2026-08-14

Bot traffic is inflating your numbers

Bot traffic is machine visits. No person, no wallet, no decision.

Péter leads growth at a software company. One month a new channel jumped 41 percent.

He moved budget into it. He told the board it was working.

Sales did not move. Not that month or the next one.

The traffic was real. The visitors were not. Scrapers had found his pricing page and kept coming back.

Who is actually visiting

Not every machine visit is hostile. There are four kinds.

• Search engine crawlers. Google and Bing need to read your site. You want these.
• Uptime and security scanners. Your own tools, checking your site every few minutes.
• Scrapers. Competitors watching your prices. Data companies collecting content for AI models.
• Fakes. Software that claims to be Google so nobody blocks it.

The first two are harmless once they are filtered out. The last two inflate your reports and tell you nothing.

On a small site a few hundred machine visits double your monthly number. There is little data to spare there.

Scraping has grown fast in the last two years. AI companies need text and your website is text.

"There are three kinds of lies: lies, damned lies, and statistics."

Mark Twain, Chapters from My Autobiography, 1907

Twain credited the line to Disraeli. Nobody has ever found it in Disraeli's writing.

Even the quote about bad data has a sourcing problem. That is almost too neat.

Six signs your numbers are inflated

You do not need a technical audit. Look for these.

• Traffic climbs but sales do not. The oldest warning sign there is.
• Visit time is zero. A real reader spends some seconds on the page. A machine does not.
• One page, over and over. Usually pricing, a product feed or your sitemap.
• A country you do not sell to. Especially a sudden burst from one place.
• Traffic at three in the morning. Steady and flat, with no daily rhythm.
• A perfectly even line. Human traffic goes up and down. Machines do not get bored.

One sign on its own means little. Three together mean you are counting robots.

What good filtering actually does

StatJolt filters bots in three layers, before anything reaches your reports.

Here is what each layer does, without the jargon.

1. It checks the known list. Most bots identify themselves honestly. A maintained list catches them.

2. It verifies the ones claiming to be Google. Anyone can label themselves Googlebot. StatJolt checks whether the visit really came from Google's own network. If not, it gets dropped.

3. It blocks visits from rented servers. This is the layer most tools skip.

That third one deserves a plain explanation.

Real people browse from home internet or a mobile network. Machines usually run in a data centre, on a rented server.

StatJolt knows which networks are data centres. Traffic from there does not get counted as a person. The list updates every month, because new providers appear constantly.

Still with me? Good, because honesty matters more than the feature list here.

No filter catches everything

Anyone promising 100 percent bot removal is selling something.

Bots evolve. Some now run through home connections and use real browsers. Those are genuinely hard to spot.

What good filtering gives you is a much cleaner baseline. Your numbers get closer to reality and they stay comparable month to month.

That last part is the real prize. A number that is wrong in the same way every month is still useless for decisions.

Two habits that protect your decisions

Compare against something that cannot be faked. Orders, invoices, signups in your own database. A bot never buys anything. If traffic rises 40 percent and orders stay flat, believe the orders.

Check the shape, not just the total. Look at how a traffic spike is spread across pages, hours and countries. Real interest is messy. Fake interest is smooth.

This is the same discipline we recommend for ad blockers and analytics, where the error runs in the opposite direction. Bots make your numbers too high. Blockers make them too low.

Both problems have the same cure. Anchor to a number you know is true.

Where StatJolt fits in

StatJolt runs all three filtering layers by default. There is no setting to switch on and no list to maintain.

Machine visits are dropped before they are stored. They never enter your reports or your comparisons.

You get a smaller number than some tools show you. That is the point. A smaller honest number beats a bigger fake one, especially when you are moving budget based on it.

Dead landing pages distort your data in a similar way. That one is here: 404 errors are burning your ad budget.

If you only read one box

Bots inflate traffic and scraping has made the problem worse in the last two years. Watch for the classic signs: flat sales, zero visit time, one repeated page and traffic with no daily rhythm. Filter machine visits before they are counted and always check your numbers against orders or signups.

Back