Chapter 11 Pay for Intention, Measure Attention

The old company bought time and hoped for attention. The collapsed company buys intention and measures attention. This chapter is about the difference, and about the few lines of an agreement that make it work.

Two of the twelve tenets carry it. The fifth says pay for intention, not skill. The tenth says hire for intention and measure attention. Stated as tenets they sound like sentiments, and every company already believes it pays for commitment and notices when it goes. It does neither. It pays for skill and seniority, in bands, and it notices commitment going at the exit interview. What follows is how a company actually does the two things the tenets say: how the pay is built, what is measured, what the aligner commits to each day, what happens when she is away, and what happens when her attention goes.

The intention premium

Start with why the number is so high.

In the old company the price of a person was set by her skill and her years. A recruiter with five years’ experience in a city earned roughly what every other recruiter with five years in that city earned, because the company was buying the same thing every other company was buying: a set of skills, and eight hours a day in which to apply them. The skills were the scarce thing, and the market priced them.

The chain has made skill cheap. The search, the assessment, the first interview, the drafting, the chasing, the scheduling are all carried by agents whose cost is a fraction of one salary. What the chain cannot supply is the thing that was always rarer than skill and was never priced, because the old company had no way to buy it separately: intention that holds. The commitment to the result that is still there on a bad week, in the month after a bonus, after the third client rejection, on a day when nobody is looking. One aligner with that intention, and a chain beneath her, produces what a department produced. She is worth to the company something like what the department was worth.

So the doctrine pays her very highly for her role at target, far above what the market pays for it. The level is deliberately high, and it does two jobs. First, it changes who the company can choose from. At the market rate a company takes whoever applies and hopes; at a price far above it, it can wait for the few whose intention does not waver, and refuse everyone else. Second, it is still cheap. A company that needs a tenth of the people can pay each of them very highly and still spend far less than it spent before, before a single result is counted.

There is good evidence that how a company pays changes both what people do and who comes to work there. When the American windscreen company Safelite moved its installers from hourly wages to pay by the windscreen in the 1990s, output per worker rose by about 44 per cent. About half of the gain came from the same people working differently. The other half came from a change in who joined and who stayed.⁴⁰ Pay is a filter before it is an incentive. The premium is set high for the filter.

How the pay is built

The formula is the same for every human in the company, from the aligner to the Keeper, and it has two parts of equal size.

A base, at the premium. Half of the pay at target is fixed. It does not move with the month. It is the company’s payment for the intention it selected, and it is set high enough on its own to be well above anything the market pays for the role.

A share, of the same size at target. The other half moves with the vector’s result against its target: it rises when the vector delivers more, and falls when it delivers less. At target, base and share are equal. Above target the share keeps rising, and a pair whose vector carries much more than its target may earn more than the master who holds the Outcome. The structure expects this. Below target the share falls towards nothing, and the base is untouched.

The master is paid the same way, on her Outcome’s result against the Keeper’s target. The Keeper is paid the same way, on a weighted score of the four core outcomes against the board’s targets. Nobody is paid on someone else’s number. Nobody is paid on activity: not on calls made, candidates sent, hours logged or meetings held. The tuner is paid on the same vector as the aligner, with a base and a share of her own, because the result is theirs together and either may be running it on any given day.

Half the pay riding on the result is a strong pull, and a strong pull needs strong guards, because the one thing the company must never buy with its share is a bad signature. An aligner who is behind target on the twenty-eighth of the month has a reason to let a weak candidate through a gate. Three things stand against that reason. Her signature carries her name, and the record of every release she made is permanent. The vector’s own inspector agent samples what she released and reports it to her master as a separate colour inside her tile. And the result the share is paid on is drawn where the customer feels it: a placement that fails, or a client who does not pay, is not a result. A bad release does not earn the share. It costs it, a month later.

Where GoodSpace stands

GoodSpace today pays the people who run its vectors a fixed salary and an incentive for each mandate closed. That is the right direction, and it is not yet the design.

It is the right direction because the incentive is paid on a result, a closed mandate, and not on activity. It falls short in two ways. The base is not yet at the premium, for the reason I gave in the last chapter: the vectors do not yet produce enough to carry it. And the incentive is paid closure by closure, not against the vector’s target. Arithmetically, a linear share of the vector’s result pays the same per closure as today’s incentive does; the difference that matters is what counts. Today a closure is counted when the candidate joins; the design counts the retained, paid result of chapter 9, and treats the vector as one result rather than a set of deals. The move from today to the design is therefore two moves, and both follow the output. As each vector’s monthly result rises towards the twenty-five of chapter 9, the base rises towards the premium, and the per-closure incentive becomes a share of the vector’s result against its target.

Measure attention, not presence

Intention is selected once, at a high price. Attention is that intention showing up, day by day, and it varies. Everyone has bad days; the people with intention bounce back from them. When intention goes, attention goes with it, and it goes in a pattern that I have learned to see: the numbers dip, the leave increases, the person becomes slow to answer, and you can see it on the face before anything is said. Chapter 3 described that pattern, and the research that shows it can be seen months before a resignation.

The old company could not see it, because it measured the wrong thing. It measured presence: the register, the badge at the door, the hours at the desk, the green dot beside a name. It bought eight hours of presence a day and got, by the best measurements we have, about three of attention. The vector buys the three, and lets the agents cover the rest.

It measures them through the work. Beside the stall alarm, which reads the flow of the chain, each vector carries a second alarm that reads the aligner’s engagement with it. It reads only work signals, all of which the vector already knows. How quickly the gates were answered. How many decisions are waiting for her, and for how long. Whether the daily review was held, and how thoroughly. Whether the quality of her releases is holding, as the inspector agent samples it. Nothing about the person herself is observed: not her screen, not her location, not her hours, not her keystrokes. The result flowing, and the decisions being made, are the proof that attention is present. A company that has to watch its people to know whether they are working has not built a vector; it has built a department with a camera.

The daily contract

Two things are fixed in the aligner’s day. Everything else is hers.

Hours, chosen and kept. The aligner chooses in advance how many hours a day she will give the vector, and when. Not set by the master, not the same for every aligner, not even the same every week; but chosen, and then kept, because the vector and the master need to know when its human will be there. At GoodSpace, as chapter 6 showed, the delivery pair’s day is a morning: the gates are cleared and the signatures given by one o’clock, and the chain works on through the afternoon and the night.

A full review, every day. At an hour she chooses, the aligner reads every alarm, clears every waiting decision, samples what the agents produced, and hands her tuner what needs to improve. The review is the one act the chain cannot perform for itself, and it is the evidence of attention. Results are judged by the month. Attention is judged by the day, and the review is how.

The master sets the number of hours her Outcome expects; the aligner sets when they fall. Neither sets how the aligner spends the rest of her day, and the doctrine has nothing to say about it.

Gates, and absence

Chapter 6 set the two rules that keep the daily contract from being a cage, and they are terms of the agreement, so they are restated here in one line each. A reversible gate opens on its own if she has not answered it in the time the playbook sets for it, usually a day; an irreversible one waits for a human however long that takes, and the stall alarm says so. Before a day away she chooses one of three settings, pause, automatic, or her tuner, and names in advance which of them the vector takes if she simply does not appear. A missed day costs the vector one day of her judgement, not one day of its flow; what she loses is the chance to inspect what went through, and she gets it back the next morning.

Form

Sometimes the attention alarm does not clear. The bad days become a bad week, and the bad week becomes a pattern. What happens then is the part of the doctrine that most people find hardest to accept and that I find most humane, because it is the part where the company is honest.

The word I use for it comes from cricket. A batsman in form scores runs; a batsman out of form does not, and everyone can see it, because the scoreboard is public. Nobody calls a meeting about it. The selectors rest him, and if form does not return he is dropped, and nobody is surprised, because the team exists to win, and he knew that when he walked out to bat.

The vector handles falling attention in the same three steps, in order.

The mirror. The attention alarm goes first to the aligner herself, and to nobody else, for a number of days the master sets in the playbook. She sees her own gates slowing, her own decisions waiting, her own review thinning. Most of the time this is enough. A bad week ends, privately, without anyone having watched it.

Rest. If the mirror does not clear it, the master rests her. The tuner runs the vector, which the tuner can do because the doctrine designed the standby in, and the aligner takes paid time away. Resting someone is the step every old company skipped, because in the old company the work stopped when the person stopped, so nobody could be rested; they could only be managed. In the collapsed company resting her costs the result almost nothing, and it gives the company’s most expensive asset, her intention, its best chance to come back.

Replacement. If form does not return after rest, she is replaced. There is no warning letter and no improvement plan. An improvement plan is supervision dressed as fairness: it asks a manager to watch a person, which is the job the doctrine abolished, and it measures effort at exactly the moment when the thing being paid for, intention, has gone. Replacement is quick, because the premium means the company is never short of people waiting to be chosen, and because the selection of the next chapter finds the next aligner in four working days. It is often a relief to the person replaced as well. Nobody whose intention has gone wants to keep pretending it has not.

The same three steps run one level up. The signals that show an aligner’s attention also show a master’s: whether her red tiles were answered, how many decisions wait on her, whether her own daily review was held. She sees them first, as a mirror. The Keeper rests her and her standby holds the Outcome. And if form does not return, the Keeper replaces her, with the same lack of drama.

The aligner’s agreement on one page

The old company hired a person with a job description, kept her with an appraisal form and policed her with an attendance policy. The collapsed company needs one page, signed by the master and the aligner when the vector is handed over. If a line cannot be filled in, the vector is not ready to be handed over.

1. The vector. Which result, for which market, with which matched vector if the result is shared.
2. The target. The result per month, drawn where the customer feels it, and what does not count as a result.
3. The base. The fixed half, at the premium. It does not move with the month.
4. The share. Equal to the base at target; how it rises above target and falls below it; when it is paid.
5. The hours. How many a day, and when, chosen by the aligner and kept.
6. The review. The hour of the daily review, and what it must cover.
7. The gates. Which are hers to sign, which default open and after how long, and which never do.
8. The absence setting. What the vector does if she does not appear: automatic, tuner, or a named person in the control room.
9. The standby. The tuner’s name, and confirmation that she can run the vector alone from tomorrow.
10. The attention signals. What the attention alarm reads, and the plain statement that nothing else about her is observed.
11. Form. How many days the mirror is hers alone; how rest is decided and paid; and that replacement follows if form does not return.

Eleven lines. Everything the old company tried to manage through a manager is on them, and nothing on them needs one.