Why the robots are excluded

Issue Mint takes the robots out of every click figure it shows you, which is why the numbers here are lower than the numbers a tool that counts everything would report. That gap is the product working, not a shortfall in it.

What gets filtered is straightforward: known security scanners, the previewers behind chat apps, scripts and monitors, and one reader appearing to hit several different links within seconds. Everything else counts.

What a corporate mail gateway does to your figures

A security gateway fetches every link in every message before the recipient sees it, to check that none of them are dangerous. It is doing its job, and it is doing it seconds after your issue arrives.

So one delivery to one reader with an IT department behind them can produce a click on every link in the issue, from somebody who never opened it. On a list with a few hundred business addresses, that traffic can outnumber the real clicks. Left in, the figure you use to decide what to cover next, or to price a sponsorship slot, is fiction.

What Issue Mint recognises

Three families of automated visitor are recognised by the name they give for themselves, plus one that gives no name at all.

• Security gateways. Microsoft Defender safelinks, Proofpoint and its URL Defense rewriter, Mimecast, Barracuda, Symantec, Forcepoint, Trend Micro, Bitdefender, Sophos, Fortinet and Zscaler.
• Link previewers. Slack, Twitter, Facebook, LinkedIn, WhatsApp, Telegram, Discord, Skype and Bing, which fetch a page to draw a preview card when somebody shares your issue.
• Scripts and monitors. curl, wget, python-requests, Go’s HTTP client, headless Chrome, PhantomJS, and anything naming itself a bot, crawler, spider, scanner, preview or monitor.

A request that gives no user agent at all is treated as a script too, because a browser and a mail client both introduce themselves and a script often does not.

The timing rule

One recipient reaching three different links within ten seconds is counted as a sweep rather than as reading, because that is not a pattern a person produces.

The rule counts distinct links, which matters: a reader who taps the same link twice on a phone, or whose client follows a link and then hands the click through, is not mistaken for a gateway. It also only applies to clicks tied to a recipient, so a busy archive page is never caught by it.

Image proxies count on an open and not on a click

A mail provider that fetches images for its users is a real open and a robot on a click, and that asymmetry is deliberate.

Gmail proxies every image in every message it displays, so treating its fetch as a robot would report almost no opens for the largest part of most lists, which is why open tracking lets those through. When the same proxy turns up on a click, though, the request cannot be the reader: a reader’s own browser makes that request, so a proxy asking for the destination is the mail system checking a link, not a person following one.

The traffic is stored, not counted

A robot’s click is written down, flagged as a robot, and then kept out of every figure Issue Mint shows you. Nothing deletes it, because a scanner’s click is real traffic and it is useful when somebody asks why a figure moved.

No screen shows you that traffic, which is worth saying plainly: there is no toggle to reveal it and no second set of numbers to compare against. If a figure looks wrong, ask us and we can look. Robot opens are treated more bluntly still and are not stored at all, since a pixel fetch from a scanner has nothing in it worth keeping.

The filter is deliberately conservative

Where a request is ambiguous, Issue Mint counts it as a reader. A false positive silently removes a real person from every figure on every page, and that is worse than admitting the odd scanner.

The same restraint shows up in the device and client breakdown on an issue report, which answers Unknown rather than guessing at a mail client that has hidden itself. A confident wrong answer is the thing worth avoiding, on both counts.

Why the lower figure is the one you want

A click figure with the robots left in cannot be acted on, and acting on figures is the only reason to have them. It tells you a link did well when what happened is that a gateway swept it, and it does that unevenly, so your best-performing link becomes whichever one had the most corporate readers.

Two consequences worth holding on to. Click tracking here reports unique clickers beside total clicks precisely so you can see the difference between reach and repetition, and the case for measuring clicks at all rests on the figure being clean. Your numbers are also not comparable with a platform that does not filter, so compare Issue Mint with Issue Mint, and treat a competitor’s higher figure as a different measurement rather than a better result.