Trigger

For a long time there were two explanations for a user who doesn't use something. They don't want to, or they can't. If it's the first, persuade them; if it's the second, make it easier. So teams poured effort into better features and simpler screens.

Cases that were neither kept turning up. When Pagedeck shipped document version comparison, first-month usage came to 4 percent. Mia interviewed twelve people who hadn't used it and the answers were nearly identical: they knew the feature was useful and they knew where it lived, but at the moment they actually needed it, it never came to mind.

So there's a place where a behavior fails to start even though someone both wants to and is able to. What was missing was a cue. Intent and ability need a third item beside them before the explanation holds up, and the word in that third slot is trigger. It's the same reason the hook model — which writes engagement as a loop of cue, action, reward, and investment — puts the cue first. The other three can be well designed and the wheel still won't turn without the first step.

Triggers come in two kinds.

External triggers are signals the product or another person sends from outside. Notifications, email, the number badged onto an app icon, a link a colleague pasted into a chat channel. They can be switched on and off, and their effect shows up immediately.

Internal triggers are a state where a user's own situation or feeling is bound to the product. Five minutes before a meeting, the thought that the agenda needs writing up arrives and the document tool gets opened automatically. Visits happen without anything being sent.

The relationship between the two is why the concept is worth having. External triggers are the bridge that holds until internal ones settle in. Repeat the same action in the same situation often enough and the user starts thinking of the product whenever that situation arrives; from then on they come without being pinged. Cut notifications before the bridge is built and the visits go with them. Pagedeck once paused all notification sending for a week and daily visits fell 42 percent — read that as roughly half its users having reached an internal trigger and half not.

Building one external trigger means deciding three things.

1. The signal: what you're announcing. A colleague commenting on your document is a signal. A new feature shipping this week is weak as a signal, because nothing happened to that particular user.
2. The timing. Comment notifications go out immediately while company-wide announcements get collected and sent once in the morning. The list of those rules is what a notification strategy holds.
3. The destination: where the tap leads. Tap a comment notification and land on the product's home screen, and the user hunts for the comment and gives up. It has to open the document at that comment for the action to finish.

The failure mode is equally predictable. Send weak signals often and users swipe past without reading, or turn the setting off. That state is notification fatigue, and people rarely turn notifications back on themselves. Since visit counts rise with every extra notification in the short run, the bill for this usually arrives a quarter late.

The first place it gets used is onboarding design. Which cues go out during the first week, and how many, largely decides whether a user ever forms an internal trigger. Pagedeck sends a user who created a first document one prompt the next day suggesting they invite a colleague, and after that only notifies them when a comment arrives. Users who created nothing get nothing, because a cue sent to someone with no available action leaves only fatigue behind.

Winning back dormant users is the second place. A message to someone who hasn't shown up for three weeks has to point at what they left behind rather than at what the company shipped. A note saying a half-written document is waiting outperforms a list of this month's updates.

The devices attached to product screens are cue-making work too. A streak, which counts consecutive days and shows the number, uses the wish not to break the run as its cue. A variable reward, where the content differs on each visit, manufactures the wish to check and so prepares the next cue on its own.

Measurement follows a fixed shape. For each cue, read the share that got tapped alongside the share that finished the intended action afterward. Read only click rate and the notification with the most provocative subject line wins, while doing nothing but spending down the credibility of the next one.

"This feature has no trigger. Nothing in the product brings it to mind at the moment it's needed."

The gap is in the cue, not in the feature.

"We're running on external triggers only. Turn notifications off and half the visits vanish."

Users haven't formed an internal cue yet.

"Put the document they left half-finished in the dormant-user email instead of the feature news."

A cue has to be about the recipient's situation, not the sender's announcements.

Pagedeck left the screens for version comparison untouched when it went after that 4 percent. It built one cue instead: when two or more people edit the same document on the same day, a line appears at the top of the document offering to show what changed. Not a notification and not an email — a signal visible only to someone already in that situation.

A month later usage stood at 23 percent. The feature and the screens were the same; the only change was where the cue sat. Mia now designs the cue alongside every new feature.

The most common mistake is reading trigger as a synonym for notification. Most visible cues are notifications, so it's an easy slide. But a notification is one kind of external trigger, and the kind that produces users who stay is the internal one. A team that equates the two ends up managing notification counts and click rates, and mistakes a state where visits collapse the moment sending stops for an improvement.

Assuming more cues means more usage is the second mistake. Add one and watch visits rise a few times and the idea forms naturally. In practice cues eat each other's value. A product going from three notifications a day to six sees per-message click rate fall by more than half, total visits rise slightly, and the share of users who turned notifications off rise alongside. What's worth increasing isn't the number of cues but how well a single cue matches the action it points at.