Email open rates are becoming a weaker measure of customer attention

Why email open rates are less reliable than they used to be

Traditional email open tracking usually depends on remote content being loaded when an email is displayed. A small tracking image allows the sending platform to record that event as an open.

Privacy technology has weakened the connection between that technical event and actual human attention.

Apple’s Mail Privacy Protection, for example, can download remote email content in the background regardless of whether the recipient actively engages with the message. An email platform may therefore record activity associated with an open even when that signal does not represent somebody deliberately reading the email.^[1]

This does not mean open rates should be removed from campaign reporting.

They remain useful for directional comparisons, particularly when similar audiences, sending conditions and devices are involved. What has changed is the confidence with which an open can be interpreted as proof of attention.

A precise-looking open rate is not necessarily an equally precise measurement of customer behaviour.

Can somebody see an email without opening it?

Yes, at least partially.

A customer does not need to enter an individual email before receiving information from it. Sender name and subject line are visible in the inbox, and many email clients also display several lines of preview content.

Research into mobile email behaviour demonstrates that these previews can meaningfully change how people process their inbox.

A MobileHCI study tested different amounts of preview information and found that showing approximately two to three lines reduced the number of transitions participants needed to make between the inbox overview and individual emails.^[2] The participants could process more information without opening every message.

This was a human-computer interaction study, not a marketing conversion study. It does not establish that reading a preview causes somebody to visit a restaurant or make a purchase.

It establishes a narrower point that matters for campaign analytics:

Useful information can be processed before an email is opened.

“Unopened” and “unseen” are therefore not necessarily the same condition.

The inbox itself is already part of the campaign

Research specifically involving tourism and hospitality email campaigns supports this broader view of the inbox.

A study published in Review of Managerial Science analysed 5,765 promotional emails sent between 2013 and 2018 to approximately 455 million recipients across 73 countries. The dataset included tourism and hospitality advertisers.^[3]

The researchers examined visible, temporal and contextual variables affecting opening effectiveness. Visible elements included information available before the recipient entered the message, particularly the sender and subject line.

The study also found that email effectiveness was not simply a matter of finding the perfect subject-line length. Audience segmentation, frequency and contextual factors also mattered.

For hospitality marketers, that is more useful than another universal rule about how many characters should appear in the subject line.

The recipient encounters the campaign in context:

• Who sent it?
• Is the brand familiar?
• Does the subject appear relevant?
• How recently has the customer heard from the venue?
• How frequently does the venue communicate?
• What information is already visible?

The open is one event inside that experience, not necessarily the beginning of it.

Subject lines matter, but be careful with famous statistics

Marketing articles frequently repeat statistics claiming that a particular percentage of consumers decide whether to open an email based on its subject line.

Many of those figures have become detached from their original research.

One widely cited source is a 2012 Chadwick Martin Bailey and Constant Contact survey. Contemporary reporting found that 64% of respondents identified the sending organisation as a factor in deciding whether to open an email, while 47% cited the subject line. The offer itself was cited by 26%, and 14% mentioned the first few lines of the email.

Over time, variations of these numbers have been repeated as universal rules about subject-line performance.

Controlled experiments provide stronger evidence.

Sahni, Wheeler and Chintagunta conducted randomized email experiments involving millions of recipients. In their main experiment, adding the recipient's name to the subject line increased opening probability from 9.05% to 10.80%. Sales leads also increased from 0.39% to 0.51%.^[4]

That does not mean every hospitality campaign should personalize every subject line with a first name.

It demonstrates that visible inbox content can change behaviour. The size and direction of that effect depend on the audience, message and context.

Email previews are becoming more complicated

The pre-open experience is also changing because email software is doing more of the interpretation itself.

Apple Intelligence can generate summaries of messages so users can understand key information without opening every email. Gmail is moving in a similar direction with AI-generated summaries and inbox assistance.

This changes the relationship between the marketer and the preview.

Historically, the visible layer was largely controlled by three things: the sender name, the subject line and the preheader or first lines of copy.

Increasingly, the operating system or email provider may decide what information deserves to be surfaced.

For marketers, that makes the boundary between “seen” and “opened” even less clean.

It also gives the opening content of an email another job. The beginning of the message is no longer necessarily written only for somebody who has already opened it. Parts of that content may be extracted, previewed or summarized elsewhere.

Click-through rate is cleaner, but a click is not the business outcome

Because open tracking has become less reliable, marketers have understandably placed greater emphasis on click-through rates.

A click is a stronger behavioural signal. Somebody had to actively interact with the message.

The limitation is different: commercially meaningful responses do not always require a click.

One useful study examined 70 randomized field experiments conducted by a large online ticket resale business. Customers were randomly assigned either to receive an emailed promotional offer or to a control group that did not receive it.^[5]

The promotions increased average expenditure during the promotional period by 37.2%.

More importantly for attribution, 90% of the incremental spending identified by the experiment did not occur through redemption of the emailed offer. The researchers also observed increased spending after promotions had expired and spillover into categories that had not been promoted.

The study does not prove those customers acted because they read a subject line or preview.

It demonstrates something broader: the action that is easiest to attribute inside the campaign can represent only part of the campaign's commercial effect.

Hospitality makes email attribution particularly difficult

Ecommerce often provides a relatively direct path:

Hospitality rarely behaves that neatly.

A customer may receive a lunch campaign, notice it while clearing notifications, discuss plans with colleagues later in the week and eventually walk into the restaurant.

There may be no click.

There may be no coupon.

There may not even be a recorded open.

Yet the campaign may still have contributed to the decision.

Research in multichannel retail provides evidence that email can influence behaviour outside the channel in which the communication was delivered.

A Journal of the Academy of Marketing Science study used transaction and marketing data across online and offline environments and supported its modelling with a randomized field experiment. It found that email could affect both online and offline sales for existing customer segments.^[6]

Retail is not hospitality, so those results cannot simply be converted into restaurant benchmarks.

The relevant principle is that communication and commercial response do not have to occur in the same channel.

For restaurants, cafés, entertainment venues and other physical businesses, that distinction matters considerably.

Return visits provide another signal

Hospitality businesses have an advantage that many digital marketers do not: where customer identity and visit data are connected, subsequent physical behaviour can be observed.

Affinect campaign analytics can connect campaign recipients with later venue visits and distinguish between customers who registered an email open before returning and customers who returned without a recorded open.

That adds context beyond the conventional email funnel.

It does not prove attribution.

Consider a customer who visits the same café every Friday.

The café sends that person an email on Tuesday. They do not register an open. They visit again on Friday.

The system can legitimately report:

It cannot legitimately conclude:

Campaign caused the subsequent visit.

The customer may have returned regardless.

This distinction becomes particularly important with regular customers, because their baseline probability of returning is already high.

Similarly, customers who frequently open campaigns may already be more engaged with the business and more likely to visit. Comparing openers and non-openers without accounting for those underlying differences can exaggerate the apparent effect of the campaign.

Return visits are therefore useful behavioural evidence, but they should not be presented as proof of causation on their own.

For more on using actual visit behaviour rather than static customer lists, see Affinect’s guide to behavioral customer segmentation.

A better way to interpret hospitality campaign reporting

Campaign metrics answer different questions.

Delivery indicates that the receiving infrastructure accepted the message.

Open provides an engagement signal, subject to privacy and technical limitations.

Click provides stronger evidence of deliberate interaction.

Redemption or booking provides a campaign-linked commercial event where that mechanism exists.

Return visit provides evidence of subsequent customer behaviour, although not necessarily its cause.

The problem starts when one of those metrics is asked to represent the entire customer journey.

An email open is not synonymous with attention.

A click is not synonymous with influence.

A subsequent visit is not automatically attributable to the campaign.

Used together, however, these signals produce a more useful picture.

Affinect’s email marketing for hospitality combines campaign interaction data with behavioural customer information and return visits so marketers can examine more than inbox activity alone.

How to determine whether a campaign actually caused additional visits

Causal measurement requires something stronger than observing that a customer returned after receiving a campaign.

One practical approach is a randomized holdout group.

A small proportion of customers who would otherwise qualify for the campaign are deliberately excluded. Their subsequent behaviour is then compared with customers who received the communication.

Instead of asking:

How many campaign recipients came back?

the business can ask:

Did customers who received the campaign return at a higher rate than comparable customers who did not receive it?

That is a considerably stronger question.

A second experiment could test the visible inbox layer directly.

Two otherwise identical groups could receive the same email body and offer but different subject lines or preview content. Instead of judging the winner only by open rate, the venue could also compare subsequent visits, bookings or redemptions.

That would provide much more useful evidence about whether pre-open content affects an offline business result.

At present, there appears to be limited published research connecting those steps specifically in hospitality.

That gap is worth acknowledging rather than filling with assumptions.

What hospitality marketers should measure

Opens and clicks should remain part of campaign reporting. Removing them would discard useful information.

They are more valuable when interpreted alongside customer and venue behaviour.

For a hospitality marketer, a more complete campaign view can include:

• delivery and bounce data;
• recorded opens and clicks;
• audience segment and historical visit frequency;
• coupon redemption or booking activity where available;
• subsequent venue visits;
• the time between communication and return;
• historical probability of returning;
• and, for campaigns where causality matters, a randomized control or holdout group.

This is also why behavioural segmentation matters. A weekly regular and somebody who has not visited for six months should not be interpreted in the same way simply because both received the same campaign. Affinect’s customer segmentation tools use visit behaviour such as recency and frequency to distinguish those customers. Where visit identity itself is the missing layer, start with customer identification.

The goal is not a better vanity metric

Email technology is becoming more sophisticated while some of its familiar marketing metrics are becoming harder to interpret.

Privacy systems can create technical opens without deliberate attention.

Inbox previews can convey information without an open.

AI-generated summaries can surface message content before the customer enters the email.

Clicks provide stronger interaction evidence, but capture only behaviour that travels through a clickable path.

And in hospitality, the eventual response may happen several days later inside a physical venue.

The answer is not to replace open rate with another single number.

It is to measure campaign performance closer to the commercial behaviour the business is trying to influence.

For hospitality, that usually means whether customers visit, return, book, redeem and spend.

An unopened email is not necessarily an unseen email. A campaign without a click is not necessarily a campaign without an effect. The difficult part is measuring that effect without claiming more than the data can prove.

References

1. Apple. Mail Privacy Protection. Apple Privacy / iCloud documentation.
2. Weaver, K.A., Yang, H., Zhai, S. & Pierce, J.S. (2011). “Understanding Information Preview in Mobile Email Processing.” Proceedings of MobileHCI 2011, 303–312. DOI: 10.1145/2037373.2037420.
3. Chaparro-Peláez, J., Hernández-García, Á. & Lorente-Páramo, Á.J. (2022). “May I have your attention, please? An investigation on opening effectiveness in e-mail marketing.” Review of Managerial Science, 16, 2261–2284. DOI: 10.1007/s11846-022-00517-9.
4. Sahni, N.S., Wheeler, S.C. & Chintagunta, P. (2018). “Personalization in Email Marketing: The Role of Noninformative Advertising Content.” Marketing Science, 37(2), 236–258. DOI: 10.1287/mksc.2017.1066.
5. Sahni, N.S., Zou, D. & Chintagunta, P.K. (2017). “Do Targeted Discount Offers Serve as Advertising? Evidence from 70 Field Experiments.” Management Science, 63(8), 2688–2705. DOI: 10.1287/mnsc.2016.2450.
6. Valenti, A., Srinivasan, S., Yildirim, G. & Pauwels, K. (2024). “Direct mail to prospects and email to current customers? Modeling and field-testing multichannel marketing.” Journal of the Academy of Marketing Science, 52, 815–834. DOI: 10.1007/s11747-023-00962-2.