The Five Places Human Hands Remain After Payment Setup, and the Range That Can Be Unmanned

After setting up payments, the state of not having got as much lighter as expected can arise. Application, screening, approval, start of use. Every procedure is complete and money arrives automatically. Even so, places where hands move on every transaction remain.

At this point the guess about cause heads toward the machinery. Perhaps the service chosen was poor; perhaps somewhere with more functions would settle it. The examination shifts to comparing replacements, and the undecided judgements around the payment never come up as candidates.

If the purpose of setting up is to reduce hands, the terminus is not the moment of approval. What remains as human hands after approval is the actual point of arrival. And what remains is settled. What has not been decided cannot be processed on the machine’s side. The places where human hands remain around a payment point to one of two things: what has not been decided, or what has not been configured.

What can be confirmed and what cannot are separated first. Three can be confirmed. Where the procedures presented come to an end. How many places human hands remain after approval, and which. And what the paring of friction is simultaneously paring away.

No line is drawn here for how far to unman. The appropriate range lands elsewhere depending on the count of transactions, the form of what is provided, and the size of the sums handled. It shifts within the same business as counts rise, besides. Hand over the position of the line as a single answer and only that answer survives the moving of conditions. What can be handed over here is the dividing line for redrawing it yourself.

The cost of misdiagnosis is worth setting down. Take the machinery as the cause and switch, and procedures are redone, screening is undergone again, and anyone receiving on a continuing basis is asked to register once more. And the same places remain where you land. Undecided judgements are not tied to machinery, so switching machinery does not move them.

What remains is cost alone. And since that cost remains as a record of “tried it, no good”, it can even look as though the material for the next examination has grown by one. What actually happened was one more comparison run against a mistaken cause. That is the reason to count the undecided side before shifting to comparison.

One point of order goes first. The judgement to switch machinery loses nothing by coming after the counting of what is undecided. Having counted, the reason for switching is concrete too. A switch made before counting pays the cost while holding no reason.

The Setup Presented Consists of Procedures Up to Going Live

The setup presented is arranged as the procedures up to going live.

Three means of payment are raised as available to a sole trader. Credit card payment, code payment, electronic money payment. Each has its own operators and procedures.

Two routes of setup are contrasted. Contracting individually with card companies, and having a payment agency handle it all. Individuals are advised toward the latter.

The flow of procedure is explained in four stages. Application, screening, approval, start of use. Screening takes days, and years of trading and financial condition are examined, runs the supplement.

The preparation listed comprises documents evidencing the business, identity documents, and details of the receiving account. For distance selling, publication of a page showing the terms of sale may be required.

On operation after setup, the treatment of expenses is touched on. Under which account the fees are handled, and how to treat the gap between the timing of deposit and the timing of recognising revenue.

The benefits presented are the widened opportunity from more payment methods and the greater certainty of collection. More supported methods mean fewer lost transactions, runs an occasional addition.

Nothing is missing from the procedures as written. The flow is as stated, and the enumeration of required documents is sound. The caution about expenses is a point genuinely required.

The sequence laid out shares one thing. Setup is treated as the work up to going live. Once usable it is complete, and operation afterwards has become a separate topic. It is organised as a problem of going live.

It is not that the organisation is an error. Going live does require procedures, and not knowing them stalls the work. Cut it off at going live and the state of not having reduced labour has nowhere to sit.

A state that goes unexplained gets no name. What has no name does not become an item for examination. So the observation that labour has not fallen drops out of the topic of procedures and gets picked up by the topic of comparing machinery. The topic that picks it up carries no item called “what has not been decided”.

The same cut runs through the sequence itself. The guidance ends at “start of use”. No stage handling what remains after starting exists in the list of procedures. A stage that does not exist cannot be prepared for. So the remaining places stay outside preparation and arise from the day after starting.

The Question the Enumerated Procedures Do Not Answer — What Remains in Human Hands

Automation reduces the work remaining, and the side that remains leans toward the difficult work.

After payment is set up, how far actually turns without hands. The turning part is clear. The operation of receiving money. The record that it was received. The execution of the deposit. The machinery does these.

The part that does not turn is clear too. Deciding what is provided at what price. Confirming that provision is complete. Answering enquiries. Judging whether to refund. Arranging records into a form usable for filing.

The difference between the two is whether judgement enters. Operations without judgement are executed by the machinery. Places where judgement enters are executed by the machinery if the judgement is already made, and remain in human hands if it is not. Which is to say the portion of remaining human hands that carries judgement grows with the quantity of what has not been decided. The portion without judgement vanishes through configuration and links.

This shape is not peculiar to payments. The cognitive engineering researcher Lisanne Bainbridge set out in 1983 the paradoxes automation carries (Bainbridge, 1983, Automatica, 19(6), 775–779). Designers automate what can be automated, so what is left to people is what could not be. The remainder is, by definition, the work that could not be entrusted to a machine.

The difference in object is worth flagging first. What this paper handles is the industrial control room, not the payments of a sole trader. What holds is the nature of the remainder. The work remaining after automation is not the work that was easiest before it.

In the payment setting this property shows with particular clarity. Amount, terms, deadline. Each can be set as a figure once decided, and leaves the setting field blank where it is not. For every blank field, thinking happens on every transaction.

Among undecided judgements are some that cannot be decided. Responses tailored to individual circumstance cannot be settled beforehand. They can be handled separately. Divide the range that can be decided beforehand from the range that cannot, and set only the former. Fail to divide, saying “individual handling is required so this cannot be automated”, and even the decidable part stays in human hands.

The dividing work itself takes time. What can be decided and what cannot does not emerge without thinking once. That time is one-off. And what the thinking yields is usable outside payments too. Deciding the terms of refund is at once a payment setting and a decision about the terms of provision.

Divide, and places classified as “cannot be decided” turn out, sometimes, to be decidable. Some of the judgement thought to require individual handling can be processed as a routine once split into three or four conditions. That realisation does not occur without the dividing work. A collapsed classification acts to stop examination.

Five Places Where Human Hands Remain Around a Payment

The places that remain divide into five.

The first is confirming that provision is complete. That money was received is recorded by the machinery; what you handed over is not. If handing over is automatic, the two are linked. If hands enter the handing over, that place remains.

The second is responding to enquiries. Whether payment completed, when it arrives, wanting a receipt. Among these, what can be answered routinely can be handled by an automatic reply. What cannot remains.

The third is refunds and disputes. With terms stated, judgement is only a matter of holding the case against the terms. Unstated, it becomes a judgement every time.

The fourth is arranging records. The record of deposits sits inside the machinery, but putting it in a form usable for filing requires reconciliation against your own ledger. Deposits arrive with fees deducted, so the revenue figure and the deposit figure do not match.

The fifth is changes to price or terms. The change itself is an operation of settings, but how to treat those already receiving on a continuing basis is a judgement. Hold existing parties at the old terms, or change everyone. Undecided, it becomes thinking on every change.

Of the five, the first and the fourth fall substantially through the choice of machinery and the configuration of links. Make provision hand over automatically and link records into accounting machinery, and the operations vanish. The second, the third and the fifth fall only by finishing the judgement beforehand.

Not all five need reducing to zero. While counts are low, some places are faster processed by hand. What takes priority is places that grow in proportion to counts. The first and the fourth arise on every transaction, so the burden grows with the count. The third is proportional to counts but low in rate of occurrence. As an order, start from what grows proportionally and arises every time.

The judgement holds that refunds, being low in rate, can wait. It is a place low in frequency and heavy in burden when it occurs. And occurring in an undecided state means judging under strained circumstances.

The cost of deciding is constant regardless of frequency. Think once and write it, and it is done. Ordering by frequency applies to places where work recurs, not to places settled once.

The order of the five carries a second reading. Sorted by the weight of judgement, the first and the fourth are problems of operation; the second, third and fifth are problems of policy. The order of starting is settled by counts, but how fast a place clears is settled by whether it is operation or policy. Operations vanish with settings; policies remain until the thinking is done.

Adding Methods and Adding Routes Are Different

Adding payment methods widens opportunity and disperses records and terms at the same time.

Card, code payment, electronic money. Three kinds, so supporting all three widens opportunity, runs the account. Opportunity does widen. Transactions failing because the payment method does not match do occur.

On that footing, the cost of adding is worth seeing.

First, the places of record increase. Passing through separate operators means deposits, fee calculations and refund procedures all become separate. The fourth of the five places — arranging records — multiplies by the number of methods.

Second, terms do not align. Deposit cycles and refund treatments differ by method. Settle refund terms as one and, where the executable range differs by method, the handling splits.

Third, the work of confirming increases, since the places to check whether payment completed become several.

So the judgement is not “more is better” but “which method is actually used”. And that judgement has material. What those already trading with you pay by. Where invoiced transactions dominate, demand for code payment is small. Where individuals dominate, methods other than card can tell.

At a stage without material, starting with one is safer. Adding can be done later, but reducing means asking those receiving continuously through that method to go through a procedure. The cost is not symmetric between adding and reducing.

Where a single operator handles several methods together, the three costs above shrink. Records and terms gather in one place. In that case the cost of adding is only the difference in fees. Rather than settling the number of methods first, confirm first the range that can be handled together.

The range that can be handled together is not settled by the payment operator alone. Application software used for bookings or courses — SaaS, as it is called — sometimes carries payment functions built in from the start. In that form the record of provision and the record of payment enter the same place, so the first place — confirming that provision is complete — clears at the same time. Look at whether the tools already in use carry the same function before beginning setup as the work of choosing a payment operator, and the object being chosen changes.

Narrow the methods and you lose those holding only the narrowed-out ones. That loss leaves no record. Leaving none, the cost of narrowing cannot be observed. The cost of adding — dispersed records and terms — is observed as daily work. Look only at what can be observed and the judgement skews toward adding.

Receipts and Records Cannot Be Rebuilt Afterwards

Arranging records looks postponable and carries a high cost of postponement.

The reason is that records are lost with time. The machinery holds them for a limited span. In a year, or in a few, old records can become unreferenceable.

And a lost record cannot be reproduced from your side. Who bought what and when exists only in the machinery’s record.

So extraction of records gets built in as periodic work. Once a month, or once a quarter. The format for extraction should be a standard one. Anything openable in a spreadsheet can be handled when the need arises later.

This work has the same shape as the extraction of a list handled in choosing a mailing platform. Hold on your side what exists only inside the machinery. The object differs; the principle is one.

There is a view that records for tax are settled by linking to accounting machinery. Linked, most of it is settled. Where the linking function exists, it is worth using.

What the link picks up, however, is amounts and dates. Who bought and what they bought may not enter the destination. That information is required later when looking from the customer’s side. Which is to say records for tax and records for seeing the business are not the same. Automate the former and the latter can remain.

Extracting records and holding them carries responsibility in their handling. Payment records contain personal information. The responsibility arises where they sit inside the machinery too. Held on your side, it takes a visible form. Deciding where they sit and who can see them is the condition of holding them.

The items extracted need not be everything. Payment details themselves are safer not held on your side. What is required is four: who, when, what, and for how much. Four suffice for looking from the side of the business. Narrowing the items extracted also lightens the responsibility. Keeping everything is not what makes it safe; narrowing to the required range is.

On the cycle of extraction, work backwards from the speed of loss. How many months the machinery holds is written in the terms of the contract. Extract at an interval shorter than that span and records do not vanish. Where it is not written, treat the absence itself as a reason for a shorter cycle. An upper limit that cannot be confirmed is safer treated as absent.

Paring Friction Is Also Paring the Room for Judgement

Simplifying the payment procedure lightens the burden and reduces the occasion for judgement at once.

Fewer steps means fewer people leaving partway. What is happening here deserves looking at from another angle.

The behavioural economics researchers Drazen Prelec and George Loewenstein argued in 1998 that the burden of paying and the pleasure of consuming bear on each other (Prelec & Loewenstein, 1998, Marketing Science, 17(1), 4–28). The burden of payment pares the pleasure of consumption directly.

The illustration offered is the taxi meter. Seeing the meter climb pares the comfort of the ride. Conversely, what has been paid for in advance can be enjoyed as though free, the burden of payment having been detached from the moment of consumption.

Here too the range of application is narrow. What this paper handles is the mental accounting of payment and consumption, not the practice of configuring a payment system. What holds is that the detachment occurs.

Moved into the payment setting, it runs as follows. Simplifying the payment procedure leaves the total cost unchanged and thins only the sensation of paying.

A thinner sensation carries two meanings. One is a lighter burden: payment no longer pares comfort, so as experience it improves. The other is a reduced occasion for judgement. If paying does not reach awareness, no occasion arises to consider whether it is needed.

The two arise from the same operation. Neither can be produced alone. So how far to pare friction becomes a question of judgement rather than of technique.

It is not that more friction is better. Unnecessary steps are simply a burden. The dividing line is drawn by whether the step being pared is used in judging. Reducing the number of input fields pares steps not used in judging. Making the fact of paying invisible pares the material for judgement.

There is the point that a customer not noticing the payment makes for a better experience, that advance payment is enjoyed more freely, and that the paper itself says so. It does. And the effect is real.

What deserves distinguishing is whether the not-noticing was chosen by them or arranged by you. Choosing advance payment is their judgement. Continuing to draw funds without notice is your design. The resulting “not noticing” is the same, but who decided differs. That difference does not surface in the quality of the experience; it surfaces in the reaction on noticing later.

The Dividing Line Is Drawn by What the Other Person Is Deciding

What may be pared is the steps that carry a decision into execution; what may not is the material used in deciding.

Input, confirmation, submission after deciding to buy. These are operations carrying a decision into execution, and steps here do not raise the quality of the judgement. The fewer the better.

What may not be pared is the material by which they decide. How much is paid. Until when payment continues. How to stop. That these are visible is not a step but material.

Where the two part concretely is the setting of continuing receipt. A continuing payment has its decision made once, and executes automatically thereafter. Judgement is not made on each execution. This is correct as design.

What becomes a problem is where the executing becomes invisible. No notice arrives, the amount is hard to check, the means of stopping is unclear. In this state the first decision keeps telling indefinitely.

So continuing receipt comes with a design that notifies on every execution. Notifying is friction, but friction as material for judgement. Notify and a certain proportion stop. That is a loss, but what is lost are people who would have stopped had they known.

There is a view that frequent notices are themselves a burden on the other person. They are. So frequency and form become objects of design. The direction for lightening the burden is not to stop notifying but to make the notice useful. Include not only the amount but what was provided in that span. Then the notice is at once a billing message and a record of provision.

Include the content of provision and it becomes apparent when nothing was provided in that span. In a thin month the notice itself tells against you. This is not a problem of the notice. A problem in the design of provision is being made visible by the notice. Treat the visibility as a disadvantage and thin the notice, and the problem is not resolved but hidden.

Not every continuing receipt requires a notice each time. Where provision appears as use each time — arriving weekly, being put in each time — provision itself performs the role of notice. Notices are required where provision recedes into the background. A form in which payment continues through periods of non-use. Here the occasion for recalling that payment continues is the notice alone.

The position of the line can be confirmed by one question. Whether paring that step removes something the other person knows. Reduce input fields and nothing they know is removed. Stop the notice and the fact of paying is removed. The same “reduce the steps”, and only the latter is paring their knowledge.

The Setup Procedure Itself Consists of Three Stages

The procedural side arranges into three stages: preparation, screening, and testing.

The first stage is preparing the application. What becomes necessary is anything that can evidence the business. A page from which what is provided is externally clear, the terms of provision, contact details, and identity documents.

Where it is treated as distance selling, certain items are required to be displayed. Trader name and location, contact details, price, timing and method of payment, timing of provision, and the terms of return or cancellation. These are checked in screening too.

This preparation is not work for the screening alone. Writing out the terms of provision is the very work of putting undecided things on the table that this article has been counting.

The second stage is application and screening. It takes days. Some cases finish in several days, some take several weeks. Working back from the day provision begins, the application goes in with margin. Additional documents may be requested. What is requested shows, as it stands, the place where material was short.

The third stage is configuration and testing. After approval, confirm that money can actually be received. Put a small transaction through yourself and look at payment, notification, record and deposit in one pass. This test finds the places not unmanned. Count where you operated and what human hands remain becomes visible.

The three stages do not necessarily proceed in order. Being asked for additional terms during screening and returning to the first stage can happen. Returning is not itself a failure. What was requested shows concretely where your preparation fell short. Adding after being told can be faster than trying to assemble everything by guesswork.

Running the third stage’s test on your own transaction incurs a fee. At small amounts this is negligible, but running the refund test as well can leave the fee unreturned. This cost is worth paying as the price of confirmation. Begin operating without once trying the refund procedure and the day it is actually demanded becomes the first attempt. Meeting the first attempt under strain, or having settled it on your own transaction. A few hundred yen separates the two.

And the record of the test becomes a procedure document as it stands. Write down which screen you operated and how, and the next refund requires no thinking. It tells in the form of one more thing finished being decided, as well.

Of the three stages, the third is the one most often skipped. Approval leaves the system usable, so transactions go through even with the test omitted. Omit it and the first live transaction doubles as the first test. Putting the attempt that can fail in front of somebody waiting, or getting it out of the way on your own side. Whether it goes through is the same; who is kept waiting when it fails is not.

In the Most Recent Transaction, How Many Times Did Your Hands Move

The human hands remaining can be read through three confirmations.

The first confirmation is tracing one recent transaction and counting the operations you performed. From receipt through to arranging the record, where and how many times your hands moved. Having counted, ask of each whether judgement entered. Operations without judgement fall through settings or links. Operations with judgement raise the question of whether that judgement can be decided beforehand.

The second confirmation is whether the terms of refund and the means of stopping a continuing payment are written where the other person can see them. Written, the material for judgement has been handed over. Unwritten, human hands arise in the form of enquiries, and the other side has no material for judging either. One omission makes a burden on both sides.

The third confirmation, where receipt continues, is looking at what reaches the other person on each execution. If nothing reaches them, the first decision is telling indefinitely. If something does, look at whether it holds only the amount or the record of provision as well.

No confirmation here looks at the functions of the payment machinery. What settles the range of unmanning is the range of what has been decided, so what deserves checking is the side of judgement.

The usable range differs across the three. The first cannot be run without a transaction having occurred; just after setup, the test transaction substitutes. The third cannot be used where receipt does not continue, and two suffice.

The question arises of what to do if counting operations yields only operations that cannot be reduced. That this is knowable is itself a result. Operations that cannot be reduced either carry judgement or exceed what the machinery supports. In the former case, deciding reduces them. In the latter, it becomes a reason to change the machinery. Having a concrete reason and feeling vaguely inconvenienced differ when the time comes to choose again.

The frequency of notices, the cycle of extracting records, the proportion unmanned. These three are not set down as figures, and not because conditions are various. All three are figures that move according to the quantity of undecided judgements. Undecided falls and the proportion unmanned rises; it grows and the proportion falls. Fix a dependent figure first and the fixed figure becomes the criterion, closing the eye that watches the quantity undecided.

On the day the approval notice arrived, what ended and what began. What ended was the procedure; what had begun was the work of putting undecided things on the table one at a time. That human hands grow with the quantity undecided. That the places remaining divide into five. That paring friction can pare the material for judgement as well. Hold those three while looking around your own payment setup, and the manual work remaining converts, one at a time, into the name of something undecided.

If only one thing gets confirmed, take one recent transaction and write out in order the screens you touched between receipt and record. To each item written out corresponds either a judgement not decided or a configuration not made.

How to assemble the machinery as a whole is handled in systemising a solo business, and the criteria for judging before choosing in accepting online payments as a sole trader.