, whether or not it converts
- Your CTR and click-to-conversion ratio look worse than they are, so you pause a campaign that might have worked
- Your landing page A/B test gets polluted — the losing variant may simply have received more bot traffic
- Your affiliate network sees low-quality conversions from your traffic and may throttle or ban your account
- Your optimization decisions (bids, geo targeting, creative rotation) get trained on fake signals
That fifth point is the expensive one. If your tracker reports a 2% conversion rate on a geo that's actually 0.4% once you strip bots, you'll keep pouring budget into a losing segment. The fraud isn't just the wasted spend — it's the compounding cost of bad decisions made from contaminated data.
For a broader view of how these signals fit together, see the affiliate marketer's guide to click fraud detection and prevention.
How datacenter IP detection actually works
Every click your tracker records carries the visitor's IP address. Detection compares that IP against a database of known hosting-provider ranges. The comparison happens in milliseconds, at click time, before the visitor is redirected to your offer.
The mechanics break down into three parts:
1. IP range matching
Hosting providers publish their allocated ranges via WHOIS and regional internet registries. Trackers maintain a lookup table of these ranges and refresh it periodically. When a click arrives, the tracker checks whether the IP falls inside any known datacenter block. This is deterministic — no probability involved — which is why datacenter detection has a low false-positive rate compared to heuristic signals.
2. ASN and organisation lookup
Beyond raw ranges, the tracker can resolve the IP to its Autonomous System Number and organisation name. An IP registered to "Amazon Technologies Inc." or "OVH SAS" is a clear datacenter signal even if the specific range isn't in a static list. This catches newer ranges that haven't propagated to every database yet.
3. Combined scoring
Datacenter detection is strongest when combined with other signals. A datacenter IP alone might be a VPN user. A datacenter IP plus a headless browser user agent plus 40 clicks from the same address in a minute is not a VPN user. That's why the best setups layer datacenter detection with IP velocity fraud checks and suspicious user agent analysis rather than relying on any single rule.
What datacenter IP detection catches — and what it misses
Be honest about the limits. Datacenter detection is a high-precision, medium-recall signal. It catches a lot, but it doesn't catch everything.
What it catches well:
- Click farms running from rented cloud servers
- Scrapers and crawlers using headless browsers on hosted infrastructure
- Automated click injection scripts hosted on VPS instances
- Bot farms that rotate IPs within a provider's range
- Low-effort fraud where the attacker didn't bother with residential proxies
What it misses:
- Residential proxy networks, which route through real consumer IPs
- Mobile proxy farms, which use genuine carrier IPs
- Sophisticated fraud that pays for clean residential addresses
- Some VPN users who are legitimate customers
This is why datacenter detection is a first line of defence, not the whole defence. The fraudsters who bother to use residential proxies are the ones you need velocity, uniqueness, and behavioural signals to catch. For the full picture of how much fake traffic typically slips through, see how much of your affiliate traffic is fake.
How to set up datacenter IP detection in your tracker
The exact steps depend on your tracker, but the logic is the same everywhere. Here's how it works in Adtraxo, where fraud detection is a built-in rule set on the Pro plan.
- Confirm your plan supports fraud detection. In Adtraxo, fraud detection and rules are Pro features ($27/month founding rate, rising to $47 for new accounts; existing subscribers keep $27).
- Open the fraud rules panel for the campaign or account you want to protect.
- Enable the datacenter IP rule. This flags any click whose IP resolves to a known hosting provider range.
- Decide the action. You can flag the click for review, block the redirect entirely, or exclude it from your conversion reporting. Blocking is the right choice for most campaigns — there's no reason to send a bot to your offer.
- Layer additional rules. Add IP velocity (multiple clicks from one IP in a short window), bot user agent matching, uniqueness conflict (same visitor fingerprint hitting the same offer repeatedly), and invalid referer. Each rule catches a different fraud type.
- Monitor the fraud report for the first 48 hours. You want to see which rules are firing and whether any legitimate traffic is getting caught. If a rule is too aggressive, tighten the threshold rather than disabling it.
For a step-by-step walkthrough of building a full rule set, see how to set fraud rules to protect your ad spend automatically.
Worked example: a PropellerAds campaign with 12% datacenter traffic
Suppose you're running a push campaign on PropellerAds for a sweepstakes offer in Tier 1 geos. After three days you've recorded 8,000 clicks and 96 conversions — a 1.2% conversion rate. That's borderline. You're about to kill the campaign.
Before you do, check the fraud report. Say it shows 12% of clicks came from datacenter IPs. That's 960 clicks. If those clicks converted at roughly zero (which is typical for bots), your real conversion rate on human traffic is 96 ÷ 7,040 = 1.36%. That's a meaningfully different number — and it might be enough to justify another day of testing rather than a kill.
Now flip the scenario. Suppose the datacenter share is 40%. That's not fraud creeping in — that's a compromised traffic source or a zone you shouldn't be buying. You'd block the datacenter rule, pull the sub-ID breakdown to see which zone is responsible, and either blacklist that zone or move budget elsewhere. The fraud report analysis guide covers how to read this breakdown properly.
Either way, the datacenter rule changed your decision. Without it, you'd have killed a viable campaign or scaled a broken one.
Datacenter detection by traffic source: what to expect
Different traffic sources have different baseline datacenter rates. You don't need exact numbers — you need to know what "normal" looks like for each source so you can spot anomalies.
- Push and pop networks (PropellerAds, EvaDav, RichAds): these sources tend to carry more automated traffic because the ad formats are cheap to spam. Expect higher baseline datacenter rates and watch for sudden spikes.
- Native platforms (Taboola, MGID): generally cleaner, but bot farms do target native because the CPCs are higher. A datacenter spike here is more suspicious.
- Solo ads: these are email drops, so clicks come from real inboxes — but the sender's list quality varies wildly. Datacenter traffic in a solo ad usually means the list was seeded with bots.
The point isn't to memorise numbers. It's to establish a baseline per source and treat deviations as a signal to investigate. A source that's always at 8% datacenter traffic is a known cost of doing business there. A source that jumps from 3% to 25% overnight has a problem.
Datacenter IP detection vs other fraud signals
It helps to see where datacenter detection sits relative to the other rules in your stack.
| Signal | What it catches | Best used with |
|---|---|---|
| Datacenter IP | Server-hosted bots, click farms, scrapers | User agent, velocity |
| IP velocity | Repeated clicks from one address | Datacenter IP, uniqueness |
| Bot user agent | Headless browsers, known bot strings | Datacenter IP |
| Uniqueness conflict | Same fingerprint hitting the same offer repeatedly | Velocity |
| Invalid referer | Clicks with missing or spoofed referrers | User agent |
Each rule catches a different slice. Datacenter IP is the cheapest to run and the easiest to justify, which is why it belongs first in your rule set. But a campaign protected only by datacenter detection will still leak residential-proxy fraud. For the full picture, see what click fraud is and how it affects affiliate campaigns.
Common mistakes when configuring datacenter IP rules
Three mistakes show up repeatedly:
1. Blocking instead of flagging during the first week. If you block datacenter traffic from day one without reviewing the report, you won't know what you're catching. Flag for the first 48–72 hours, review the report, then switch to blocking once you're confident the rule isn't catching legitimate VPN users.
2. Relying on datacenter detection alone. As covered above, residential proxies defeat it entirely. Layer the rules.
3. Ignoring the sub-ID breakdown. A campaign-level datacenter rate of 15% might hide one zone at 60% and the rest at 2%. Without sub-ID tracking (sub1–sub5 in Adtraxo, a Pro feature), you can't isolate the bad zone — you'd have to blacklist the whole campaign. Sub-IDs are how you find the specific placement or source that's compromised.
There's also a tracking-domain angle worth knowing: some ad networks and fraud systems flag certain tracking domains more aggressively than others. A custom tracking domain can reduce false flags and keep your links clean. See custom domain tracking: does it reduce ad network fraud flags? for the details.
How Adtraxo handles datacenter IP detection
Adtraxo's fraud detection runs at click time, before the redirect, and covers datacenter IP, IP velocity, bot user agent, uniqueness conflict, and invalid referer as separate rules you can toggle and threshold individually. Flagged clicks show up in the per-campaign analytics view with geo and device breakdown, so you can see whether the fraud is concentrated in one country, one device type, or spread evenly.
The AI Campaign Optimizer agent can also surface anomalies — if a campaign's datacenter share spikes, that's the kind of pattern it's built to flag. Combined with S2S postback conversion tracking, you get a clean picture of which clicks were real and which converted, rather than guessing from click counts alone.
If you're weighing trackers, Adtraxo fraud detection vs Voluum fraud detection walks through the differences. For context on pricing: as of October 2026, Voluum's cheapest plan is $119/month billed annually, RedTrack's Solo plan is $169/month billed monthly or $118/month billed annually, ClickMagick's Starter is $79/month billed monthly or $66/month billed annually, and BeMob has a free plan with paid plans from $49/month. Prices change, so check each vendor's page. Tools like Binom and Keitaro have their own licence pricing — check their pricing pages directly.
Frequently asked questions
Does datacenter IP detection block legitimate VPN users?
It can, which is why you should flag rather than block during the first few days and review the report. Most consumer VPNs terminate in datacenters, so a small share of legitimate traffic will look like datacenter traffic. In practice, the volume of genuine VPN conversions is usually low enough that blocking datacenter IPs is still net-positive — but check your own data before committing.
What's the difference between datacenter IP detection and residential proxy detection?
Datacenter detection flags IPs allocated to hosting providers. Residential proxy detection is harder — the traffic comes through real consumer IPs, so the IP itself looks legitimate. Catching residential proxies requires behavioural signals like IP velocity, uniqueness conflicts, and conversion patterns rather than IP classification alone.
How much datacenter traffic is normal in an affiliate campaign?
There's no universal number — it varies by traffic source, geo, and vertical. The useful approach is to establish a baseline per source and treat deviations as the signal. A source that consistently runs at a certain datacenter share is a known cost; a sudden spike is a problem worth investigating. Pull the sub-ID breakdown to see whether the spike is concentrated in one zone.
Can I use datacenter IP detection on the Adtraxo Free plan?
No. Fraud detection and rules are Pro features. The Free plan covers 10 tracking links, 5,000 recorded clicks a month, 1 custom domain, and basic stats with no card required. To run datacenter IP detection, IP velocity, bot user agent, and the other fraud rules, you need Pro at $27/month (founding rate; it will rise to $47 for new accounts, and existing subscribers keep $27). Pro includes a 30-day money-back guarantee and you can cancel any time.
What should I do when dat
Track your affiliate campaigns for free
Clicks, conversions, fraud detection — all in one tracker.
Get started free →Not ready for an account? Get the Tracking Domination Playbook — free, just your email.