Notification Preferences and Billing You Don't Have to Fight

Two things shipped recently that nobody would put on a landing page. Detailed notification preferences, and a billing flow that lets people cancel without calling anyone.

They are worth writing about anyway, because they are where trust is won or lost.

Alerting only helps if it survives contact with real life

An alert system can fail in two ways. It can miss things, which is the obvious one. Or it can be right, too often, at the wrong hour, until somebody mutes it.

The second failure is more common and much harder to spot, because from the inside everything looks fine. Alerts are firing. Delivery is succeeding. Nobody is reading them.

So the controls we built are about letting families stay reachable on their own terms rather than ours.

Channel selection per severity. SMS, push and email can each be assigned to whichever severity levels warrant them. A sensible default is Critical by SMS, Warning by push, and Informational as an email digest, but the point is that the family decides rather than us.

Quiet hours that Critical breaks through. This part has to be exactly right. If quiet hours hold back everything, they are dangerous. If they hold back nothing, they are pointless and people switch notifications off entirely. Critical alerts ignore quiet hours by design, and everything else waits for morning.

Per guardian settings. The sibling who wants everything and the sibling who wants only emergencies are both making a reasonable choice. Forcing one policy on a circle guarantees that at least one person is either overwhelmed or under informed.

The principle underneath is that the goal is not maximum delivery. It is alerts that still get read in month six.

Billing, without being clever about it

Trial management and the full subscription lifecycle are in, including cancellation, self service, in the same place as everything else. No retention phone call, no support ticket, no flow that buries the cancel button three screens deep.

This is deliberate, and worth explaining, since the short term incentives point the other way.

We are asking families to put sensitive information about a vulnerable parent into this product. That request only makes sense if the product is plainly trustworthy in the places where trust is cheap to check, and billing is the cheapest place anyone can check. A company that makes cancelling difficult has told you something about how it will behave in the places you cannot easily check.

Making it easy to leave is a fairly direct way of saying we would rather earn the renewal.

Where this stands

Both are complete on the backend with interfaces shipped. There is a notification preferences page covering the full severity by channel grid and quiet hours, and a plan management page covering trial status, current plan and cancellation.

If either one is confusing in practice, that is exactly the feedback we want now, while changing it is cheap.