Are Competitors Clicking My Ads? How to Tell From IP Data
Are competitors clicking my ads? Google says it happens. How to tell from IP data who owns the network behind repeat Google Ads clicks, with live lookups.

The same address shows up in your landing-page log four times this week, always in office hours, and your nearest rival is across town. Every guide says to exclude it.
But Google's own help page says one public address can be "hundreds or thousands of devices". The networks a rival's staff most likely click from (phones, home lines, a corporate security gateway, iCloud Private Relay) show you the network's owner, never the employer.
Below: what Google says about competitor clicks and repeat IPs (pages re-read on 16 November 2026), what seven kinds of network look like in a live GeoIPHub lookup, the registry check that is the only way an IP can name a company, and a short Node script that groups your log by GCLID and network.
Can competitors click your Google Ads? Google says they do
Yes. Google names competitor clicks as invalid activity on two pages. Its About invalid traffic page, under "Other invalid user activity", includes "competitors who manually click to increase your spend". The Ad Traffic Quality home page describes "other advertisers who click on ads to waste their competitors' budgets".
What Google does about it: "Google catches most invalid traffic shortly after the ad interaction. If Google identifies an interaction as invalid before the end of the billing cycle, you don’t pay for it." Clicks it finds later come back as what Google credits after billing.
Google said in 2007, in a Click Quality team post, that its filters recognise "the vast majority of invalid clicks that come from a competitor who is repeatedly clicking on your ads". How common competitor click fraud is remains unknown: Google publishes no figure, and no vendor figure we found shows its method.
Google's invalid traffic page also gives an innocent cause for a jump in clicks: "If a competitor stops ads, your ads might show more often and receive more clicks."
Same IP clicking your ads again and again? Count GCLIDs before you count people
A repeat IP in your log is weaker evidence than it looks: some lines aren't separate paid clicks, and the rest may be different people sharing one address. Google's pages give five reasons:
- Same GCLID, charged once. Google's advertiser guide says: "Even if you find several entries with the same GCLID, you will only be charged for at most one click."
- Genuine repeat clicks are billed. "Customers may click your ad several times when, for instance, comparison shopping or conducting research" (third-party data differences). "As long as the click patterns don't fit a profile of abuse or invalid activity, however, they'll be counted".
- Two links in one ad. "You won’t be charged for more than 2 clicks per ad impression" (clicks and clickthrough rate).
- One IP, many devices. "Most networks such as universities, offices, coffee shops, and mobile carriers use Network Address Translation (NAT). NAT allows hundreds or thousands of devices to share one public IP address." (About invalid traffic)
- Carriers rotate. The advertiser guide says "most mobile carriers and some Internet Service Providers (ISP), rotate IP addresses among their users."
The IP comes from your own records; for form visitors, Google suggests you "record the IP address, the GCLID-parameter (if applicable), the referrer URL and user agent". Then treat the IP as evidence, not a verdict.
The rule: one row per distinct GCLID, then group the rows by who owns the network.
How to tell if a competitor is clicking your ads: seven network lookups
A lookup names the network operator (asn.org) and its type, and that answer usually rules a competitor out. We looked up seven addresses, one per kind of network a click log contains. You can look up any address in your log the same way; the free tier is 1,500 lookups a day with every field.
These are public infrastructure stand-ins; none clicked anyone's ad. Looked up 16 November 2026, 06:40 to 06:41 UTC, keyless.
A company's own registered block
131.107.1.1 sits in Microsoft's corporate network.
{
"ip": "131.107.1.1",
"asn": {
"asn": 3598,
"org": "MICROSOFT-CORP-AS - Microsoft Corporation",
"user_type": "none",
"connection_type": "unknown"
},
"detection": {
"is_hosting": false,
"is_relay": false,
"is_vpn": false,
"is_mobile": false
},
"scoring": {
"fraud_score": 0,
"recommended_action": "allow",
"coverage": "unobserved"
}
}
What it tells you: the network itself names the organisation. The 0 comes with coverage: "unobserved", which means no data, not clean. Microsoft stands in for a company big enough to hold its own block; most competitors aren't.
An ISP's customer pool
The second is a Comcast home broadband line, shown as the network only:
- Address: one in Comcast's 73.162.0.0/16 (RDAP name BAYAREA-CPE-43, registrant Comcast IP Services)
asn.asn: 33651asn.org: "CMCS - Comcast Cable Communications, LLC"asn.user_type: "none";asn.connection_type: "unknown"detection.is_hosting,is_relay,is_vpn,is_mobile: all false;threat.is_cgnat: falsescoring:fraud_score0,recommended_action"allow",coverage"unobserved"
What it tells you: asn.org and RDAP both name Comcast. BAYAREA-CPE-43 is a customer-premises pool, so the registry names the ISP, never the customer.
A mobile carrier
The third is a T-Mobile USA phone address, also network only:
- Address: one in T-Mobile USA's 172.32.0.0/11 (RDAP TMO9, 172.32.0.0 to 172.63.255.255)
asn.asn: 21928asn.org: "T-MOBILE-AS21928 - T-Mobile USA, Inc."asn.user_type: "cellular";asn.connection_type: "mobile";asn.carrier: "T-Mobile USA"detection.is_mobile: truethreat.is_cgnat: true;threat.cgnat.type: "mobile";threat.cgnat.evidence: "mobile_carrier"scoring:fraud_score0,recommended_action"allow"
What it tells you: threat.is_cgnat: true with cgnat.type: "mobile" means many subscribers can share the address, so repeat clicks from it prove nothing. That is why a shared address never earns an exclusion slot.
A corporate security gateway: labelled business, shared by many companies
165.225.72.1 is a Zscaler egress address in Frankfurt, and the trap in this list.
{
"ip": "165.225.72.1",
"asn": {
"asn": 62044,
"org": "ZSCALER-EMEA - Zscaler Switzerland GmbH",
"user_type": "business",
"connection_type": "datacenter"
},
"detection": {
"is_hosting": true,
"is_relay": true,
"relay_provider": "Zscaler",
"relay_type": "enterprise_gateway"
},
"scoring": {
"fraud_score": 24,
"recommended_action": "allow"
}
}
It came back user_type: "business". That label sounds like an office, but our API gives it only to corporate security gateways and business VPNs; Microsoft's own block above returned "none". So business never means a rival's office, and relay_type: "enterprise_gateway" tells you what the address really is: a gateway shared by every company that routes through it.
Zscaler's own published range file lists 165.225.72.0/22 under "city : Frankfurt IV" (fetched 06:51 UTC). RDAP names the registrant as "Zscaler, Inc.", not any customer.
Staff of any company that routes web traffic through Zscaler can leave from this address, your rival's included, and yours if you use Zscaler. Excluding it hides your ads from all of them. Our relay and VPN detection also loads egress lists for other gateway vendors, such as Netskope and Cisco Umbrella, but doesn't cover every gateway.

A cloud server
159.89.100.1 is a DigitalOcean server.
{
"ip": "159.89.100.1",
"asn": {
"asn": 14061,
"org": "DIGITALOCEAN-ASN - DigitalOcean, LLC",
"user_type": "hosting",
"connection_type": "datacenter"
},
"detection": {
"is_hosting": true,
"hosting_provider": "DigitalOcean",
"hosting_type": "cloud"
},
"scoring": {
"fraud_score": 15,
"recommended_action": "allow"
}
}
What it tells you: a server, not a person at a desk. Clicks from it are a bot question, not a competitor one.
A VPN
185.213.155.73 is a Mullvad VPN server in Frankfurt.
{
"ip": "185.213.155.73",
"asn": {
"asn": 39351,
"org": "ESAB-AS - 31173 Services AB",
"user_type": "hosting"
},
"detection": {
"is_vpn": true,
"vpn_provider": "Mullvad",
"vpn_type": "consumer",
"vpn_evidence": "provider_list",
"is_hosting": true
},
"scoring": {
"fraud_score": 70,
"recommended_action": "step_up"
}
}
What it tells you: the IP names the VPN, not the clicker. A rival on a VPN looks like every other Mullvad user.
iCloud Private Relay
104.28.46.99 is an iCloud Private Relay egress on Cloudflare.
{
"ip": "104.28.46.99",
"asn": {
"asn": 13335,
"org": "CLOUDFLARENET - Cloudflare, Inc.",
"user_type": "consumer_privacy_network",
"connection_type": "datacenter"
},
"detection": {
"is_relay": true,
"relay_provider": "iCloud Private Relay",
"relay_type": "consumer_relay",
"is_hosting": true
},
"scoring": {
"fraud_score": 24,
"recommended_action": "allow"
}
}
What it tells you: one Apple egress shared by many Safari users (how Private Relay egress works). It is also flagged is_hosting: true, so check is_relay before is_hosting.
Can an IP address identify a competitor's company?
An IP address can name a competitor's company only when it sits in a block registered to that company, and the registry (RDAP), not the lookup's network name, confirms it. Our RDAP queries through rdap.org at 06:51 UTC:
- 131.107.1.1:
NET-131-107-0-0-1, 131.107.0.0 to 131.107.255.255, registrant "Microsoft Corporation". - 165.225.72.1:
NET-165-225-72-0-1(ZSCALER-FRA4), 165.225.72.0 to 165.225.72.255, registrant "Zscaler, Inc.".
One names the company. The other names the gateway vendor, never its customers. See who owns an IP address: ASN and RDAP.
Most small competitors can't be found this way. ARIN's policy manual requires each IPv4 reassignment "containing a /29 or more addresses" to be registered with the customer's name, but: "To maintain the privacy of their residential customers, an organization with downstream residential customers holding /29 and larger blocks may substitute that organization’s name for the customer’s name". So a business on a dynamic line or a block smaller than a /29 usually shows only its ISP. That is ARIN's policy; RIPE NCC, APNIC, LACNIC and AFRINIC set their own.
Auction insights doesn't fill the gap. Google says the report "lets you compare your performance with other advertisers who are participating in the same auctions that you are." Nothing in it links a competitor to a click or an IP. As our investigation guide concludes, you can't find the person behind an IP address from public data.
Nor can you capture a rival's IP from their website or emails: both usually lead to a host's or mail provider's servers.
Group your click log by GCLID and network
The script below reads a CSV log with click_time_utc, gclid and client_ip, keeps one row per distinct GCLID, looks up each unique IP once, and checks RDAP only for networks with no shared, anonymizer or hosting flag.
import { readFileSync, writeFileSync, existsSync, mkdirSync } from "node:fs";
import { setDefaultAutoSelectFamilyAttemptTimeout } from "node:net";
setDefaultAutoSelectFamilyAttemptTimeout(3000);
const [logPath, cacheDir = "./lookups"] = process.argv.slice(2);
const MIN_CLICKS = Number(process.env.MIN_CLICKS ?? 3);
const MIN_DAYS = Number(process.env.MIN_DAYS ?? 2);
const offline = process.env.OFFLINE === "1";
mkdirSync(cacheDir, { recursive: true });
const [header, ...lines] = readFileSync(logPath, "utf8").trim().split(/\r?\n/);
const cols = header.split(",");
const rows = lines.map((l) => Object.fromEntries(l.split(",").map((v, i) => [cols[i], v.trim()])));
const clicks = new Map();
for (const r of rows) if (r.gclid && !clicks.has(r.gclid)) clicks.set(r.gclid, r);
const n = (k, word) => `${k} ${word}${k === 1 ? "" : "s"}`;
const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
async function cached(file, url, headers = {}) {
const path = `${cacheDir}/${file}`;
if (existsSync(path)) return JSON.parse(readFileSync(path, "utf8"));
if (offline) throw new Error(`no cached copy of ${file}`);
await sleep(2500);
const res = await fetch(url, { headers });
if (!res.ok) throw new Error(`${url} answered ${res.status}`);
const body = await res.json();
writeFileSync(path, JSON.stringify(body));
return body;
}
const key = process.env.GEOIPHUB_API_KEY ? { "x-api-key": process.env.GEOIPHUB_API_KEY } : {};
const lookup = (ip) => cached(`${ip}.json`, `https://api.geoiphub.com/v1/lookup/${ip}`, key);
const rdap = (ip) => cached(`rdap-${ip}.json`, `https://rdap.org/ip/${ip}`, { "user-agent": "click-log-check/1.0", accept: "application/rdap+json" });
function classify(j) {
const d = j.detection, t = j.threat;
if (d.is_relay) return `shared: ${d.relay_provider} (${d.relay_type}), never exclude`;
if (d.is_tor) return "anonymizer: Tor exit, names nobody";
if (d.is_proxy) return "anonymizer: proxy, names nobody";
if (d.is_vpn) return `anonymizer: VPN ${d.vpn_provider}, names the VPN, not the clicker`;
if (t.is_cgnat || d.is_mobile) return "shared: carrier network, never exclude";
if (d.is_hosting) return `server: ${d.hosting_provider}, a bot question, not a competitor one`;
return null;
}
function registrant(r) {
const e = (r.entities ?? []).find((x) => (x.roles ?? []).includes("registrant")) ?? r.entities?.[0];
const fn = e?.vcardArray?.[1]?.find((x) => x[0] === "fn");
return fn ? fn[3] : e?.handle ?? "unknown";
}
const groups = new Map();
for (const ip of new Set([...clicks.values()].map((c) => c.client_ip))) {
const j = await lookup(ip);
const label = classify(j);
let id, name;
if (label) {
id = `${label}|AS${j.asn.asn}`;
name = `AS${j.asn.asn} ${j.asn.org}: ${label}`;
} else {
const r = await rdap(ip);
id = `rdap|${r.handle}`;
name = `AS${j.asn.asn} ${j.asn.org}: no shared or anonymizer flag, RDAP ${r.handle} ${r.startAddress} to ${r.endAddress}, registrant "${registrant(r)}"`;
}
const g = groups.get(id) ?? { name, fixed: !label, ips: new Set(), gclids: new Set(), days: new Set() };
for (const c of clicks.values()) if (c.client_ip === ip) {
g.ips.add(ip);
g.gclids.add(c.gclid);
g.days.add(c.click_time_utc.slice(0, 10));
}
groups.set(id, g);
}
console.log(`${rows.length} log lines, ${clicks.size} distinct GCLIDs, ${groups.size} networks`);
for (const g of [...groups.values()].sort((a, b) => b.gclids.size - a.gclids.size)) {
const enough = g.gclids.size >= MIN_CLICKS && g.days.size >= MIN_DAYS;
const verdict = !g.fixed ? "not a candidate" : enough ? "CANDIDATE: compare the registrant with the competitor" : "watch";
console.log(`\n${g.name}\n ${n(g.gclids.size, "GCLID")} on ${n(g.days.size, "day")} from ${n(g.ips.size, "IP")}: ${verdict}`);
}
Save it as group-clicks.mjs and run it:
node group-clicks.mjs clicks.csv ./lookups
OFFLINE=1 node group-clicks.mjs clicks.csv ./lookups
MIN_CLICKS=5 MIN_DAYS=3 GEOIPHUB_API_KEY=your_key node group-clicks.mjs clicks.csv ./lookups
Unflagged addresses group by RDAP handle, so one registered block is one group.
Synthetic log; addresses are the public infrastructure from the lookups above:
click_time_utc,gclid,client_ip,keyword
2026-11-09T09:12:04Z,SYNTH-GCLID-01,131.107.1.1,emergency plumber
2026-11-09T09:12:31Z,SYNTH-GCLID-01,131.107.1.1,emergency plumber
2026-11-09T14:40:10Z,SYNTH-GCLID-02,131.107.1.1,boiler repair
2026-11-10T10:05:55Z,SYNTH-GCLID-03,131.107.1.1,emergency plumber
2026-11-11T11:20:02Z,SYNTH-GCLID-04,131.107.1.1,emergency plumber
2026-11-09T08:01:00Z,SYNTH-GCLID-05,165.225.72.1,boiler repair
2026-11-09T12:30:00Z,SYNTH-GCLID-06,165.225.72.1,emergency plumber
2026-11-10T09:45:00Z,SYNTH-GCLID-07,165.225.72.1,boiler repair
2026-11-10T15:10:00Z,SYNTH-GCLID-08,165.225.72.1,emergency plumber
2026-11-11T10:00:00Z,SYNTH-GCLID-09,165.225.72.1,boiler repair
2026-11-10T20:14:00Z,SYNTH-GCLID-10,104.28.46.99,emergency plumber
2026-11-11T21:02:00Z,SYNTH-GCLID-11,104.28.46.99,emergency plumber
2026-11-10T13:13:00Z,SYNTH-GCLID-12,185.213.155.73,boiler repair
2026-11-11T13:15:00Z,SYNTH-GCLID-13,185.213.155.73,boiler repair
2026-11-11T03:00:00Z,SYNTH-GCLID-14,159.89.100.1,emergency plumber
2026-11-11T03:00:02Z,,159.89.100.1,
2026-11-11T04:00:00Z,SYNTH-GCLID-15,185.220.101.1,boiler repair
Output, run under Node 24.20.0 on 16 November 2026 against lookups saved at 06:40 to 06:41 UTC:
17 log lines, 15 distinct GCLIDs, 6 networks
AS62044 ZSCALER-EMEA - Zscaler Switzerland GmbH: shared: Zscaler (enterprise_gateway), never exclude
5 GCLIDs on 3 days from 1 IP: not a candidate
AS3598 MICROSOFT-CORP-AS - Microsoft Corporation: no shared or anonymizer flag, RDAP NET-131-107-0-0-1 131.107.0.0 to 131.107.255.255, registrant "Microsoft Corporation"
4 GCLIDs on 3 days from 1 IP: CANDIDATE: compare the registrant with the competitor
AS13335 CLOUDFLARENET - Cloudflare, Inc.: shared: iCloud Private Relay (consumer_relay), never exclude
2 GCLIDs on 2 days from 1 IP: not a candidate
AS39351 ESAB-AS - 31173 Services AB: anonymizer: VPN Mullvad, names the VPN, not the clicker
2 GCLIDs on 2 days from 1 IP: not a candidate
AS14061 DIGITALOCEAN-ASN - DigitalOcean, LLC: server: DigitalOcean, a bot question, not a competitor one
1 GCLID on 1 day from 1 IP: not a candidate
AS60729 TORSERVERS-NET - Stiftung Erneuerbare Freiheit: anonymizer: Tor exit, names nobody
1 GCLID on 1 day from 1 IP: not a candidate
Zscaler has the most clicks, 5 GCLIDs on 3 days, and is still not a candidate: the lookup flags it as enterprise_gateway, so the script skips RDAP for it. Microsoft passes because no flag is set and it has 4 GCLIDs on 3 days. The script doesn't judge the registrant; you do. "CANDIDATE" never means "exclude"; it means compare the registrant with the competitor's legal name and check the block is static.
Notes from our runs:
- Limits. Each request takes one address, so the script caches responses and waits 2.5 s between new ones (the anonymous limit is 30 a minute). rdap.org refused Node's default User-Agent with a 403, so it sends its own.
- GCLIDs only. With auto-tagging on, "only the log entries with a GCLID parameter are clicks from Google Ads". Rows carrying
wbraidorgbraidinstead (how iOS click IDs differ) are skipped too. - Our thresholds. 3 GCLIDs on 2 days are ours, not Google's; set
MIN_CLICKSandMIN_DAYS. - Weak spots. The parser splits on commas (no quoted fields), and a lookup describes the address now, not at the click. A business VPN (
vpn_type: "business") can be a company's own gateway; the script still files it as an anonymizer, so check those rows by hand.
When excluding a competitor's IP range is worth it
To stop competitors clicking your ads from their office, exclude their range, but only when all four are true:
- The RDAP registrant is the competitor itself.
- The block is static: a company allocation or assignment, not a customer-premises pool.
- No relay, carrier, VPN or hosting flag is set.
- The script shows it against several distinct GCLIDs on more than one day.
Google's own exclusion example is your company, not a rival: "people working at your company regularly check your website's organic search results", so "you decide to exclude your company's network IP address." Exclude your own office and your agency's first: your own team is a repeat clicker you can confirm.
Even a registered office block is NAT for everyone inside, visitors included; exclude it only because nobody outside that company sits behind it.
Then see how to enter a range and which addresses earn one of your 500 slots.
What to do when the IP names nobody
When the registrant is an ISP, a carrier, a relay operator, a VPN or a security vendor, the IP has told you everything it can, and it isn't a name. Google's process comes next, not an exclusion:
- Ask Google to investigate. Send the pattern, with GCLIDs and dates, through Google's invalid traffic investigation request within its 60-day window.
- Score visits, not addresses. Judge each landing-page visit on its lookup flags, as our click fraud by IP guide shows with a hosting click and a Tor click.
- Don't click back. Clicking a rival's ads is the same invalid activity you suspect them of.
The address that clicked four times this week is far more likely to be a gateway, a carrier or your own office than a rival's registered block. Now you can check which: run your log through the script, or paste one address into the IP lookup tool.
Frequently Asked Questions
Can competitors click my Google Ads?
Yes. Google lists "competitors who manually click to increase your spend" as invalid traffic, and says you don't pay for invalid clicks it catches before the billing cycle ends. Clicks it finds later can come back as credits.
Can I see the IP address of someone who clicked my Google ad?
Google's advice is to record it yourself on your own site: the IP address, the GCLID, the referrer and the user agent. Match log lines to paid clicks by GCLID, then look up each address once, for example with a GeoIPHub lookup.
Is competitor click fraud common?
Nobody publishes a reliable figure. Vendor estimates of the competitor share of click fraud range from under 1% (Polygraph) to 17% (Lunio, 2021), and neither shows its method. Google publishes no such number and says there is no industry standard for a normal invalid traffic rate.
Does auction insights show which competitor clicked my ads?
No. Auction insights shows which other advertisers compete in the same auctions. Nothing in it links a competitor to a click or an IP address.
Can I find a competitor's IP address from their website?
No. A website runs on a hosting provider's server, so its address belongs to the host, not the office. A GeoIPHub lookup of a hosting address shows the provider's network, such as DigitalOcean, with is_hosting set to true.
Can I block a competitor from seeing my ads?
Only on one network. If their clicks come from a static block registered to their company in the IP registry (RDAP), repeating against distinct GCLIDs on more than one day, that range can go on your IP exclusion list. It stops your ads on their office network, not on their phones or home lines. VPNs, iCloud Private Relay and corporate security gateways are shared, so excluding them blocks other people instead.
The team behind the GeoIPHub API
Written by the GeoIPHub team, the people who build the GeoIPHub IP intelligence API and its scanner. Drafts are prepared with AI assistance and edited by the team; API examples are real lookups against the live API unless labelled illustrative.


