On this page
What is bot traffic on a website?
Bot traffic is every request to your site sent by an automated program rather than a person with a browser. Some of it is welcome: search engines have to crawl a page before they can rank it. Some of it is noise or abuse: spam referrers, scrapers copying your prices, and scripts posting to your contact form.
Website bot traffic shows up in two places. Your server logs see every request, because every request has to reach the server. Your analytics tool only sees the bots that run its JavaScript tag or send hits straight to its collection endpoint. That gap explains why the two sources rarely agree, and why you need both to get the full picture.
Good bots vs bad bots: which ones to keep
Before you block or filter anything, sort the bots into two groups. Blocking a good bot can cost you search visibility, and leaving a bad one in place distorts every number you report.
| Bot | What it does | Keep it? |
|---|---|---|
| Search engine crawlers | Fetch pages so Google, Bing and others can index them | Yes, after checking it is the real crawler |
| Uptime and performance monitors | Ping your pages on a schedule to warn you about outages | Yes, filter them out of reports |
| Link preview fetchers | Load a page when someone shares a link in a chat app | Yes |
| Referral spam bots | Fake visits that plant a domain in your referral report | No |
| Scrapers | Copy content, prices or listings in bulk | No |
| Form and login bots | Post spam, create fake accounts or try stolen passwords | No |
A user agent string is easy to fake, so a request that says it is Googlebot may not be. Google explains how to verify Googlebot with a reverse DNS lookup followed by a forward lookup. If you want to know how bots fake browser signals more generally, see our guide on what a headless browser is.
How much internet traffic is bots?
Roughly half. Imperva's yearly Bad Bot Report has put automated traffic close to or above half of all web traffic in recent editions, with a large share of it classed as bad bots. Cloudflare Radar publishes a live split between human and automated traffic across its network.
Your own share depends on your site. A small blog with a public contact form can see a much higher bot ratio than a busy shop, simply because there are fewer people to dilute the scripts.
How to detect bot traffic in analytics and server logs
Work through the signs below in this order. The first ones are the quickest to spot and the easiest to filter. The last one needs your server logs.
Referral spam sources
Referral spam is fake traffic that exists only to put a domain into your referral report, hoping you visit it out of curiosity. Open your traffic acquisition report, sort by session source and look for domains you do not recognise, odd names that promise SEO or traffic, and sources with many sessions but almost no engagement. Do not open those domains to check them.
Zero-second sessions
Real people scroll, click and stay at least a few seconds. Many bots load one page and leave at once. In Google Analytics 4, a session counts as engaged when it lasts longer than 10 seconds, has a key event or has two or more page views. A source or landing page whose engagement rate sits near zero, with an average engagement time of zero seconds, is worth a closer look.
Traffic at odd hours
Your visitors follow a daily rhythm in their own time zones. Bots often do not. In an exploration, add the hour dimension and compare it with your usual pattern. A flat line through the night, or a sharp block of sessions at 3 a.m. local time from one source, points to automation.
Single-page spikes
A sudden jump on one page, with no campaign, mention or newsletter behind it, is a classic sign of web traffic bots. Look at the landing page report for the spike day. If one page took almost all of the extra sessions, all with the same browser, screen size or location, treat it as fake website traffic until proven otherwise.
Data center networks in your server logs
People browse from home broadband and mobile networks. Bots usually run on cloud and hosting servers. Google Analytics 4 does not show the network a visit came from, so this check happens in your server logs. Count requests per IP address, then look up who owns the busiest ones.
# Top 20 IP addresses by request count in an Nginx or Apache access log
awk '{print $1}' access.log | sort | uniq -c | sort -rn | head -20
# Check who a busy address claims to be (reverse DNS)
host 203.0.113.10If a busy address belongs to a hosting provider and requests pages at a steady, machine-like pace, it is almost certainly a bot. Our sister site Network Tools can help with DNS and network lookups when you check a handful of addresses by hand.
How to filter fake website traffic from your reports
Google Analytics 4 already excludes traffic from known bots and spiders on its own, and you cannot turn that off. Google describes it in its known bot-traffic exclusion help page. Everything that slips through needs your own filters.
- List the bad sources. Write down the spam referrers, hostnames and landing pages you found in the checks above.
- Build a comparison or segment. In reports, add a comparison that excludes those sources. In explorations, create a session segment that excludes them, plus sessions with zero engagement time if they are clearly junk.
- Save it and use it every time. Report from the cleaned view, and keep the raw view for spotting new patterns.
- Exclude your own monitors. If an uptime checker runs your analytics tag, exclude its source or stop it from loading the tag.
- Block at the server for repeat offenders. A scraper that hammers you from one hosting network can be rate limited or blocked in your web server or firewall. This removes it from both logs and analytics going forward.
Filters and segments change what you see, not what was collected. Historic data stays as it is, so note the date you started filtering when you compare periods.
Does a captcha remove web traffic bots from analytics?
No. A challenge runs only where you place it, usually inside a form. Bots that load a page, trigger your analytics tag and leave never reach the form, so they still land in your reports. A captcha will not lower your bot sessions or fix your referral report.
That is not a weakness, just a different job. Analytics cleanup is about reporting. A captcha is about stopping a bot from doing something on your site: posting a comment, creating an account or trying a password.
Where bot traffic detection fits: your forms
Forms are where bot traffic costs you real money and time: spam in your inbox, fake signups in your user table, stolen card tests at checkout. This is the one place a challenge fits, and PrivCaptcha is built for it. The widget gives each visitor a puzzle with proof-of-work, and your server redeems the resulting token with one verify call before it saves anything. Each token works once and only for the site that issued it.
See what the dashboard countsServed, solved, failed and verified, per site and per dayHow to compare form submissions with verified counts
The PrivCaptcha dashboard counts four things per site and per day: challenges served, solved, failed and verified. A verification is counted when your server successfully redeems a token. That makes a simple bot check possible:
- Count your stored submissions for one form and one day.
- Read the verified count for the same site and day in the dashboard, or export it with the CSV button.
- Compare. More stored submissions than verifications means some posts reached your handler without a redeemed token, for example a route that skips the verify call or code that fails open on an error.
- Watch the failure rate. The analytics page tracks failure and abandonment rates. A spike in failures is the clearest single sign that bots are working the form.
- Check solved but never verified. Tokens that were solved but never redeemed point to a form whose server side does not check them at all.
If you protect several forms, give each one its own site in the dashboard so the counts line up one to one. For the setup itself, follow the quickstart, and for the attacks that target forms, read about spam comments and contact forms.
Ready to stop bots where they cost you?
- Every form post solves a challenge and pays proof-of-work
- Your server verifies a single-use token before saving anything
- Daily verified counts to check against your own submissions
Frequently asked questions
Bot traffic is any visit made by software instead of a person. It includes useful crawlers from search engines and uptime monitors, and harmful bots such as referral spammers, scrapers and scripts that submit forms. Your server logs record all of it, while your analytics tool only sees the bots that run its tag.
Check in this order: unknown referral sources, sessions with zero engagement time, traffic at hours your audience sleeps, sudden spikes on a single page, and requests from hosting or cloud networks in your server logs. One sign alone proves little, but several together on the same source almost always mean automation.
Referral spam is fake traffic sent only to plant a domain in your referral report so you visit it. You cannot delete it from collected data, but you can hide it: build a comparison or segment that excludes those sources and report from that view. Never open the spam domains to check them.
Spam hits in your analytics do not change rankings by themselves, because search engines rank what they crawl, not your reports. Bots can still hurt indirectly: scrapers copy your content, heavy crawling slows your server, and fake numbers lead you to wrong decisions about which pages work. Never block real search engine crawlers.
No. A captcha runs only on the forms where you add it, so bots that load a page and leave still appear in your analytics. Use filters and segments to clean reports. Use a captcha to stop bots from submitting forms, then compare stored submissions with the verified count in your dashboard.