An open is counted only when a hidden image loads, so readers with images off are missed and some people who never read are counted. Watch clicks on the link that matters, and replies.

The number on the report

After every newsletter, the email service shows a report, and the first figure on it is the open rate: the share of recipients who opened the message.

Owners watch it closely. When it rises they credit the subject line, and when it falls they wonder whether people are tired of hearing from them.

The figure is less solid than it looks. It is not a count of people who read the email, but of times a particular small image was downloaded.

Once that is clear, most of the worry about it falls away, and attention can move to numbers that say more.

How an open is counted

Each email carries a tiny invisible image, a single pixel, with an address unique to that recipient. When the email program downloads it, the email service records an open.

No image download, no open. Whether the person read every word makes no difference to the count.

Many services also record an open when someone clicks a link, since a click proves the message was opened.

And anything that downloads the image counts as an open, whether or not a person was looking at the screen.

Readers who never show up

The first is the one that skews business lists. Plenty of office workers read every newsletter in Outlook without ever downloading an image, and the report records each of them as someone who ignored it.

Opens that are not readers

The count also runs the other way. A preview pane that shows each message as the reader moves through the inbox can download the image, recording an open for a message glanced at for a second.

A newsletter forwarded to a colleague is counted against the original recipient every time the colleague opens it.

Some security filters used by businesses open links in incoming mail to check them before delivery, which can show up as clicks, and sometimes opens, from people who never saw the message.

Gmail also fetches images through its own servers, hiding the reader's device and location, so reports on where and how people read are unreliable.

What clicks show

A click is firmer evidence. Someone read enough to want more, and chose a link.

Clicks have limits too. A reader who got everything from the email itself has no reason to click, however closely they read it.

Clicks are most useful counted on one link, the one the email exists for: the booking page, the new arrivals or the event. Total clicks across a dozen links say less.

Add campaign tags to that link so Analytics shows what those visitors did next, which is the question the open rate could never answer.

Replies, the number nobody counts

For a small business, a reply is the most valuable thing a newsletter can produce. It means a person read the message and wants something.

Most email services do not report replies, because replies land in your own inbox rather than passing through their system. So they go uncounted, even when they are the point.

Invite them on purpose: reply to reserve one, reply with your question, reply if you would like a call. Then keep a simple tally for each send.

A worked example

A private liquor store in a suburban shopping centre sends a weekly email to a couple of thousand customers who signed up at the till: new arrivals, a featured bottle and the Saturday tasting.

The open rate looked weak, and the email service suggested removing everyone who had not opened anything in six months. That came to several hundred addresses.

Before deleting them, the owner read through the names. Several belonged to regulars who mentioned the tasting emails at the counter most Saturdays.

The mechanism was simple. Many of them read the email at work in desktop Outlook, with images blocked, and the store wrote its emails as text, so they could read everything without loading anything.

The list stayed. Each email now ends with an invitation to reply and have a bottle held for the weekend, and staff keep a tally of replies and of tasting visitors who mention the email.

The open rate is still on the report. Nobody at the store makes decisions from it any more.

Comparing an email with itself

The open rate still has a use, compared only with itself. Same list, same kind of email, same day of the week: a sharp drop from one send to the next means something changed.

Often what changed is delivery. A sudden fall can mean messages started landing in spam folders or in Gmail's Promotions tab, which is worth looking into.

What the figure cannot do is stand against other businesses' rates, or averages published for whole industries. Different lists, readers and email programs make those comparisons close to meaningless.

Cleaning a list without losing readers

The costliest mistake with open rates is deleting people because of them. Advice to remove anyone who has not opened in a set period assumes the count is accurate, and on many business lists it is not.

Before removing anyone, look at what else they do. A subscriber who clicks, replies, buys or mentions the email in person is a reader, whatever the report says.

Remove addresses that bounce and anyone who asks to leave. Treat silence with more care.

The counter-case

For some lists the open rate sits closer to the truth. A list made mostly of personal Gmail and phone readers, whose images load by default, will report opens reasonably well.

A large sender testing two subject lines on the same list at the same moment can also learn something from the difference, since both versions face the same distortions.

A small business list is rarely either. Treat the open rate as a rough signal, and judge each email by clicks on its main link and the replies it brings.

What to watch

  1. Stop comparing your open rate with averages.
  2. Pick the one link each email exists for.
  3. Tag that link for Analytics.
  4. Invite replies in every send.
  5. Tally replies and mentions by hand.
  6. Compare each send with your previous ones.
  7. Check activity before removing quiet subscribers.

Step four is the one that changes what the email is for. A message that asks for a reply gets read differently, and every reply is a reader the report could not see.

Writing the email so it works before any image loads is covered in newsletters read with the images turned off.

Frequently asked questions

How does an email service count an open?

A tiny invisible image in the email is downloaded from the service's server, and that download is recorded as an open. Without it, many services count a reader only after a click.

Why do some readers never appear as opens?

Desktop Outlook holds back images until the reader asks for them, plain text versions carry no image, and some phones load images only on request. All these readers look like non-openers.

Can opens be counted that never happened?

Yes. Preview panes can load images for messages glanced at for a second, forwarded copies count against the original recipient, and some security filters check messages first.

Which number should a small business watch?

Clicks on the one link each email exists for, tagged so Analytics shows what happened next, and replies. Both show that a person read the message and wanted something.

Should I remove subscribers who never open?

Not on the open rate alone. Check first whether they click, reply, buy or mention the email in person. Remove bounced addresses and anyone who asks to leave.

Is the open rate useful for anything?

As a comparison with your own previous sends. A sudden drop on the same list and the same kind of email can mean messages started landing in spam.

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.

Judging your newsletter by its open rate?

Send us a recent report. We will show you which numbers in it reflect real readers and which do not.

Start a Conversation