Your list is right. Forty-one rows open. Eleven of those rows are things other people owe you, and four of them come due this week. Every date on it is correct. Then Thursday arrives and the list stays closed. A client calls at 8:30 about something else entirely. The call runs to 9:40. By the time it ends there are nineteen new messages and somebody in the doorway holding a printout. The list was accurate the whole time. It simply sat there, because a list waits to be opened.
The chapter before this one ended on the distance between knowing when to chase and being told when to chase. This one is about the telling. What reaches you, on what day, carrying what, so a wait survives a week like that.
- A list answers a question you have to remember to ask. A bad week is exactly the week you forget to ask it.
- The day to chase sits earlier than the day the thing is due. The gap between them is arithmetic you can do in advance.
- A reminder you can act on inside a minute gets dealt with. One that takes longer becomes a small new job.
- Reminding the person you are waiting on is a different instrument with its own evidence, and that evidence says the wording matters far less than the template pages imply.
The difference between a record and a prompt
A list is a record. Its job is to be correct, and a good list is correct. It holds what you owe and what you’re owed. It holds the dates, the conditions of satisfaction, and the sentence each row came from. Everything the first three chapters of this series asked you to write down is in there.
A record waits to be consulted. Consulting is an act, and an act needs your attention on the day. So the whole arrangement rests on one step that stays unwritten. On the Thursday, in the week that made the delegation necessary, you remember to look at the place where you wrote down what to look at.
A written plan does real work, and the evidence for it is worked through in email task management, without writing it down. What matters here is the half of the finding that usually gets left off. In E. J. Masicampo and Roy F. Baumeister’s studies, an unfinished task kept working on people after they wrote it down. What settled it was a specific plan. Letting people formulate specific plans for their unfulfilled goals “eliminated the various activation and interference effects” that the unfinished task had been producing, and their account of why is that the pursuit is suspended and taken up again at the time the plan named.
Read that as a trade, because that’s what it is. You stop carrying the thing around. In exchange, something has to happen at the time you named. When the hour passes quietly, you’ve stopped worrying about the problem while the problem stands. That’s a worse position than the one you started in, and it feels considerably better.
So a reminder is the half of the arrangement that makes writing things down safe. It has three properties worth arguing about. When it arrives, what it says, and where it lands.
The day to chase is earlier than the deadline
Almost everybody sets the follow-up for the day the thing is due, and almost everybody half knows how that goes. A reminder that arrives on the due date isn’t a prompt to chase. It’s a notice that you’ve already missed.
The day you want is the due date minus three intervals, and you can estimate all three.
How long they take. The person, rather than the task. Some clients answer inside an afternoon. Some take four days. One takes four days and then asks a question. What ordinary reply times actually look like, and why the mail that asks for something is the slowest of all, is the subject of unanswered emails, and the threads that come back. The number to use here is what this person has done the last three times.
How long you take after it lands. The thing arrives, and then your work on it starts. A signed form has to be checked. A schedule has to be read into the file. A report has to be looked at properly by somebody who might disagree with it.
What happens if the answer is wrong. The margin most people skip. What comes back is the old version, or unsigned, or an answer to a different question. Then you need a second pass at it, and the second pass is where the whole thing either survives or falls over.
Tax practice has the cleanest example of this in American professional work, because the last step is a legal precondition rather than a preference. Publication 1345 is the Internal Revenue Service’s handbook for authorized e-file providers, and it’s specific about it. Anytime an electronic return originator enters the taxpayer’s PIN, the originator “must, prior to submission of the return, complete an IRS e-file Signature Authorization form which must be signed by the taxpayer.” The order is fixed. The originator may enter the PIN early, “but the taxpayers must sign and date the appropriate form before the ERO originates the electronic submission of the return (or extension form).”
Notice what that does to the arithmetic. The deadline you’re working to is a filing date that holds firm. Before it sits your transmission. Before that sits a signature traveling back from a person who sets their own pace, by whatever means they choose. The same handbook lists those means as hand delivery, mail, private delivery service, fax, email or a website. Each of them is a different number of days. So the reminder to chase that signature belongs well ahead of the filing date. Take the filing date, then subtract the client’s return trip, your review of what came back, and one full cycle in case the form is signed in the wrong place.
Most of what you’re waiting on sits outside any statute, and the same subtraction still applies. Usually it goes undone, so the follow-up lands on the day of, and the chase turns into an apology.
One consequence is worth saying out loud, because it runs against the instinct. The more important the deadline, the earlier the reminder should fire. That’s the opposite of how people set them. A trivial wait can be chased late, because being late on it costs a shrug. The wait with a hard date behind it is the one that needs a prompt while there’s still room for the answer to be wrong.
What actually sets that date
Four things move it, and urgency sits outside all four.
Who you are waiting on. A colleague eight feet away is one wait. A county surveyor is another, and so are two different clients. If you’ve been closing rows deliberately since the last chapter, you already know who returns things two days early and who has to be asked twice. That’s the single best input available.
How big the ask is on their side. A question answerable in one word comes back quickly or stays silent, and a week of silence there means something. A request that needs somebody to pull three files and reconcile them runs late at a different point. Chase that one early and you buy a status update rather than an answer.
Whether the date behind it can move. A board meeting on the twelfth holds. An internal target you set yourself bends. Treat the two the same and you end up chasing everything at one pitch, which teaches people to read all of it as soft.
Whether there’s another way to get the fact. The portal shows you the filing status. The bank feed shows you the payment. Then the reminder is a prompt to go and look. It’s free to everybody else. It can fire much earlier and more often than a message ever should.
Naming these has a narrow purpose, and computing a date for every row is no part of it. All four are known at the moment you set the wait, in the same half minute as the name and the deadline. Three weeks later they’re almost impossible to reconstruct. That’s the same argument the previous chapter made about the handoff, applied to the one field it left out.
What a reminder has to carry
Here’s the part that quietly decides whether the whole system holds.
A reminder is a message from you to somebody with less information than you have. That somebody is you, on a Thursday, between two other things. The context of three weeks ago is gone. Write it the way you’d write to a colleague you respect and whose time you value, and it gets acted on. Write it as a memo to yourself and it gets postponed, because you’re the one who has to reconstruct what it meant.
Five things, and a good one carries all five in a line or two.
What you asked, in the words you asked it. The whole sentence, rather than “Hallett file” or “chase Marcus.” A row that reads ambiguously three weeks on is a decision you have to make again from scratch, and the decision you actually make is to look at it later.
Who owes it. One name, for the reason the first chapter gave.
The date behind the date. The reminder fires on the eighth. It fires on the eighth because the board meets on the twelfth. Carry both dates and you can decide in two seconds whether this one can slip. Carry only the eighth and it sits there, impossible to weigh against anything else on your Thursday.
What you would do about it. Written when you set the wait, while the decision is still cheap. Call the office, send the second note, go ahead anyway and say so. Under time pressure that same decision gets expensive, which is the whole reason it belongs in the reminder rather than in the moment.
One step back to the thread. The conversation itself, one step away. Checking whether they answered while you were looking elsewhere costs ten seconds, rather than a hunt through your sent mail.
The measure for all of it is a single question. Can you act on this reminder in under a minute, with everything you need already in front of you? A reminder that clears that bar gets dealt with on the day it arrives. One that misses it becomes a decision to look at it properly later, and later is a word this series has already spent three chapters on.
Where the reminder should land
The last property is placement, and it settles whether the rest of this holds up in a real week.
A reminder about an email should arrive where the reply arrives. That sounds obvious, and it goes the other way routinely. The natural place to put a follow-up is wherever your list lives, and for most people that’s somewhere else: a task app, a calendar entry, a note. The moment the reminder is in one place and the answer is in another, you’ve given yourself a reconciliation job. Every prompt now costs a context switch to check whether it’s still true. Prompts that cost a context switch get batched, and a batched chase is a late chase.
The mainstream instruments each solve part of it.
A snooze puts the message back in front of you at an hour you name. The shape is right and the condition is wrong. It returns on the clock, whatever the other person has done, and it stays entirely private to you. Snoozed email, and where later actually goes takes that apart, including the failure where you snooze the same thread four times because you’re waiting on an event and keep expressing it as a date.
A flag with a reminder date sits closer to a list than to a prompt, and it inherits the list’s problem. It marks the message. Then it waits to be looked at.
An add-on that returns a sent message if nobody replied is the explicit version of the right idea, and it works. It covers whatever you thought to mark on the way out, which is a real limit rather than a quibble, and Point compared with Boomerang works through where that lands.
Gmail’s nudges are the version most people already have without asking for it. Google’s own help page describes them plainly: “At the top of your inbox, you may see suggestions (or ‘nudges’) for emails you should reply to or follow up on.” The setting behind the second half of that is described just as plainly: “At the top of your inbox, you can find sent emails that you may need to follow up on” (checked September 6, 2026). One more line from the same page is worth knowing before you rely on any of it. Conversation view has to be turned on before the reminders appear at all.
Nudges are a real feature and they catch real things. What they offer is a guess about your sent mail, rather than a record of what you’re owed. The system decides which sent messages look like they need following up, and it decides when. The surveyor’s confirmation and a note to a mailing list read the same to it. There’s nowhere to say that the board meets on the twelfth, so this one has to surface on the eighth. A nudge arrives with the reason left off, so you supply the reason yourself every time one appears. That’s a fair description of the whole category, and of why putting a reminder on an email so often comes down to a snooze and a hope.
The reminder that should not have fired
Two failures matter, and they are opposite.
The first fires at an hour when the chase is out of reach. Friday at 4:40, when the client’s office has gone quiet for the weekend. The morning you spend at a client’s premises with your phone in a drawer. The week you’re away, which produces the particular species of prompt that arrives, gets dismissed, and takes the wait with it. A reminder is a date on which the chase is possible. A system that fires blind teaches you to dismiss things by reflex, and that habit is what kills every one of these arrangements.
The second is the reminder you keep pushing. Three or four repeats of the same prompt is a diagnosis rather than a discipline problem, and the diagnosis is that you set a date to stand in for something that was always an event. A confirmation landing, a portal changing state, a document arriving: every hour you pick for one of those is a guess, and it keeps arriving ahead of the event. What that wants is a standing watch instead of another date, which is setting your first watch in plain words.
Both failures spend the same currency, and it’s the scarce one here. A reminder is a claim on your attention that you granted in advance. The granting stays rational while the claims turn out to be worth it. Two or three wrong ones in a month and you start reading the prompts the way you read a smoke alarm at a hotel, which is to say you stop reading them. Whether ten of them arriving together does the same damage is where this series ends, in one alert instead of ten.
Reminding the person you are waiting on
Everything so far is about a reminder aimed at you. The other kind points outward, and in a working week it turns up as the same job repeated. The survey sitting unfilled. The client questionnaire. The annual confirmation request. The list of items you asked for and got two thirds of. One message, many recipients, most of whom simply haven’t gotten to it yet.
There is more evidence about this than about almost anything else in the series, because researchers have to solve it to get their own work done. The Cochrane methodology review Methods to increase response to postal and electronic questionnaires pools 758 randomized trials. It’s by Philip Edwards, Ian Roberts, Mike Clarke, Carolyn DiGuiseppi, Benjamin Woolf and Chloe Perkins. The Cochrane Database of Systematic Reviews published it in 2023, with its evidence current to December 2021. It reviews research questionnaires rather than business mail, and it’s led from the London School of Hygiene and Tropical Medicine, so what it offers an American practice is direction rather than rule. Take the direction. Its own plain-language summary is honest about the range it covers, which runs “from patients, doctors, university students, and professors, to marketing managers, accountants, and grocery store managers.”
Three of its results are worth carrying.
Reminding works, and changing the channel works far better than changing the wording. Four trials with 15,143 participants compared a telephone reminder to no reminder. The odds of a response roughly doubled, at an odds ratio of 1.96, with a confidence interval running from 1.03 to 3.74 that is wide enough to be honest about. A telephone reminder added on top of a postal one, in four smaller trials, increased the odds by more than half again.
Sending more of them buys much less than you would think. Six trials with 7,520 participants compared a higher-frequency follow-up interval to a lower one. The odds of response went up by about a tenth: 1.13, from 1.02 to 1.25. Real, and roughly a tenth as valuable as picking up the phone once. That’s the single most useful sentence in this whole section, because the instinct when a reminder fails is always to send another one sooner.
The wording tricks are thinner than the template pages imply. A trial with 231 participants tested “Action Required” as the subject line of an email reminder against “Questionnaire reminder.” No evidence of an effect, at an odds ratio of 0.61 with a confidence interval from 0.23 to 1.63. One small trial with an interval that wide leaves the question open in both directions. What it settles is where your afternoon should go, and the subject line is a poor place for it. Brevity did show up: for electronic questionnaires the review found the odds of response tripling with a brief covering letter rather than a detailed one.
There’s a cost on the other side that the response rate hides from you. Mette Trier Damgaard and Christina Gravert ran a field experiment with a charity and reported it in the Journal of Public Economics in 2018 as The hidden costs of nudging. Reminders worked, and they found “reminders increasing intended behavior (donations), but also increasing avoidance behavior (unsubscriptions from the mailing list).” Charitable giving sits a long way from a client relationship, and a mailing list sits a long way from your client list. The shape transfers anyway. The people who respond to your third reminder are visible. The ones who quietly decide to deal with you less stay out of sight.
So the practical version, for a survey or a questionnaire or a list of outstanding items, is short.
Send one reminder, three or four days after the original rather than three weeks. Send it to the people still outstanding and to them alone. Make it shorter than the first message, and put the actual ask in it rather than a reference to the ask. Say what the deadline is and what happens after it. Then, if that one falls flat, change the channel instead of the schedule. Call the four people who matter rather than mailing the forty again. And keep the record that you asked. The question that comes up later is whether anybody chased, and a dated line answers it before the sentence is finished.
One caution belongs here rather than at the end. A reminder sent to a room is a weaker reminder than the same words sent to one person, for the reason set out in the first chapter. If four of the forty genuinely matter, give those four a message each with their names on it. And the one-to-one case, where a real relationship is on the other end and the second message has to do work a template can only gesture at, is the follow-up email that gets answered. Chasing documents specifically, where the reason it’s still outstanding is usually the whole problem, is managing client document requests without the chasing.
What Point does with a date you never set
Point is the mail client itself, which settles the placement question by construction. The prompt and the reply arrive in the same feed. Checking whether a chase is still on is reading, rather than switching.
Say on the thread, in whatever words you’d have used anyway, that the next move is theirs. The wait is written down with a date on it, alongside everything you owe. Point works that date out from the thing itself, and it lands where chasing still leaves room, ahead of the day it’s due. When the day comes, the thread rises back to the top of your feed. The conversation sits one step away, and the ask is there in the words it was made in. Writing the message stays yours, which after the section above is where it belongs.
The same holds for work you handed to a colleague in the team chat, which sits in that feed too, and for what a client, a vendor or another firm owes you. All of it gathers into one list, across every conversation and every business you run. That matters most for exactly this, because a second list is the one that stays closed on the Thursday.
Where the thing you need is an event rather than an answer, say that in plain words and you get a watch instead. Point speaks once, when the thing happens, in place of a guessed date every few days. Setting your first watch in plain words is how that is written.
Every type of action carries its own level for how far Point may go without you, from suggest and stop, through prepare it and leave it for you, to get on with it. Each one begins in the middle position, so moving any of them up is a decision you go and make. Point’s work is recorded in order, in sentences you can read straight through, and taking an action back begins at the entry that recorded it. The one exception is mail already delivered, which is sitting on another company’s server and past recall. The rest of the inventory is on the benefits page.
Deciding that a wait is worth chasing stays with you. Some things you’re owed should be allowed to die, and that judgment stays exactly where it was.
The Thursday it is wrong
So the arrangement works. Three prompts on Thursday morning. The signature still out, seven days from a filing date, with the client’s usual four-day turnaround already subtracted. The surveyor, on the day you and Marcus agreed his silence would start to mean something. And the March invoice question, quiet for twenty-four days.
Two of those are right, and you deal with them before 9:00.
The third is the invoice question, and the client answered it on Tuesday. She answered by phone, which is why the thread stayed quiet. She called about something else, mentioned it in passing, and said go ahead at the original figure. You did, and the invoice went out that afternoon. The row is still open because closing it was a separate act, and that one stayed undone. So a system you’ve started to rely on has told you to chase a client for something you already have.
That costs more than the thirty seconds of checking. It’s the first entry in a ledger you’ll keep without meaning to. Four or five entries in, you start opening the prompts with a question rather than an intention. Every arrangement in this series depends on the list being right about what’s still open, and the list has just been wrong in the only direction that matters.
Which turns out to be a solvable problem, and the solution sits well away from more discipline about closing rows. It is the software noticing what you already did.