Your Server Logs and Your Analytics Disagree. Bots Live in the Gap.
The server logs show 100,000 requests to the product page. The analytics platform shows 60,000 pageviews. The gap is 40,000 requests, and the team explains it away: ad blockers, privacy settings, bots that do not run JavaScript. The first two are real. The third is the one that matters, because the gap is where the bots live, and the size of the gap is the size of the problem nobody is measuring.
Why the two counts differ
Server logs count every request that reaches the server, including the ones that never render a page: scrapers, credential stuffing attempts, vulnerability scanners, and bots that fetch the HTML and leave. Analytics counts only the requests that execute JavaScript and set a cookie, which means it counts the browsers and misses the machines. The gap between the two is not an error. It is a measurement of the traffic that does not behave like a browser, and some of that traffic is legitimate, search crawlers and uptime monitors, and some of it is not.
The legitimate share is small and stable. Search crawlers identify themselves, uptime monitors hit a handful of pages on a schedule, and both are easy to name. The rest of the gap is the problem: requests that look like browsers in the user agent and behave like machines in every other signal. They fetch pages at regular intervals, they never request the images, they never move a mouse, and they never come back through the same path a human would.
What lives in the gap
Three kinds of traffic live in the gap, and they cost different things. Scrapers take prices, inventory, and content, and they cost margin, because a competitor with your prices can undercut you by a dollar and win the sale. Credential stuffing bots take login attempts, and they cost accounts, because a successful stuffing attack is a breach with a customer's name on it. And click fraud bots take ad clicks, and they cost money directly, because every fake click is a charge for a visit that never happened. The gap does not distinguish between them. The gap only says how big the problem is, and a gap that grows month over month is a problem that is growing, whether or not anyone has named it yet.
The gap also hides the bots that do run JavaScript. A headless browser executes the page, sets the cookie, and shows up in analytics as a visitor, which means the analytics count is inflated too, and the real gap is bigger than the measured one. The server logs are the ground truth for volume. The analytics are the ground truth for behavior. Neither one alone tells the story.
The check that names the bots
Compare the two counts weekly, by page, and watch the gap. A stable gap is infrastructure: crawlers, monitors, and the ad blockers that have always been there. A growing gap is an attack, or a scraper that found a new target, or a botnet that moved in. Then look inside the gap: requests with no image fetches, requests at regular intervals, requests from datacenter ASNs, and requests that hit the login page and nothing else. Each of those is a signature, and a signature is enough to name the traffic and block it. For the traffic taxonomy that decides whether it gets refunded, see GIVT vs SIVT and the MRC split, and for the headless browsers that hide inside the analytics count, how headless browsers fake a real browser.
Frequently asked questions
Why do server logs and analytics never match?
Because they count different things. Server logs count every request that reaches the server, including bots that never render a page. Analytics counts only requests that execute JavaScript and set a cookie. The gap is the traffic that does not behave like a browser.
Is the whole gap bot traffic?
No. Some of it is legitimate: search crawlers, uptime monitors, and ad blockers. The legitimate share is small and stable. A gap that grows month over month is the signal that bot traffic is growing.
Can bots show up in analytics too?
Yes. A headless browser executes the page, sets the cookie, and appears in analytics as a visitor. That inflates the analytics count and makes the real gap bigger than the measured one.
What signatures name the bots inside the gap?
Requests with no image fetches, requests at regular intervals, requests from datacenter ASNs, and requests that hit the login page and nothing else. Each is a signature, and a signature is enough to name the traffic and block it.