The place is the crossing.
Rosa made her call on the Thursday in March: this is a Ben one, he has the file. She made it where the work lived, in her inbox, and that was the only place it existed. To reach Ben it had to be carried by hand, across a gap, into a different application, by a person who had already mentally finished with it. The carry leaves no record, so nobody can check it. It’s manual, so it happens when somebody remembers.
The gap is structural. It sits between surfaces, and it’s where four years of their work had been falling. Rosa put it better when it landed on her: it was never a gap between her and Ben. It was a gap between her inbox and everywhere else.
Put that way, it stops sounding like a personality problem. The message arrived in one place. The decision about it got made in that place. The person it concerned was somewhere else. Between the two sat one partner’s memory, working at eleven in the morning between two calls.
What follows is that same Thursday with the crossing taken out. First, though: why the crossing gives way, rather than the two people standing on its edges.
Why the carrying is the part that fails
There are two ways to forget something, and they fail differently.
The first is losing hold of a fact you once had. A name, a figure, where you put the keys. The second is losing hold of an intention. You knew perfectly well what to do. The moment to do it arrived with nothing to bring it back, and it passed. Psychologists call that second one prospective memory, and its everyday form is small and familiar: the lunch left on the counter, the dry cleaner’s driven past on the way home.
Which of the two does more damage in skilled work is a question somebody has counted. Jessica Lang Nowinski, Jon Holbrook and Key Dismukes, working at NASA Ames, took a random sample of 1,299 reports filed by airline crews to the Aviation Safety Reporting System during 2001. They read every one. Seventy-five of them described a memory failure by a pilot. Of those seventy-five, exactly one was a case of forgetting a fact. The other seventy-four were forgotten intentions. They presented the work at the International Symposium on Aviation Psychology in 2003, and they were careful about what it meant. Vulnerability to these errors, they wrote, “is not an indication of lack of expertise, but rather a function of the way normal human memory processes operate in situations involving routine, well learned behaviors.”
When they sorted the seventy-four by what had gone wrong, one group led by some way. Twenty-seven of them came down to the absence of an adequate cue.
The same paper is exact about what a cue has to be. Two things, and both of them. It has to sit close enough to the intention that noticing it calls the intention to mind. And it has to be the kind of thing you’ll actually notice at the moment the intention needs doing. Hold both and it’s a reminder. Hold one and it’s a thing you happened to see, whatever anybody meant it to be.
One pattern in their sample was crews landing without a clearance. Thirteen reports, and in all but one of those the error was forgetting to switch radio frequency and call the tower. Normally a controller hands a crew straight to the tower, and that’s a cue arriving at the moment the action is due. In the reports the authors quote, the crew had been told to switch later instead, at a fixed point still ahead of them. The instruction was heard and understood. It simply arrived well before the moment it could be acted on, and by that moment there was nothing left to bring it back.
Rosa’s Thursday went exactly that way. The message was a cue and a good one, tied about as closely to the thing that needed doing as a cue can be. It reached her in a parking lot with six minutes and a meeting on either side of it. She read it at the one moment she was unable to act on it, and by the time she could act, the cue was behind her. The decision she had made was living in her head, which has no surface at all.
So the useful question sits with the crossing, rather than with how two careful people might be more careful.
Why telling each other more has a ceiling
The remedy most small firms reach for first is to tell each other more. That has a ceiling, and it’s worth seeing where it sits, because Rosa and Ben spent four years finding it.
Look at who the crossing asks to do the work. The person carrying a message across is the person who has just finished thinking about it. They read it, they made the call, and the call was that it belongs to somebody else. By the time the carry falls due they’re already moving. The step lands on the one person in the building who is done with the subject, which is most of why it’s fragile.
Then look at when. Rosa has two ways to do it and both of them cost something. She can do it now, which lands it on Ben inside a filing block, at a moment chosen by her morning rather than his. Or she can do it later, when he’s free to be interrupted, which means the sentence has to survive the rest of her day. In March one of them didn’t.
So the practice has been choosing between Ben’s concentration and Rosa’s memory, several times a day, for four years, and nobody ever put it in those words. Tell each other more and the first bill goes up. Tell each other more carefully and the second one does. Ben’s instinct that a livelier inbox is a worse inbox is him noticing which of the two prices he has been paying.
There’s a third option. The telling stops having to be an event. Her note lands in the same feed as the client mail and gets weighed against the rest of it, so it arrives the way everything else arrives, and she can let go of it the moment she has said it. What’s worth interrupting Ben for, and what waits until he next looks, becomes something he sets. The two of them stop negotiating it in the kitchen.
Which is to say the question is whether an intention can be given somewhere to live besides a head, and somewhere to arrive besides a shoulder.
A Thursday that goes differently
Rosa’s day keeps its shape. She still reads mail on a phone in a parking lot with six minutes before a meeting. She still makes the same call she made in March: this is a Ben one, he has the file.
What changes is where that sentence goes. She says it where the message is. Her mail and the practice’s own chat sit in one live feed, so telling Ben something happens in the same place she read it, while she’s still standing there.
Her note then gets the treatment a client’s mail gets. It’s read and weighed and ranked and summarized, and the commitment inside it comes out as a task with a date on it. So does the ask buried in the joinery owner’s forward, the one about signed accounts before the bank will release the loan. That comes out as a task with a due date and a link back to the message it came from, the original a click away. Neither partner wrote anything down. Rosa’s six minutes stayed six minutes.
Then there’s the part of the work where nobody does anything. The accounts go to the bank and the bank goes quiet. That’s normal, and it’s also how a week disappears. Either of them can ask that thread, in plain words, to come back if the other side stays silent. On the day they picked, it returns to the top of the feed with a link to the thread. Waiting becomes a dated thing instead of a thing to remember, which is the same move as the task and worth about as much.
Ben opens his mail at eleven, as he has for four years. He wants a quiet inbox, and what he gets is a shorter reconstruction. The Halvorsen matter arrives as one idea rather than five places. The mails, the task and next week’s meeting are drawn together under a single brief, with the sources listed beneath it. The calendar sits in the same window as the mail, so the meeting is right there too. His eleven o’clock is a read rather than an excavation.
There is exactly as much work as before. The relief is that the day stops asking two accountants to be the connective tissue between their own tools. That’s most of what Point is doing on an ordinary Thursday, with benefits as the flat version, and it’s the first thing that speaks to what went wrong in the week they took fourteen hundred off a fee.
Two partners who do not read the same way
One surface for two people raises the obvious problem, and it’s the one that sinks most attempts at this. Rosa and Ben work differently. Build a single place to suit one of them and the other quietly stops opening it.
Rosa reads in gaps. Six minutes in a parking lot, one thumb, standing up, usually with somewhere to be. She needs a short line per thread telling her what the thing is. She needs a way to deal with the two that are due now, and a way to put the rest down and still have them. On a phone that comes out as a single column of one-line summaries. A tap opens more. A swipe files something once it’s done. A thread she sets aside comes back at a time she picks.
Ben reads in a block. One window at eleven, no browser and no row of tabs, a reading pane where he left it, and counts in front of him so he can see the size of the morning before he starts it. He works down a list rather than around it. What he needs is depth on demand. The one-line version for the forty threads that are somebody else’s, the full original with its real headers for the three that are his. He makes that choice thread by thread, rather than settling it once on behalf of them both.
That’s one feed read at two depths, and it’s the part which makes a shared surface survivable in a firm of two. Each partner keeps their own habits. The spreadsheet had that problem underneath its more famous one: as well as needing to be fed, it asked both of them to work in a way neither of them worked.
The same idea runs through how the feed is ordered. Ranked by what needs you is the default, and it’s what Rosa reads in the parking lot. Plain time order, newest first, is one tap away, and it’s what Ben wants when he’s going through a day properly. Switching changes the view and leaves everything where it is, so each of them can have the order they want.
What folds into one surface, and what does not
The title of this chapter is doing slightly more work than it has earned, so take it apart.
Six applications don’t become one. Rosa and Ben still have their tax software and their ledgers, and they’ll have them next year. A mail surface doesn’t prepare a return, run a workflow or host a client portal. A tool that tells a small firm otherwise is selling something that firm will have to unpick later, usually in February.
What actually folds in is narrower and more specific than the pitch tends to be. Mail, from the accounts they already have. The practice’s own chat, so talking to each other happens where the work is. The calendar, in the same window rather than a tab away. Tasks, which until now lived in two heads. Files that arrive as attachments, gathered into one list with a link back to the message they came on, which holds what came through the mail. And one search that looks across mail, tasks, calendar and files at once, so finding something starts with what you remember about it rather than with which application kept it.
That last one is worth a further line. In a practice, the thing you’re hunting for is usually a figure inside something attached to the email. Ben can ask what a supplier’s invoice came to, or what the bank actually asked for, and get the figure back with a link to the document it was read out of. The answer arrives with its source attached. That’s the form an answer has to take in a firm that may later have to show where a number came from.
So the count is the wrong measure, and it always was. Rosa and Ben lost a week’s fee because a decision about a client’s work had to leave one application inside a human being and get into another one. Whatever shortens that journey is worth paying for. Whatever merely reduces the number of icons is decoration.
The category question for accounting firms sets out which kind of tool does which job, and why the products claiming to be the inbox and the practice-management suite and the tax software tend to lose to the ones that pick.
Why this is not the spreadsheet again
Ben’s first objection is the right one, and anybody who has run a small firm has earned the right to make it. They’ve been sold one place for everything before. It was a spreadsheet, and it worked for about three weeks.
The reason it failed doubles as a test, so it’s worth being precise. The spreadsheet was a second place, and second places have to be fed. By the Thursday in March it was eleven days stale, which is what a hand-kept record always is in the week you’d most want to read it.
The test, then, is one question. Does staying current require a person to stop what they’re doing?
That’s the whole difference here, and the only one that matters. The summary is written before anybody opens the thread. The task comes out of the message that asked for it. The follow-up is set in the same sentence that promises it. Each of those happens where the work already is, so there’s a single place and it keeps itself.
Your mailbox stays as tidy as it was, which is the other way these things quietly die. Filing a thread in the feed files it in the mailbox as well, and nothing is deleted. Point’s own sorting shows up on the messages in Gmail, so a partner who opens Gmail directly sees the same sorted mail. A surface that leaves you with two of everything is a surface you abandon in March.
And the load holds or grows. A client relationship acquires moments over the years and rarely sheds any, and more moments is more crossings. Discipline bought Rosa and Ben three weeks at the old volume.
Why a date and a trigger beat a resolution
A record that keeps itself is necessary, and there’s a second half to it. Something can be written down and still sit there, because a recorded intention and an acted-on intention are two different objects.
What separates them has been measured. Peter Gollwitzer and Paschal Sheeran spent years on the difference between a goal intention and an implementation intention. A goal intention is a person deciding to do something. An implementation intention is that same person deciding in advance when and where and on what signal. In their summary of that work for the US National Cancer Institute they report on the 94 studies in their 2006 review. Forming an if-then plan had a medium-to-large effect on whether the thing actually got done, d = .65. It had a comparable effect on the specific failure of never getting started, d = .61. The mechanism they give for it is the plain one. An if-then plan hands control of the action to a cue picked in advance, rather than leaving it to be remembered.
Which is the cockpit finding arrived at from the other side. One study counted what happens when the cue is missing. The other measured what happens when somebody puts one there deliberately.
It’s also, fairly exactly, a description of the useful things on Rosa and Ben’s Thursday. A task carrying a due date and a link back to the message it came from is an if-then plan whose if is made of a date. The follow-up on the bank is one written in ordinary words: if the bank stays quiet, bring this back to me on Friday. In both cases the remembering has moved off the partner and onto the thing itself, and the partner was the component that was failing.
Two caveats, because the research says rather less than a vendor would like. These are studies of people forming plans for themselves, rather than of software forming plans for them, and nobody has run that experiment on an accounting practice. And the effect is on getting started and on staying on track, which is a different thing from getting the work right. What carries across is the fix itself rather than the number attached to it: an intention tied to a cue you’re going to meet beats an intention held in good faith.
What a calm surface still asks of you
This leaves a two-person practice the size it is, and the parts it leaves alone are worth saying out loud. A partner who is told that everything is handled will believe it once.
The work stays the same size. Sixty clients write as much as they wrote before. Rosa still reads mail standing up between meetings, and Ben still wants a quiet hour at eleven. Both of those are working habits rather than problems to be solved.
The judgment stays with them. A reply comes back written the way they would have written it, and then it waits. Nothing is sent until one of them sends it, unless they choose to raise that setting themselves for a kind of work they’re comfortable handing over. The setting sits where they can see it, and they can turn it back.
Chasing works the same way, and the wording matters. A quiet thread comes back on the day you asked for it with the thread attached, and Point nudges you to follow up. Whether anything goes to the bank, and what it says, is still a decision one of them makes.
And once a message has actually gone, it has gone. That’s true of every mail system, including this one. What you can take back is what Point did on your behalf. Filing a thread, preparing a draft, creating a task: each one is on the record with the time it happened, and each is one click from being reversed.
The last thing they keep is where they already are. They connect the mail they already have and keep the address, the history and the contacts, so the info@ that Halvorsen’s owner has written to for four years goes on working, and everybody writes to it as they always did. That holds whatever is on the letterhead: design studios and training providers connect the same mailbox the same way.
What Ben said in the kitchen
He wasn’t sold.
“It sounds like more tools,” he said, and then half answered himself, which is how he concedes. The spreadsheet had failed because it was a chore to keep, and he could see this one keeps itself. He said so holding the same cup he had been holding for a minute.
Then he got to the one that mattered, and Rosa stopped putting things away.
“Say all of that works,” he said. “Say it’s all in the one place and I never have to go looking. On that Thursday you were in a parking lot and I was in a filing block. We’re both in the same feed now, both of us looking straight at the same message. What stops us doing precisely what we did? What stops me assuming it’s yours while you assume it’s mine?”
Rosa had no answer, and the feed in front of her had none either. Six applications folded into one surface leaves two heads where they were. The crossing was closed. The message, the task and the meeting were in one place, and each of those is a fact about the work. What either of them was actually doing about it was still, exactly as in March, inside their own head.
“One place is not one picture,” Ben said, and reached for his coat. “We could both be sitting in it, looking at the Halvorsen thing, and both still waiting for the other one to pick it up.”