Original research · May 24 – Aug 29, 2026

Most of your website traffic isn’t human — and the biggest bot is the platform you buy ads from

We analyzed 46,903 sessions recorded by our first-party tracking pixel on US local-business websites. 68% of all sessions were automated — and the single biggest non-human visitor wasn’t a shady scraper. It was Meta.

1. Who the bots actually are

Of all detected bot sessions: Meta’s crawlers produced 19,168 sessions — about 49% of the total and nearly double Googlebot (11,243). Generic scripts and scrapers added 5,383 more. If your business advertises on Facebook or Instagram, the platform crawls your site relentlessly — and every one of those visits inflates your “traffic” in naive analytics.

2. Bot share by industry — a 4.7x spread

The dirty secret of “average bot rate” statistics is that no business is average. Across four verticals in our dataset, bot share ranged from under a fifth to nearly nine in ten sessions:

The pattern: sites that run paid social and get scraped for product data (retail, B2B) drown in crawler traffic; local consumer sites with mostly direct and social human traffic stay cleaner.

3. The human traffic looks nothing alike either

Parents researching childcare are 91.6% mobile; buyers of B2B services are 93% desktop. An entertainment venue’s audience comes back constantly (60.2% returning); a considered-purchase retail audience almost never does (13.7%). Any “average” benchmark that ignores vertical is telling you about someone else’s business.

4. Why this matters for your marketing math

If 68% of raw sessions are bots, every rate computed against raw traffic — conversion rate, cost per visit, engagement — is understated by roughly 3x. Businesses conclude their site “doesn’t convert” when the human-only math says it converts fine and the problem is volume. Bot-filtered analytics isn’t a nice-to-have; it changes the decision you make.

Methodology

Data: 46,903 sessions recorded May 24 – Aug 29, 2026 by Karbon’s first-party pixel (k.js) on US local-business websites, classified server-side by user-agent and behavioral analysis. Per-vertical figures use the four sites with meaningful volume; vertical labels are anonymized categories, not client names. One important caveat: this measures JavaScript-executing bots only. AI-training crawlers such as GPTBot generally do not run JavaScript and are therefore undercounted here — the true non-human share is higher than reported. Aggregates are cross-client and anonymized. We refresh quarterly.

Citing this report? Link this page and credit “Karbon Agency, Bot Traffic Report 2026.” Journalists can request the underlying aggregate tables via the contact page. See also: State of Local Business Web Traffic 2026.