You look at the open rate on your latest newsletter, it shows 45%, and it looks like an excellent result. The problem is that machines account for a large share of those opens. Apple's servers and a range of security bots open your emails automatically, with nobody reading them, and the figure climbs. Any decision you take on that number rests on data that does not reflect what really happened.
In this article you will see how an open is counted, why yours is inflated, why that causes real damage, and which metrics to look at instead if you want to know whether your email programme actually works.
How is an open counted?
To understand the problem you have to start with how an open is measured, because everything follows from there. Your email marketing platform places a tracking pixel in every email, an invisible one-by-one-pixel image. When the email is opened and that image loads, an open is recorded.
The flaw is that anything capable of loading that image counts as an open, even with no human involved. That is where the machines come in.
What causes it
Three culprits are behind an open rate that runs higher than your real readership:
- Apple Mail Privacy Protection (MPP). This is the main factor. Since 2021, with iOS 15, Apple preloads the images in every email arriving at its Mail app through its own servers, before the user has even opened them. That fires the pixel, so every email sent to an Apple Mail user counts as opened whether anyone reads it or not. Estimates put the inflation at between 30% and 45% of opens. It also affects the Mail app regardless of the address behind it, so an email from a Gmail account read in Apple's app is affected too.
- Security bots and filters. Plenty of antivirus and corporate security systems open emails automatically and click the links to check they are not a threat. This inflates opens and clicks alike, and the clicks are the more dangerous trap, because the click is usually the metric you lean on.
- Image caches and proxies. Some providers preload or cache the images in emails, which can register opens that have nothing to do with a real read either.
People genuinely opening your email.
Preloaded by Apple's servers, with nobody reading them.
Systems opening the email to check that it is safe.
Why this is a genuine problem
This goes beyond a number that runs a little high. An inflated open rate breaks things that do matter:
- You cannot trust the open as a signal of subscriber interest.
- Your open-based automations stop working properly, resending a campaign to everyone who "did not open" being the obvious case, because many of those non-openers did open, and the reverse happens just as often.
- It distorts your subject line A/B tests, since both versions carry the same inflated opens and the real difference disappears into the noise.
- It muddies the engagement signals that mailbox providers watch, which can affect your deliverability over time.
What to look at instead of the open rate
The good news is that machines leave the metrics that matter untouched, because those require a human action:
- Clicks and CTR. A click requires somebody to press a link, which makes it far more reliable than an open. One caveat: security bots also generate clicks, so it is worth filtering them, something most platforms allow.
- Conversions and email revenue. This is the only thing that matters at business level. How many sales and how much revenue each campaign generates, or better still, how much per recipient. No machine buys anything.
- Unsubscribe rate. The pace of unsubscribes is an honest signal of fatigue or of a mismatch with your audience.
- Real activity on your site and purchases. To measure engagement properly, look at who visits and buys after receiving your emails rather than at who "opens" them.
How to adapt your flows and your segmentation
With that clear, here is how to adjust your email marketing:
- Filter bot clicks. Tools such as Klaviyo and Omnisend let you identify and discard a good share of automated clicks, so your data stays clean.
- Segment on behaviour rather than on opens. Build your active subscriber segments on clicks, site visits and purchases instead of on who opens.
- Switch your automations from "did not open" to "did not click". It is a far more reliable signal of genuine disinterest than an open that may never have happened.
- Account for Apple opens. If your tool allows it, identify or filter Apple MPP opens so the data stops misleading you.
The STRAT view
The conclusion is the same one we apply to any metric. The open rate is a vanity metric, the email equivalent of gross platform ROAS, a number that looks good without telling you whether you are making money. Do not throw it out altogether, it still works as a quick deliverability check and as a trend line within a single list, but stop treating it as your headline metric.
Frequently asked questions
Why did my open rate suddenly jump? Apple MPP is almost always the reason. Since 2021, Apple preloads the emails arriving at its Mail app and fires the tracking pixel, so everything reaching those users counts as opened. If your rate jumped with no obvious cause, that is usually it, rather than a real rise in interest.
Do bots inflate clicks as well? Some of them do. Security and antivirus systems open emails and click the links to check that they are safe, which generates clicks with no person behind them. That is why it is worth filtering bot clicks, something most email platforms allow.
Should I stop looking at the open rate? Not completely. It still works as a quick deliverability check and for spotting trends within the same list. What you should avoid is using it as your headline metric, or comparing your current data with anything from before 2021, because the two no longer measure the same thing.
Which email metric is the most reliable? Clicks and, above all, the conversions and revenue each campaign generates. They require a human action, so machines cannot inflate them. At business level, revenue per recipient is the most honest metric you have.
Can bots be filtered out? Partly. Email marketing platforms such as Klaviyo and Omnisend let you filter a good share of bot clicks and work with click-based segments. Apple MPP opens are harder to isolate, because Apple mixes the preloads in with real opens.
Does Apple MPP affect my deliverability? Indirectly. MPP does not penalise your delivery, but by inflating opens it muddies your engagement signals and can lead you into poor segmentation decisions that do end up affecting your sending reputation. That is why it pays to segment on clicks and behaviour instead of on opens.
If you suspect your email marketing is being measured with inflated data, or you want your flows and segmentation built on real metrics, we can review it with you. At STRAT we analyse your email programme so you can make decisions based on what actually generates revenue, with ecommerce email automation judged on the turnover it brings in rather than on an open rate.
