Your first morning already does something for you. The mail is weighed and in order when you sit down. Threads carry a summary you can act on before you open them. One or two things you were waiting on land, and you hear about them without going to check. The hour back comes later. You’re also slower at everything this week, and that’s the part that makes a good first week read like a letdown.
- The first value costs you nothing. Ranking, summaries and a tidied contact list run off the history already in your mailbox, and they’re ready the hour the reading finishes.
- Anything that needs a new habit from you arrives later, and that’s the normal shape of it. Habits take far longer than seven days, and the research puts numbers on how much longer.
- Week one comes out roughly flat on time, by design. You’re learning where things are, and that costs about what the sorting saves. Judge the week on evidence.
- Ten minutes on the first afternoon moves the week further than anything else you can do. A list of names. One thing to watch for. One sentence about how you like things done.
- Take two numbers and a list before you connect. A first week needs something to measure against, and memory is too rough for the job.
What works before you have taught it anything
Value arrives in three tiers. They’re sorted by how much of you each one needs.
Tier one runs off your history. Everything needed to put your mail in a sensible order is already sitting in the mailbox. Who you answer within the hour, and who you answer in three days. Which threads you always open. Which sender has written four hundred times, and which one wrote once last April. All of that is a reading of what you’ve already done, so it’s ready before you say a word. A thread summary works the same way, off the thread itself. So does merging the four versions of the same client that piled up across nine years of address books. This tier lands on the first morning, as soon as the reading-in finishes, and it’s most of what a first week actually delivers. How the ordering is arrived at is how AI email triage works, and what a summary is for before you open anything is reading the gist before you open a thread.
Tier two needs one thing from you, and it takes a minute. The names where being late costs you. The one thing you keep checking for. The way you sign off to clients. Each one is a sentence from you and a week of return. That’s why the first afternoon is worth more than the rest of the week put together.
Tier three is built out of watching you work. Replies that arrive sounding like you rather than like a template. Knowing which threads you file unread. Knowing that the message from the bank is noise and the near-identical one from the same bank matters. This tier is made from your corrections. It starts from scratch on Tuesday, and it’s comfortably better by the end of the month.
Almost everything written about an AI inbox describes tier three, including the writing by the people who make them. That’s the interesting part, and it’s the part that arrives after week one. Get the tiers straight in advance and your first week reads correctly. It’s the tier one week.
The week you are also slower
Here’s the arithmetic most pages leave out. The sorting saves you something on Tuesday morning. Hunting for the archive control costs you something back. So does looking for the search box, and reaching twice for a keyboard shortcut that does something else now. In week one that second number is large. It shrinks quickly, and it’s still with you on Friday.
So the honest expectation for week one is flat. Maybe a shade under flat. What switching email tools really costs budgets four to ten hours per person of working slower, spread over two weeks rather than lost in a block. Those are the same hours, counted from the accounting side.
The shape has a name, and it’s been measured at a scale far above a six-person practice. Erik Brynjolfsson, Daniel Rock and Chad Syverson call it the productivity J-curve. Their paper opens on the reason for it: “General purpose technologies (GPTs) like AI enable and require significant complementary investments. These investments are often intangible and poorly measured in national accounts.” That mismeasurement, they argue, “can lead to underestimation of productivity growth in a new GPTs early years” (The Productivity J-Curve, American Economic Journal: Macroeconomics, January 2021). They were adjusting national accounts rather than advising firms of six, so borrow the shape and leave the scale. What the shape says is that the dip is the investment, not the verdict.
In a small practice that intangible investment is easy to name. It’s knowing where things are. It’s the names you wrote down and the corrections you made. It’s real work, it’s invisible on every ledger you keep, and it’s finished far sooner than it is for an economy.
Two practical consequences. Leave a flat first week out of the partners’ meeting, because a week of paying tells you very little either way. And put the week you connect somewhere quiet. A week of being slightly slower is cheap in June and expensive on April 10.
Two numbers and a list, before you connect
A first week is worth most when it produces a comparison. Otherwise Friday leaves you with a feeling, and a feeling loses to whatever you feel in November.
Write it down for one reason: recall is genuinely rough about this. A meta-analysis pooling 106 effect sizes set what people said about their own screen time against what their devices had logged. It concluded that “self-reported media use correlates only moderately with logged measurements” and that “self-reports were rarely an accurate reflection of logged media use,” at a correlation of r = 0.38 (Nature Human Behaviour, 2021). That research covers digital media use in general rather than email in particular. It’s also the closest thing there is to a direct answer on whether people can estimate their own time at a screen, and the answer is a shaky one.
So capture three things, on two ordinary mornings before you connect anything. Twenty minutes total.
How long the first pass takes. Start a timer when you open your mail. Stop it when you know what today asks of you. That’s the moment to catch, rather than the empty inbox, which is a different and less interesting event. Use two mornings rather than one, because Monday runs differently everywhere.
How many threads you opened only to find out whether they mattered. A tally on paper, one mark each. This is the number that moves most in a first week, and the one almost everybody skips, because opening a message to see what it is barely feels like work. It’s the whole cost of an unsorted inbox, and you can count it.
Everything you’re waiting on right now. One line each: who, what, and when you asked. It’s a list rather than a number, and it’s the most valuable of the three. The section after next says why.
Deciding in advance what would make the answer yes is a separate discipline, and it belongs to trying a new inbox without the risk. That’s the question you’re testing. These three are what you test it against. For the wider version of the same measurement, taken across a firm rather than one person, there’s how much time firms lose to email.
One caution about the timer, so Friday reads fairly. In week one you’re slower at navigating and faster at deciding. The two pull in opposite directions inside the same number. That’s why the tally and the waiting list are the honest pair.
Ten minutes that bring the rest of it forward
Three inputs, in order of what they return per minute spent. All three belong on the first afternoon, because each one changes what happens to mail arriving after you enter it.
The names. Ten minutes with a piece of paper. Write down the people whose mail has to reach you. History gets you most of the way there on its own. The part it leaves out is the part that matters: the new client who has written twice, the attorney on a deal still in flight, the person at the state board. Those names carry the least history and the highest cost of being late, and that combination is one a system has to be told. Marking the people you cannot miss is the fuller version, and the day-by-day walkthrough is day four of the first two weeks.
One thing to watch for. Take the top line off the waiting list you wrote. Hand it over as a standing request, in one sentence: tell me when the signed engagement letter comes back. It costs ten seconds, and it retires a checking habit you’ve been running for weeks. Details are in telling your inbox to notify you when.
One sentence about how you like things done. The thing you’d tell a new assistant on their first morning. Client replies get my cell number. Keep my mornings clear until ten. This is the first deposit into tier three, and day one is the day to make it, because everything after it inherits it.
What to leave alone is just as specific. Leave a setting where it arrived until you’ve watched it fail, because in week one that decision is a guess, and Thursday hands you a better one for free. Leave your filters, labels and tidying exactly as they are, for the reason the switching cost page gives: one new variable you can diagnose, and four you can only guess at. And let the tier three features arrive in their own time. They’re coming, and they come at the same speed either way.
The win you will not notice happening
The largest thing a first week does for you is an absence, and absences stay quiet. A message that would have sunk stayed where you could see it. A chase you’d have made on Thursday was already settled, because the thing landed on Wednesday and you heard about it. A forty-message thread you’d have opened before a call, you read as one line instead, and you were right.
All of that passes without a mark on the day. Skipped work is hard to remember, which is why people who have had a genuinely good week describe it as fine.
Two ways to see it anyway, both cheap.
Read the waiting list on Friday. The one you wrote before you connected. Go down it, mark what closed, and beside each closed item mark whether you had to push it. The items that closed on their own are the clearest evidence a first week produces, and counting them takes two minutes. Chasing what has gone unanswered is the mechanism underneath, when you want to know how the loop got closed.
Scan the old inbox once, top to bottom. Ten minutes at the end of the week, in the app you still have installed. You’re looking for one thing: anything you’d have handled this week and missed. This is the honest miss-check, and an empty result is worth more than a week of feeling good. Once a week is the dose, and more often than that is a checking habit under a new name. How much watching is enough before you stop is AI email triage, and whether you can trust it, and the running record you can read instead of watching is the activity log behind every action.
Reading a flat week correctly
Suppose Friday comes and the week feels quiet. That has four causes, and each one calls for a different response, so it’s worth working out which one you have.
It’s early, and what you were waiting for was tier three. The commonest case by a distance. The drafts learn you over weeks, and Friday is too soon for that. Response: sit tight. Read the tally and the waiting list rather than the feeling.
Everything was set to suggest. With the software only showing you what it would have done, you did the entire week’s work yourself. That’s a finding about the setting, and the setting is yours to move. Trying a new inbox without the risk makes that argument in full, and setting how much your inbox does on its own is what the positions mean.
You changed four things at once. New app, new filters, tidied labels, new booking link. Every result now has four possible causes, and a week like that stays unreadable. Response: hold everything still, and give it a clean week.
Your inbox was already fine. Some firms genuinely have this handled. If everything reaches you, and you get to the end of the day’s mail calmly, then a flat week is the correct and permanent answer. The time you’d recover is already yours. That case is priced honestly in what switching email tools really costs, and the answer there is to keep your money.
Flat and wrong look different, and it’s worth knowing what wrong looks like. Wrong is a client landing below a newsletter twice after you’ve corrected it. Wrong is a summary you know to be inaccurate, on a thread you sat in the middle of, where you already knew what it said. Those are verdicts, and they arrive early. A quiet week is something else.
What the first week is buying for month three
Every switching estimate carries an hour or two of teaching. That’s the line on the bill that returns more than it costs, and almost all of it is spent in the first month. Week one is when you’re least fluent and most likely to disagree with something, which makes it the richest week for corrections you will ever have. Corrections are how tier three gets built, and when a summary learns from the corrections you make is the mechanism underneath that.
One distinction is worth making here. It separates a correction that compounds from one that stops at a single thread.
A fix moves one thing. You drag a thread back into the feed because it belongs there. Good, necessary, and it teaches very little, because a single instance is ambiguous. Was it the sender, the subject, the timing, the client?
A deposit names the reason. Anything from this client goes to the top, whatever it’s about. Keep the state board in the feed. Sign client replies with my cell number. These are sentences rather than gestures, and they apply to every message that has yet to arrive.
Make five deposits in week one and month three is a different month. Make fifty fixes and you’ve had a tidy week. The whole difference is the habit of saying why, and the days when everything still surprises you are the days to build it.
There’s an honest limit on what a week can do to how you work. In a University College London study, 96 volunteers each took on one new daily habit in a fixed context, and were tracked for twelve weeks. “The time it took participants to reach 95% of their asymptote of automaticity ranged from 18 to 254 days” (European Journal of Social Psychology, 2010). The study was about eating, drinking and exercise rather than email. The useful part is the spread rather than the numbers: even the fastest person took eighteen days, which is more than double a first week. The new way still feels new on Friday, and the study is why. So judge your first week on what happened while you carried on working exactly as you always have. That’s tier one, and this is the week it gets tested.
The first week, in order
- Two mornings before you connect, capture the three baselines. The timed first pass, the tally of threads you opened to find out if they mattered, and the list of everything you’re waiting on.
- Connect on a quiet afternoon and then leave it be. The reading-in takes a while, and what it’s doing shows up tomorrow.
- Spend ten minutes on the names, including the ones with barely any history. Those are the ones history has to be told about.
- Set one watch, taken off the top of your waiting list.
- Write one preference sentence, the thing you’d tell a new assistant on their first morning.
- Leave everything else exactly as it is all week. Filters, labels, booking links, and any setting you have yet to watch fail.
- Keep the old app installed with its notifications off. One source of alerts, two ways to look.
- Make deposits rather than fixes. When you disagree with something, say why, in a sentence that applies to the next one.
- On Friday, read the waiting list. Mark what closed, and mark what closed on its own.
- Scan the old inbox once, top to bottom, looking for anything you’d have handled and missed.
- Write two lines down. Where the timed pass sits now against the baseline, and where the tally sits now. Those two lines are the whole output of the week, and in eight months they save you starting this from scratch.
What Point does in your first week
Point reads the Gmail or Microsoft 365 mailbox you already have, so your first afternoon is a secure sign-in rather than a setup. Your mail stays where it is, your address stays the same, and there’s nobody to tell, which is the argument of why a new inbox never means a new email address. Then Point reads back through what’s already there. That’s the wait, and the wait is the whole of day one.
The tier one week shows up the next morning. What needs you today sits at the top of the feed, ahead of whatever happens to have arrived most recently. Every thread carries a plain summary you can act on before you open it. Duplicate contacts get merged, the same client under four addresses into one. All of it is a reading of mail you already had, so all of it is ready before you say a word.
Two things make that visible from where you already work. The judgment Point forms is written back as ordinary Gmail labels, Now, Later and Other, and as Outlook categories on Microsoft 365. So you can read the sorting from your real mailbox as easily as from a new app. Filing runs both ways too: archive a thread in Point and it leaves the inbox on the provider’s side as well, still there if you go looking for it. That’s what makes Friday’s scan of your old inbox a genuine check rather than a formality, and it’s set out properly in how Point writes back into your real inbox.
The tier two inputs are the ten minutes this page argues for. You mark the people who matter. You ask to be told when something lands, in a sentence rather than a rule builder, and you hear about it the moment it does. You say once how you like things done, and it’s kept as a plain line you can read and change later, rather than a setting three menus down. If you’d rather have the week’s shape handed to you each morning, a short briefing gathers what needs you today, on a schedule you choose.
While that runs, every kind of work sits at review out of the box. Review is the position where the work is prepared and waits for your yes. Drafts wait for you by default, until you raise the dial for that kind of work yourself. What was done is listed in plain language, with a time against each entry. You can reverse what Point did from that same list, so a thread filed wrongly on Tuesday is back in the feed as soon as you say so. Undo covers what Point did, up to Point’s own side of the line, which undoing what Point did is direct about.
A second business stays in its own lane from the first morning. A firm adds seats one person at a time rather than all at once, which is worth knowing when you’re deciding whether week one is one person’s week or six people’s. Point is in private beta and priced per seat, so what that comes to is a question for the team rather than a number to read off this page. Everything Point does is the flat inventory, and it’s the right list to hold your own three baselines against. The paced, day-by-day version of everything above is your first two weeks with Point, which starts on the afternoon you connect.
Common questions
How long before a new email tool actually saves me time?
Something is saved on the first morning. Learning where everything is costs you about the same back, so week one comes out roughly flat. The gap opens in weeks two and three as the navigation cost falls away, and it keeps widening for a month or two as the tool learns your corrections. If you want one number to plan against, treat the first week as evidence rather than hours, and expect the first clear time difference somewhere in week two.
What should I expect on the first day?
A sign-in, then a wait while the reading-in works back through the mail already in your account. It’s a quiet day, and judgment comes later. Spend ten minutes of it writing down the people whose mail has to reach you. Set one thing to watch for. Say one sentence about how you like things done. Then leave every other setting exactly where it arrived.
Is it normal to feel slower in the first week?
Yes, and it’s the expected shape rather than a warning sign. You’re paying for the navigation you already have in the old app and are still building in the new one. The habit research says the same thing from the other end. In one study, the fastest anyone took to make a single new daily habit feel automatic was eighteen days. A week is a short run at feeling natural, whatever the tool is.
How do I tell whether it is working, rather than whether I like it?
Take three baselines before you connect, and read them on Friday. Time your first pass on two ordinary mornings. Count the threads you open only to discover whether they matter. Write down everything you’re waiting on. On Friday the tally and the waiting list do most of the work: the threads you got to skip, and the items that closed while you got on with something else. Liking it is a poor predictor in either direction.
What is the single highest-value thing to do on day one?
Write down the people whose mail has to reach you, including the ones with barely any history behind them. History supplies the regulars on its own. The new client who has written twice, and the attorney on a deal still in flight, are the two you supply yourself, and those are exactly the names where being late costs the most.
Should I change the settings in the first week?
Barely. A setting you change before you’ve watched something go wrong is a guess, and the evidence is a few days away and free. The one exception is anything obviously wrong on the first afternoon, such as notifications firing from two apps at once. Fix that, leave the rest, and in particular leave anything that sends exactly where it starts.
Does the first week look different for a firm than for one person?
The value arrives on the same schedule. What you’re watching is different. For one person the first week is about time and misses. For a firm it’s about whether handoffs stopped needing a conversation, and that shows up later, because it takes more than one person moving. The mechanics of getting six people there, and how to read week three rather than week one, are in how to roll out a new tool across your practice.
What if nothing happens at all in the first week?
Work out which of the four causes you have. It was early, and what you wanted was the part that learns you. Everything was set to suggest, so the week’s work stayed with you. You changed four things at once, and a week like that stays unreadable. Or your inbox was already fine, in which case a flat week is the right answer, and the honest response is to keep what you have.
The short version
- Value arrives in three tiers. What runs off your existing history lands on the first morning. What needs one sentence from you lands the same day. What’s built from watching you work arrives after this week.
- Expect week one to be flat on time. The dip is what buys weeks two and three, and it’s the same hours the switching estimate already budgets for.
- Write down two numbers and a list before you connect. A timed first pass on two mornings. A tally of threads opened just to find out if they mattered. Everything you’re waiting on. Recall is too rough to stand in for them.
- Ten minutes on the first afternoon does more than the rest of the week. The names, one thing to watch for, one sentence about how you like things done.
- The biggest first-week win is an absence, and it stays quiet. Read the waiting list on Friday, and scan the old inbox once for anything you’d have handled and missed.
- Say why when you disagree with something. A fix moves one thread. A sentence that names the reason applies to every message that has yet to arrive, and that’s what month three is made of.
- A quiet week is fine. A client ranked below a newsletter twice after you corrected it is the bad week, and that one arrives early enough to act on.