What 2,002 reminders say about what people forget
Reminder apps are sold as a way to manage the future: the dentist in March, the renewal in November, the birthday you always miss. So when we counted what people actually set on Pokobot, we expected a calendar.
We got a scratchpad instead. Between 4 July and 30 August 2026, 92 people set 2,002 reminders. Of the one-off ones with a future due time (450 of them, from 79 different people), 87.8% were due within twenty-four hours of being typed, and half were due within three. Almost nobody is using this to remember next month. They are using it to survive the next few hours.
Here is the whole dataset, including the parts that undercut the headline.
The shape of the data
Everything below comes from Pokobot's own database on 30 August 2026, covering 4 July to 30 August: 2,002 reminders from 92 people who set at least one. It is a small dataset from a young product, and we would rather show you its edges than round them off.
Two splits matter for reading the rest. A reminder is either one-off (“call the landlord at 4”) or recurring (“every day at 9pm”), and those turn out to be almost different products used by almost different people. 455 reminders were one-off, 1,547 repeated. But 79 of the 92 people set a one-off, while only 28 ever set anything recurring. The recurring pile is big because a small group leans on it hard, not because most people do.
Almost nothing is set more than a day ahead
Take the 450 one-off reminders with a future due time, set by 79 different people. This is the broadest, least skewed slice we have, and it is startlingly flat against the near end of the clock.
34.9% are due within the hour. 50.7% within three hours. 79.1% within twelve. 87.8% within a day. The median gap between typing a reminder and it arriving is about three hours, and that number barely moves (to 3.7 hours) if you drop the single heaviest user entirely.
At the far end, the drop is near total. Out of 450 reminders, exactly 10 were set more than a week out. Ten. The future beyond next Tuesday is, empirically, not what people are using this for.
remind me to call the plumber back in 2 hourstake the clothes off the line at 6text mom before her flight at 9:15The recurring reminders are a different product
The other 1,547 reminders repeat, and their distribution is lopsided in a way worth naming. 850 are daily. 675 fire more than once a day, some every twenty minutes, and those 675 come from just 12 people. Weekly and monthly barely register: 15 and 7 respectively.
One habit dominates. 594 reminders (29.7% of everything) are some version of “drink water,” set by 27 different people. Hydration is not a rounding error in this dataset, it is the single largest category, and any statement we make about recurring reminders is partly a statement about people trying to drink more water.
That is also why weekly and monthly look so thin. If you only need something once a month, a calendar already handles it and you never open a chat to ask. The reminders that arrive here are the ones a calendar is too slow or too formal to catch.
drink water every 2 hours between 9am and 7pmstretch every 45 minutesmedicine every day at 9pmNobody schedules anything for 3am
Sorted by the hour they are due, the reminders trace an ordinary waking day. Only 54 of 2,002 fall between midnight and 6am, which is 2.7% across six full hours. Volume climbs from 6am, peaks around noon, dips mid-afternoon, and rises again to a second peak at 9pm.
This sounds obvious until you build a bot that repeats things every twenty minutes. A fast repeat set at 2pm is useful; the same repeat at 2am is an alarm clock nobody asked for. 493 reminders (roughly one in four) now carry a daily active window for exactly that reason, so a fast repeat goes quiet overnight and picks up in the morning.
The feature exists because the data said the night was empty on purpose.
What we think this means
The reminders people set are overwhelmingly things they already know they have to do, and will still forget, within the next few hours. That is not a planning problem. It is a working memory problem, and it shows up as a very specific behaviour: a thought arrives mid-task, gets typed in eight words, and comes back at the moment it is useful.
It explains why the channel matters more than the parsing. A reminder due in three hours has to interrupt a day already in progress, which a notification in an app you opened twice this month cannot do. The reminders in this dataset arrive as WhatsApp messages, in a thread that is already open.
It also explains what people do not use it for. There is almost no long-range planning here, no project structure, no nested lists. The median person in this dataset set three reminders in two months. They are not power users building a system. They are people who did not want to lose one specific thing.
What we are not claiming
This is 92 people over 58 days, and the honest caveats are large enough to state plainly.
The sample is concentrated: the five heaviest users account for roughly half of all reminders, and the single heaviest for 316 of them. It is geographically narrow, with the overwhelming majority of accounts in one timezone. Every percentage here is at the reminder level, never the person level, because the two tell different stories and the reminder-level one is flattered by whoever was most active that week.
We are also not showing you retention, completion, or category mix as headline numbers, because at this sample size each is fragile to a handful of accounts going quiet. The one finding we would defend is the lead-time one: it holds across 79 separate people and survives removing the heaviest user, which is more than we can say for the rest.
Questions, answered
- How far in advance do people set reminders?
- Barely any distance at all. Across 450 one-off reminders set by 79 people on Pokobot, the median gap between creating a reminder and it firing was about three hours. 87.8% were due within twenty-four hours, and only 10 of the 450 were set more than a week ahead.
- Do most people use recurring reminders?
- No, though they generate most of the volume. Only 28 of 92 people ever set a recurring reminder, but those repeats account for 1,547 of 2,002 reminders. A small group of heavy repeat users produces most of the traffic while the typical person sets a handful of one-offs.
- What do people set reminders for most?
- Drinking water, by a wide margin: 594 reminders from 27 different people, nearly 30% of everything in the dataset. After that it thins out quickly into medicines, exercise, calls and follow-ups, chores, and work check-ins.
- Is this dataset big enough to draw conclusions from?
- For the lead-time finding, we think so, because it comes from 79 separate people and holds up when the heaviest user is removed. For anything about categories or retention, no. Those are dominated by a few accounts, which is why they are described here rather than published as headline statistics.
- Can I set a reminder for a few hours from now?
- That is the most common thing people do here. Message Poko on WhatsApp with something like “remind me to call the bank in 2 hours” and it arrives as a WhatsApp message. It is free for 7 days with no card and no signup.