How to measure newsletter engagement now that open rates are unreliable

An open is counted when the email client downloads a tiny invisible image, the tracking pixel, from your email platform. Apple announced Mail Privacy Protection on 7 June 2021, saying it "stops senders from using invisible pixels to collect information about the user". It came with iOS 15, released on 20 September 2021. Apple's privacy page says the feature downloads remote content in the background by default, "regardless of whether you engage with the email".

So for an Apple Mail reader with the setting on, your platform sees the pixel load whether they read the issue or never looked at it. Kit says it "can't tell whether the loading of the pixel was due to the recipient opening your email" or Apple loading it automatically.

Litmus, which measures email client market share from the opens it tracks, says opens from Mail Privacy Protection users "are not considered reliable opens", and cites studies that put the feature at roughly 55 to 60% of all email opens. Your list will differ, but if your open rate jumped in late 2021 and stayed high, read that jump as pixels loading, not new readers.

Raw clicks have the opposite problem: machines add them. Image blocking, meanwhile, hides real opens.

Every metric built on raw opens or clicks inherits these errors, including click-to-open rate and "inactive subscriber" segments. Use bot-filtered clicks wherever your platform offers them.

Each platform documents its own approach:

None of these platforms claims to catch every bot. Use the same filtered figure for every issue and compare your issues with each other, not with an industry benchmark.

Make click-to-delivered rate your main number and read the others around it. Each has a blind spot:

Count unsubscribes and complaints per issue, not per month. A monthly average hides the one issue that lost 40 readers, and that is the issue you want to look at.

A table you fill in after each issue is enough:

To read it, compare each issue with your recent ones. A higher click-to-delivered rate than usual with normal unsubscribes means the issue worked. A jump in unsubscribes or complaints means it missed, whatever its clicks. Replies and referrals tell you which of the good issues readers cared about most.

Every number above is behaviour you infer. To know whether an issue helped, ask "Was this issue useful?" with a one-click Yes / No, in the email or on the web version, and add the share of Yes answers per issue to your scorecard. Our post on getting newsletter feedback compares the ways to ask.

Asklet adds a short survey to the web version of each issue, such as "Was this issue useful?" with a one-tap Yes / No, and shows the answers per issue by the path of its web page. On Ghost you add it with an HTML card or once in your theme; on beehiiv with an HTML Snippet block. It does not work inside the email itself or on Substack.

It sets no cookies and stores nothing that identifies the reader, so it cannot tell you which subscriber answered or split answers by segment. Use your email platform's own polls for that.