From visit to habit: how newsletters and push can bring readers back

A campaign can lift clicks and barely change how often people keep up with a site. Extra visits on the same days can carry editorial and advertising value. But to know whether a reader now follows the outlet more regularly, you have to look at how many days and across how many weeks they come back.

Newsletters and push notifications offer something more valuable: the chance to set up moments when the audience expects something useful from that outlet. The morning email that helps someone start the day, the roundup from one desk, and the alert about a change that matters can all become part of that routine.

Automated scheduling keeps the work going without asking someone to pick and build every send by hand. When the selection reads how stories are performing and applies the criteria the newsroom set, automation also helps decide what goes out. The gain depends on doing both jobs at once: choosing what fits and sending on schedule.

The challenge is to find out whether this setup improves your relationship with the reader or only increases the number of messages they receive.

The research backs the channel, not your version of it

A 2024 study in the Journal of Media Economics tracked 16,284 new digital subscribers at four American newspapers. Signing up for a newsletter in the first month was associated with more days of visits over the next two months, and with a lower risk of canceling later on. One of the models estimated 36.1% more active days among the newsletter subscribers. Study by Kim, Kim, Choi, and Malthouse.

For notifications, there is experimental evidence, though it is old and short-term. In 2016, the Center for Media Engagement at the University of Texas at Austin split 420 participants across conditions with different news apps and with alerts on or off. About two weeks later, 27% of the group assigned to receive alerts reported using the app daily, against 12% in the other group. The study covers apps and self-reported use, not web push and not habits formed over months. Experiment on news notifications.

These studies give you grounds to invest in these channels. You still have to test how you set it up in your own newsroom: who receives it, what information earns a send, how often, and what it does to the relationship.

Design a campaign that has something to offer

Take a hypothetical regional publisher. The team decides to build a newsletter about the changes that affect daily life in the city, sent at 7:30 a.m. on weekdays. Readers know they will find a selection of stories on transportation, public services, and city government decisions, explained clearly enough that a reader knows what changed.

The publisher also tests a scheduled push at 12:30 p.m. for the people who opted into that channel. Its job is to single out one thing worth knowing that day. Urgent news, like an unplanned service outage, goes through a different editorial call and does not have to wait for that slot.

A first setup could work like this:

Those times and caps belong to the example, not to some ideal frequency. Live sports coverage, a markets briefing, and a weekly newsletter each need something different.

One more distinction matters: scheduling a message you already wrote is not the same as scheduling a selection. With a selection, the system reads the pool of eligible content at the appointed time, applies the criteria, and builds the send. The team stops repeating part of the daily work of hunting for stories, picking links, and assembling the message.

In PageSignal, newsletter scheduling and push scheduling bring together content filters, selection criteria, and the choice between automatic sending and human approval. That combination lets you run campaigns that use your own site data to distribute stories inside the rules your team has set. Fact-checking, judging urgency, and deciding how wide the coverage should be still call for your own editorial process.

Automated selection has to reckon with the limits of a ranking

If the local campaign simply takes the most-read story across the whole site, one big national story can occupy every send. The product would stop keeping the promise it made to its readers, even with the clicks running high.

So narrow the pool first: decide which stories are worth distributing at all. A neighborhood story may draw fewer visits across the whole site and still be the better fit for that newsletter. Performance criteria come in after that choice.

Even inside one desk, volume needs interpretation. A story published twenty minutes ago has had less chance to be read than one promoted since the morning. A high rate built on few visits can swing a lot. And a weak result may reflect a lack of promotion, which is exactly the problem the campaign was supposed to help solve.

In the pilot, the team can keep human approval on and record why it overrides a selection. Frequent changes mean the rules still do not describe the product well. When the selections hold up across different days and different subjects, there are better grounds to automate the sending and keep watching it.

A campaign also needs a rule for when not to send. When no story fits, the answer is not to fill the slot with whatever is available just to keep the schedule.

With push, silence can protect attention for the next alert. With the newsletter, the newsroom has to decide what to do without breaking the expectation of a regular edition, which can mean preparing suitable content or announcing a change in how often it goes out. The two cases call for different decisions.

Measure the operational gain inside your own routine. Time how long it takes to pick the stories, build the edition, review it, and get the send out. Then, with automated scheduling in place, compare editions of similar size and complexity. Count the time spent supervising and fixing selections: a faster build can demand more review.

Time saved needs somewhere to go. It can pay for a better edition, the reporting you are missing, or the analysis of results. Spend it only on multiplying sends and the outlet may trade manual work for list fatigue.

When more clicks argue for sending less

Now suppose the publisher wants to go from one scheduled push a day to two. The goal in this pilot is to raise reading frequency across the weeks while holding on to the people who allow notifications. Comparing one month against the next would be weak: the news cycle, the size of the list, and where readers come from all change too.

A more informative test splits the eligible participants at random into two groups and holds the other conditions comparable. Send frequency is the main change: one group stays on a single daily send, the other gets two. Run it across different days of the week, long enough and with enough people to support the decision.

Clicks rose 50%. The gap in visit days was small, and until you check its statistical precision it supports no conclusion about a real gain in frequency. Meanwhile, more participants turned notifications off.

Against the goal this pilot set, the results shown do not justify raising the frequency. The prudent call is to keep the old cap while you look into whether the second send adds anything useful. The publisher may be packing more clicks into the same days, at the cost of losing permission to reach part of the list.

The average number of visit days also has to be read alongside how those days fall across the weeks. Six days bunched at the start of the month mean something different from six days spread across all four. A second measure for this pilot would be the share of participants who visited the site in three or four separate weeks, always with each group's original total as the denominator.

The test only works if you can follow the participants for the whole period, including those who stop receiving push. That takes an identifier compatible with your data collection permissions, through login or another consistent mechanism. If you only count clicks on the messages, you cannot tell how many days the reader reached the site by another route. Running the experiment may take instrumentation and tools beyond the ones that handle the sends.

The risk of fatigue is not merely theoretical. In a 2025 Reuters Institute survey across eight countries, Brazil among them, 43% of the people who were not getting news alerts said they had turned them off. Too many messages, and messages that were not useful, show up among the reasons. That percentage describes only that group, not all users and not a campaign followed over time. Reuters Institute: use and rejection of notifications.

The habit can also live inside the email

Not every newsletter has to produce a visit to be worth something. An edition can offer a complete read that the subscriber folds into a routine. Forcing clicks by holding back essential information can damage the very product the subscriber signed up for.

Then your metrics have to cover how long people stay on the list, replies, what they say about the product, and, where it applies, subscription retention. Read the open rate carefully: Apple's Mail Privacy Protection can load remote content in the background, so an open does not necessarily mean the email was read. Apple documentation on Mail Privacy Protection.

Nor does a habit require the reader to start coming back with no prompt at all. Reading a daily newsletter, or a visit that started with a useful alert, can be a steady part of the relationship. The question is whether the product still delivers value and whether the frequency you chose can carry that relationship.

Automation lets you keep distribution running with fewer repetitive tasks and steadier criteria. It pays off most when that capacity helps the outlet become something the reader expects. That takes watching both what the sends produce and what the list stops accepting.