What Is Arranged in the Stages of a Marketing Funnel Is Involvement, Not Products

What stands in the stages of a funnel is not a product, not a price, not a measure. What stands there is the depth of involvement the other party holds towards you.

The form generally presented is a diagram of the flow from awareness to purchase, drawn as a funnel. Wide at the top, narrow at the bottom, with numbers falling as you descend. It is easy to read and widely used. And the premises the shape places have been passed down with almost no examination.

There are three premises. That people move in one direction from top to bottom. That the boundaries between stages are settled by proximity to purchase. And that the lower you go, the higher the value.

None of the three match how relationships actually get made. People move back and forth; what settles the boundaries between stages is the depth of involvement rather than proximity to purchase; and time continues after the purchase, in most cases for longer than before it.

When the premises do not match, the measures assigned to each stage stop matching too. Arrange by proximity to purchase and only one reason to move up a stage emerges: they have not bought yet. That reason contains no information from the other party’s side. In a design with no such information, the answer at every stage becomes an intervention that pushes them to the next one.

Cut by involvement and the question changes. Not how do we move them to the next stage but what is this person trying to judge right now. One question points at your own next action; the other points at what belongs in their hands.

The individual measures — how to write a subject line, how to price a trial, how to split a mailing — live in their own articles. What is taken on here is the single point of what the stages get cut by, and how the eighteen articles arrange themselves from that cut. Choose measures while the cut is unsettled and you cannot adjudicate whether the choice was good.

Conversely, some things get no answer here. How many stages are appropriate varies with the field and the interval between purchases, so it cannot be settled from this side. Nor do these eighteen hold a method for measuring, from outside, which stage a person’s involvement currently sits at. What they hold is the procedure for taking the boundaries from observation.

Two principles run through these eighteen articles: cut the stages by involvement, and relationship before selling. Neither comes out of the funnel diagram. What the diagram carries is a shape, and the three premises that came attached to it. Neither principle says to stop using the diagram. A wide top and a narrow bottom does match how the count actually falls. What gets swapped is what the stages inside that shape are divided by.

Every Example of a Marketing Funnel Carries the Same Premise

The standard account organises the funnel as an instrument for making purchasing behaviour visible. The basis on which the stages get cut is never once questioned inside that account.

Source to source, the description barely varies. A marketing funnel is a diagram of the flow from a customer learning about a product to buying it. Funnel means a funnel, and the inverted-triangle shape is described as its characteristic. The basic stages number four: awareness, interest, comparison, purchase. Numbers fall as you descend, which produces the inverted triangle.

Extended examples are shown as well. One puts sharing and spreading after the purchase to make an hourglass; one makes purchase and reconsideration circulate; one draws improvements in customer experience lowering the cost of new acquisition as a loop. Two uses are presented: count the numbers at each stage to find where people are dropping, and assign a measure suited to each stage. The expected effect is visibility of purchasing behaviour — know how many sit at each stage and where the blockage is, and where to act is settled.

The set holds up as a working framework. Counting stage by stage is genuinely useful, and it is a fact that different stages call for different information. What is off is not the assignment of measures but the basis on which the stages get cut.

The extended examples do not change the basis either. Adding sharing and spreading after the purchase moved the purchase to a midpoint, but the stages are still cut by how much each contributes to your transaction. So in the later stages too, the question stays how do we move them onward. Changing the shape of the diagram solves only the problem of treating purchase as the terminus.

There is no defect in the arrangement itself. Seen from the selling side, the order is natural. What moves is that the arrangement supplies only one reason to climb a stage. The reason becomes they have not bought yet, and no information from the other party enters it.

In a design whose stages carry none of the other party’s information, measures also get chosen without regard to their state. At every stage the answer becomes an intervention that pushes to the next one. And when that intervention fails, the only explanation left standing is that the intervention was too weak. That is the route by which no move other than raising the intensity ever appears.

The two uses presented also carry an order inside them: measure first, then assign measures. Being able to measure is presupposed. But becoming able to measure requires that the stages are defined and that records exist by which people can be sorted into them. At a stage with no records, this order cannot be used as given.

Where the Funnel Diagram Came From

The funnel shape came from somewhere other than the staged account. Receive the two as a single diagram and how far it applies becomes unreadable.

In 1898 Elias St. Elmo Lewis, an advertising practitioner, organised the roles of advertising copy into stages: catch attention, create interest, convince. He did not himself call it by an acronym. The acronym settled in 1921, when C. P. Russell used it in a trade journal.

And the funnel shape itself was brought in by William Townsend in Bond Salesmanship in 1924. A book about selling bonds.

The origin yields two facts worth holding.

First, the staged account and the funnel shape were made by different people at different times. The staged account was about how to write advertising copy. The funnel shape was about how to manage a salesperson’s prospects.

Second, the context in which the funnel shape was born was individual selling. In bond sales you contact many prospects and contract with few, so a shape that is wide at the top and narrow at the bottom fits the reality. That the number of prospects is fixed at the outset; that the narrowing runs one way; that a contract closes a round. It was a diagram for a scene where those three held together.

So this diagram came from the specific scene of face-to-face individual selling, and many of the scenes where it is now used differ in their conditions.

If the diagram works, the origin does not matter — the objection holds, and indeed it does work. That said, knowing the origin tells you how far it applies. Where movement is one-way, where things end in a single round, where the number of contacts is fixed at the outset, the shape fits well. Where there is back-and-forth, where the relationship continues after the purchase, where contact originates from the other side, the fit weakens.

A gap between the circulated form and the originally built form is common among well-known frameworks. Among the articles gathered here, the cost of acquiring a customer sets out a case of the same type.

A difference of origin does not make a framework wrong. Becoming useful in another setting does happen. What can be said is that the range in which it applies can be inferred from the origin. Without the origin, you cannot tell why it fails to apply where it fails to apply.

And a framework whose origin has been forgotten stops being doubted. Something whose maker and purpose are unknown gets treated as having been there from the start. What was there from the start gets used as a premise and never becomes an object of examination.

The Three Premises the Funnel Shape Carries In

The shape of the diagram carries three premises on its own. The premises are never stated in words, so whoever receives them does not notice having received anything.

The first premise is that movement runs one way. A funnel flows only from top to bottom. Returning, and entering partway, cannot be expressed inside the diagram. In practice people move back and forth. They advance to the comparison stage and go back to gathering information. Some enter at an intermediate stage.

The second premise is that the boundaries between stages are settled by proximity to purchase. Awareness, interest, comparison, purchase. The basis is how close each sits to your transaction. Seen from the other side, that basis carries no meaning. The other party is moving in relation to their own problem, not in relation to your transaction.

The third premise is that the lower you go, the higher the value. The shape of the diagram presents the bottom as the destination. In practice, purchase is one event partway through a relationship. There is time before it and time after, and the time after is frequently longer.

The three premises bear directly on how measures get chosen. Think of it as one-way and measures become things that push to the next stage. Think of it as ordered by proximity to purchase and the lower stages get priority. Think of the lower as more valuable and the upper stages become an object of investment.

The third premise has a sequel. Once the upper stages become an object of investment, their results get measured only by their contribution to the lower ones. Value that stands on its own at that stage does not get counted. What is not counted stops being made. That is the route by which upper-stage content comes to be built only as an entrance to what lies below.

The upper stages are an entrance, so of course they are built to push people down. It looks obvious. That said, content built only as an entrance leaves nothing with the person who finished reading there. Nothing left means the name is not remembered either. It does not come to mind when the need next arises, so its work as an entrance weakens too.

None of this says the diagram should be discarded. Thinking in stages is itself useful. What can be said is that you had better not take what the stages get cut by from the shape of the diagram. The diagram shows only a shape, and the cut can be settled outside it.

And the three premises hold one another up. One-way, so the bottom looks like the destination; the bottom is the destination, so the arrangement runs by proximity to purchase. Remove only one and the remaining two pull the shape back.

Cutting the Stages by Involvement Settles the Information to Hand Over, Not Your Own Move

The first principle these eighteen articles place concerns what the stages hold. What stands in the stages is not a product, not a price, not a measure, but the depth of involvement the other party holds towards you.

Involvement is the degree to which this matter has become their own. At shallow stages the content is not examined closely; judgement runs on legibility and on how the source appears. At deep stages the content itself gets examined.

Research on persuasion has distinguished these two modes of processing. Richard Petty and John Cacioppo separated a route that scrutinises the content itself from a route that judges on peripheral cues, and showed that higher personal relevance draws on the former, and that attitudes formed by the former persist longer and resist counterargument (Petty & Cacioppo, 1986, Advances in Experimental Social Psychology, 19, 123–205).

What the framework gathers is experiments on attitude change, not measurements of business funnel design. The reach stops there. What is shown is only a direction: the same content gets processed differently according to the recipient’s level of involvement.

Cut the stages by involvement and the design question changes. Cut by proximity to purchase and the question is how do we move them to the next stage. Cut by involvement and the question is what is this person trying to judge right now. The first settles what you do; the second settles what you hand over.

And involvement does not deepen through intervention. It deepens when a connection forms on their side: this is my problem. What makes the connection is the material you handed over together with their own circumstances. So there is no operation that raises a stage. What you can do is place material from which a connection can form.

Does that leave measures moving nothing? Some things do move: the quantity of material that arrives, the range of people it reaches, and the form of the material. What does not move is the connection itself. Without this distinction, you try to move what cannot be moved, and keep raising the intensity. Where that route leads is handled in the product launch.

Nor does any of this say deeper involvement is better. Some situations are done with shallow involvement, and demanding depth there is demanding the other person’s time. The adjudication is which level of involvement what you are handing over requires. Hand over a criterion for judging and deep involvement is needed. Hand over an answer that ends in one round and shallow will do.

Get the correspondence wrong and it fails to land regardless of the quality of the content. Place content that needs deep involvement at a shallow stage and it passes unread. Place shallow-stage content at a deep stage and it gets treated as already known.

Relationship First, Selling Second — What Happens When the Order Is Reversed

The other principle is an order. Relationship comes first and selling comes second.

This is a statement about sequence rather than a moral claim. Selling to someone with whom no relationship has been established requires supplying a reason from outside in order to establish one. Deadlines, bonuses, price conditions. None of them came out of the relationship.

A reason supplied from outside demands the same intensity next time. Keep meeting the demand and the reason moves entirely to the transaction side. A relationship that has fully moved ends the moment the condition stops.

The reverse order runs like this. With the relationship established first, all the selling stage requires is an account of what you are offering. An account needs no intensity, so repeating it does not wear anything down.

How this order looks varies with the age of the business. Early on there appears to be no time to build a relationship, and in fact the time until results appear is longer when the relationship goes first. And that difference in time settles the choice. Looking only at the short-term numbers, the form that puts selling first tends to look better.

This structure runs through all eighteen. Most of the articles on indicators handle the same problem in a different place. When an indicator moves, whether the reason sits on the relationship side or the transaction side cannot be read from the indicator itself.

Putting selling second does not guarantee anything on its own. Even with a relationship established, no purchase occurs if what you offer is not needed. What can be said is that reversing the order creates, by exactly that much, an ongoing need to add reasons from outside. The adding cannot be stopped.

The difference in order also changes the kind of work on your side. Put selling first and the centre of the work becomes the design of interventions: what to send when, how to present it, which conditions to attach. Put the relationship first and the centre becomes the design of what you hand over: what you get recognised as handling, and whether what you handed over stays with them.

The two demand different uses of time. The first is work with deadlines; the second is work without them. In an environment where only work with deadlines gets done, the order drifts automatically to the first. It happens without any recollection of having chosen.

There are periods with no room to put the relationship first. That said, the order is not all-or-nothing. Rounds that put selling first and rounds that put the relationship first can be separated. Separate them and the ratio can be changed later. Run on one alone and the very concept of a ratio goes missing.

The First Layer — Two Articles That Settle Where the Stages Are Cut and Where the Entrances Merge

The shape of a funnel is settled by two things: where the stages get cut, and where the entrances converge. Two articles take one each.

The cut sits in the principles of funnel design. Build the stages in order of price and the reason to climb becomes they have not bought the more expensive one yet, which carries none of the other party’s information. The boundaries between stages are not settled by you; they are taken from observation. Watch for the point at which the kind of question they ask changes and the boundary appears on its own. The point where the kind of question changes is the point where the stage of involvement changed.

The routes are handled in customer journey design. How many entrances to hold, and where to converge them, are separate decisions. The reason to converge on one is not to lower the burden of choosing. It is whether the history accumulates in one place. Build a receiver per entrance and the same person splits across the records. From split records you cannot read what a person went through to get here.

What runs through both is that the operation that settles the shape constrains everything after it. Change the cut of the stages or the position of the convergence later, and the records up to then stop being usable.

These two have to be settled earliest, and are therefore settled when the least information is available. The order in which information arrives after the decision cannot be avoided. So what matters at this layer is not deciding correctly but recording what the decision rested on. With the grounds recorded, you can correct it once information arrives.

The two decisions sit on different axes. The cut of the stages is the vertical design; the position of the convergence is the horizontal one. The vertical expresses how far a person has come. The horizontal expresses which entrance they came in through.

Confuse the two and differences of entrance get treated as differences of stage. Put came from search and came from a referral at separate stages and people at the same involvement get managed separately. Records that have split cannot be joined afterwards.

Initial involvement genuinely does differ by entrance. People arriving through a referral frequently hold deep involvement from the start. What can be said is that this is a difference of starting value and not a difference of stage.

Settling this layer early carries a price too. The form you settled limits what becomes visible for a while. Cut into three stages and only three kinds of people appear in the records. A fourth stage may exist and will not show up. So the occasion to doubt the settled form does not arrive from outside.

The Second Layer — Three Articles on the Route That Reaches People, on What Is Handed Over First, and on When

What you hand over at the start of a relationship stays afterwards as the standard. Three articles stand behind that first handover.

The premises of list building enters from the side of the route of reach. What you hold is not a name but a route that reaches that person. And a route stands only on their continuing consent. The verb collect imports the metaphor of stock; stock does not decrease and consent does. That gap produces the misreading involved in treating a list as an asset.

What gets handed over first sits in the design of a trial offer. The substance of being seen as cheap is not the amount but the set of things it gets compared against. The axis that divides is completeness, not quantity. Hand over a shrunken version of the real thing and what gets experienced is a thin version of the real thing; hand over a part executed completely and what gets experienced is the quality of the real thing. At the same price, what is experienced differs.

The product launch is the method of concentrating the moment of purchase. At its centre is not the release of information but the external supply of a reason to buy now. The costs are two: it demands the same intensity next time, and a judgement settled by a deadline does not remain with them as a criterion for judging.

What runs through the three is that what you hand over first becomes the standard for everything after. The range handed over free becomes the comparison when it next becomes paid. An experience settled by a deadline requires a deadline next time too.

The first design does not have to be perfect. It can be changed. What cannot be changed is the standard already left with the people you handed to. So changing it requires an account of why it changed.

The three have a settled order of work. Holding a route comes first, what to hand over comes next, and when to hand it over comes last. Reverse this and you design what to hand over and then look for a means of delivery. Look for the means afterwards and you choose from what already exists, which limits the options to borrowed routes. Keeping the order is not a claim that things move faster. The period in which the route gets built is also a period with nothing yet to hand over.

And the third method can be executed even where the first two do not hold. Being executable, it is the method most easily started out of order. Executed out of order, no relationship remains after the concentrated purchase. Holding no route, you cannot make the next contact; and since what you handed over did not remain as a standard, it does not get used in the next judgement either.

The Third Layer — Four Articles on the Indicators That Diagnose Where People Drop

The indicators used in diagnosis tell you where things are dropping and not why. Four articles cover how to read them.

In where to begin doubting when the conversion rate is poor, the order of diagnosis is what is at stake. The conversion rate is not a single indicator but a product of several proportions. And since records accumulate automatically only downstream, diagnosis leans structurally downstream.

What the phrase not selling points at gets separated in the diagnostic procedure for what is not selling into five: not seen, not read, not taken as addressed to them, not now, not decidable by themselves. The response differs across all five.

In the cost of acquiring a customer, the definition of one unit varies with the stage at which you measure. And ease of measurement and weight of meaning run in opposite order.

What how to read the drop-off rate names is the structure by which only the side that stayed is observable. The reasons of those who left cannot be obtained after they have left.

One point is shared by all four: an indicator shows a place and not a reason. Where things are dropping is legible; why they dropped is not inside the indicator. And all four handle, in different places, the same consequence: put the measured stage in the position of a target and that stage becomes the object of optimisation. The consequence is unavoidable, so choosing which stage to make a target is where the design actually happens.

None of these articles concludes with measure more. Increase the number of things measured and each addition becomes an object of optimisation. What is needed is choosing which to measure.

One criterion for choosing: is it settled what you would do if that number moved? A number with no answer changes no behaviour when you look at it. Line up numbers that change no behaviour and the material for judging appears to grow while only the number of things to look at actually does.

The wish to record everything anyway, in case it becomes useful, does not go away. Recording and displaying are separate. Records may be many; the things looked at regularly are better few. Separate them and you can still go back later. Leave them unseparated and everything recorded becomes an object of the monthly review. As the review grows, the time spent on each shrinks, and none of them get looked at deeply.

The Fourth Layer — Four Articles on the Indicators That Measure Whether the Relationship Continues After Purchase

Indicators that measure what happens after a purchase move the same way whether the relationship thickened or the exit got harder. Four articles bear on that distinction.

What moves the repeat rate is the design of the relationship, not measures rereads a second purchase as an observation that the relationship is still running. Put the reason on the transaction side and the condition has to be renewed.

The conditions under which a contract continues sit in what settles the retention rate is a shared framing, not the substance. The retention rate counts has not stopped yet, not is continuing. A route that holds the number up through difficulty of leaving is always open.

Folding a future estimate into the present is handled in customer lifetime value. The folded number then goes on to settle how the person in front of you gets treated. Treatment produces the result, so the prediction goes out and makes itself true.

The point at issue in raising the amount a customer spends is what the stages get cut by. Cut by price and the upper stage becomes more of the same thing.

What runs through the four is that the same number rises in two opposite states. Whether the relationship thickened or leaving got harder, the number moves in the same direction. From the side of the number they cannot be told apart.

The method of telling them apart is common to the four as well: actually open the exit and see. If the number does not move once opened, what supports it is the relationship. If it does move, what supported it was the difficulty of leaving.

Resistance to executing this is high, because it means running an operation that might depress the number purely in order to confirm something. And unless it is run, the material for adjudicating whether the number rests on relationship or on friction never comes into hand at all.

The scope can be narrowed. Telling only some of them the renewal date in advance, and stating plainly how to stop, is enough to reach an adjudication. Limit the range and the downside is limited too.

These four cannot be used until the business has reached a certain age, because continuation and repurchase do not occur without elapsed time. What is visible at an insufficient age is individual events rather than proportions. Was there anyone who came back a second time without being prompted? Was there anyone who wrote back specifically about the substance? While the denominator is small, that single case is more usable for the next judgement than any ratio.

The Fifth Layer — Three Articles on Designing the Reason a Relationship Continues

A design that makes the reason for continuing always holds a period in which it points the opposite way from the short-term numbers. Three articles divide up how to hold that period.

Where repeat measures return to after running their course counts what remains once a list of measures has run its course. Many items on such a list put the reason on the transaction side. Where the course returns to sits outside the measures.

Segmented delivery moves the question over to the axis of division. Dividing by attribute contains an inference from what was observed of a group to an individual. And what settles the axis is availability rather than importance. Attributes can be taken at registration; states cannot. That gap settles the axis of classification.

In the dividing line of educational marketing, the substance of what gets provided is where the line falls. Someone handed an answer comes back to ask again; someone handed a reason judges for themselves next time.

What runs through the three is that the short-term numbers point the opposite way from the long-term state. A reason manufactured by a measure works faster; dividing by attribute takes less effort; handing over an answer produces higher satisfaction. And all three carry the same practical consequence: knowing that they point opposite ways is what lets you keep a design when the short-term numbers are bad. Without that, the only material for judging is the short-term numbers.

None of this says the short-term numbers can be ignored. Without the business continuing there is no long term. What can be said is that knowing there is a period in which the two point opposite ways is the condition for holding the design.

Of the three, the first answers what is holding things up, the second whom to address, the third what to hand over. The three are not independent. If what you hand over is only answers, making the axis of division a state changes nothing, since every state receives the same answer. Conversely, once you are handing over reasons, different states need different reasons, so dividing acquires a point. The order is therefore the third first and the second after.

This order also points the opposite way from the short-term numbers. Dividing can be done this week; designing what to hand over takes time. Settle the order by ease of starting and it inverts again.

All three also share a long lag before you can confirm whether they worked. The result of handing over a reason appears at the next occasion on which the person judges, not immediately after handing it over. A design whose confirmation takes time tends to lose when evaluated on short-term numbers. Settling the evaluation period in advance is the condition for holding this layer.

The Sixth Layer — Two Articles on How a Written Email Arrives and Gets Opened

The delivery layer looks easy to improve and holds a low ceiling on effect. Two articles set out that ceiling.

What writing an email newsletter pins down is where opening gets settled. Before opening, all that is visible is the sender and the subject line, and what settles whether the subject line gets read is the sender’s name. And where that name sits in a person’s order of importance was built by each message received in the past. How today’s message gets opened is a consequence of the past several months.

What the open rate actually is unpicks what the indicator counts. It counts loads, not readings. Since 2021 this measurement has been inflated by the recipient’s environment.

What runs through the two is that the numbers in hand move for reasons other than your operations: the composition of the list, the mechanics of measurement, how the denominator is taken. So what is usable at this layer is only the trajectory of your own numbers measured under the same conditions. Comparing with others carries no meaning beyond confirming an order of magnitude.

This layer looks easy to improve. Subject lines get written every time, so they can be improved every time. Improvement never finishes, so any amount of time can go into it. And the effect of the time spent is small compared with the layers upstream. Improve how things arrive and, if what you are delivering does not connect, the position of the name does not rise.

None of this says the layer is unimportant. If nothing arrives, nothing made well upstream matters. What can be said is that the ceiling on this layer’s problems is low. The effect of making something arrive that was not arriving is large; the effect of improving how something already arriving arrives is small. The two get treated as the same improvement and differ in size.

So how far does this layer need to go? Far enough that you can confirm whether things are arriving. The number that failed to arrive is known, the number opened is comparable under the same conditions, and the list can be viewed split by registration period. With those three in place, further precision changes no judgement.

And this layer yields a strong sense of improvement, because you write every time and numbers come back every time. The sense being strong, more time gets allocated here than to the layers upstream.

The way to restore the allocation is to settle it from the ceiling side. Mark off in advance the time this layer gets, and the surplus flows automatically upstream. Leave it as until it gets better and it never ends, because it keeps getting better. Work that never ends absorbs the other layers’ time without limit.

In What Order to Take On the Six Layers

The eighteen articles need not be read in the order written. That said, the order in which they get worked on as a design carries dependencies.

What must be settled first is the first layer. The cut of the stages and the position of the convergence are premises for everything else. Unsettled, measurement has no fixed meaning.

The second layer comes next. What gets handed over first builds the standard for what follows. What is handed over here cannot be taken back, so settling it early is cheaper.

The third and fourth layers come into use once records have accumulated. At an early age there are not enough cases for a proportion to be readable. Judge by proportions while the count is small and you cannot see that what moved was one or two people.

The fifth and sixth layers get adjusted while running. Neither is settled once and left.

This order carries a practical difficulty. The first and second layers get settled when the least information is available. Information arrives once the third and fourth become usable. So the order runs settle, then measure. Measuring and then settling is not available.

None of this makes the first decision a guess. Only the grounds cannot rest on measurement; grounds themselves can still be placed. What business this is, whose what it solves, how far your range extends. None of those require measurement.

And starting from measurement means using numbers produced by the present form to evaluate the present form. A measurement whose age does not fit cannot separate not moving from not moving yet.

The order has a practical exception: rebuilding a business already running. There the records already exist. That said, those records were produced by the present form, so they say nothing about anything outside it. The current stage numbers cannot be used to judge whether to change the cut of the stages.

What they can be used for reaches as far as locating where things are not working inside the present form. Beyond that, whether to change the cut itself is a design judgement rather than a measurement.

Changing the cut also carries a price. From the moment you change it, the records up to then stop serving as a basis for comparison. Whether the numbers afterwards got better or worse is also unreadable for a while. Whether that unreadable period can be absorbed is the practical condition for changing.

And the more records there are, the harder this distinction becomes, because plentiful material makes it look as though the answer will come out of the material. What is not there does not come out however much material you gather.

What Necessarily Occurs in the Course of Designing a Funnel

Look across the eighteen and the same shape of event recurs. There are four.

First, things that work fast and things that work slowly get presented in the same list. Lists of measures, lists of advice, lists of indicators. Once lined up, they look to the chooser like the same kind of thing. And the choosing settles into order of least effort. Least effort tends to overlap with shortest effect. Fast-working things generate operational work, and the work generated takes time from the slow-working side.

Second, the measured stage becomes the object of optimisation. This is unavoidable. The act of measuring turns that stage into a target. So the question is not what to measure but what may be placed as a target. The moment it is placed as a target, the number stops holding the meaning it had before.

Third, indicators move identically in two opposite states. Rising because the relationship thickened, and rising because leaving got harder. Leaving having achieved the purpose, and leaving in disappointment. The method of telling them apart has the same shape every time: put the other person in a state where they can freely leave, and see what happens. Since this depresses the short-term numbers, it is the confirmation with the highest resistance to being executed.

Fourth, at every layer, the most effective confirmation is the least executed operation. Open the exit, stop the measure, send without changing the subject line. Each either depresses the short-term numbers or carries the possibility of doing so.

None of the four is the kind of thing knowledge lets you avoid. Even knowing, in a busy period the lower-effort option gets chosen. What knowing achieves is that awareness remains after the choice. With awareness remaining, you can return months later. Without it, a structure nobody remembers choosing has been built.

A single line of note is enough to keep the awareness. Write down this month I chose the fast-working one. Written, you can read it back in three months and see that it was a choice. Unwritten, all that is visible in three months is the result. From a result you cannot read whether it was the consequence of a choice or the only thing available.

Being hard to execute is not a matter of capability. It is a structural matter: as long as a number is held as a target, an operation that depresses it always looks like a loss. So the response also sits on the structural side — settle the confirmation period in advance, and take the number out of evaluation for that period.

Reading Whether the Stages Are Cut by Involvement, in Three Ways

Whether the cut is right can be read. What you look at when reading is the cut itself rather than the headcount at each stage.

The first way is whether each stage can be described without touching price or proximity to purchase. If all you can say is people who have not bought and people who bought once, the stages are cut on the transaction side. If you can say people still sorting out what the problem is, they are cut by involvement.

The second way is whether the same person ever moves back a stage. If the design makes returning unrecordable, the one-way premise has been implemented as it stands. People do return in practice, so a design that cannot record returns cannot show the reality.

The third way is how many stages exist after the purchase. If it ends at purchase, the diagram is arranged in the order of selling. If stages of involvement exist after the purchase, it is arranged in the order of the relationship. Of the eighteen, seven handle what comes after the purchase for this reason.

Headcount plays no part in any of the three ways. Since the question is whether the stages are cut by involvement, what has to be confirmed is the cut.

The three share an applying condition: all of them require several stages and a record that has run for some period. At an early stage there is nothing to confirm. What can be done then is to write out the stages you presently hold in words other than proximity to purchase. If you cannot, they are cut on the transaction side. If you can, those words become the starting point for observation.

You could read the third way as requiring stages after the purchase, and object that it fails to apply to businesses centred on one-off transactions. The way it applies changes. Even in a one-off transaction there is referral and there is return, and those two are separate stages of involvement. The absence of stages is frequently caused not by the transaction being one-off but by the later stages never having been designed. Undesigned, what happens after the purchase goes unrecorded, and unrecorded means it does not exist.

That said, there are fields that genuinely end in one round. There the third way cannot be used, and what remains is the first and the second.

The funnel diagram carries two things as one: the shape that is wide at the top and narrow at the bottom, and the cut of the stages. The shape comes from bond sales in 1924, and the cut is settled outside the diagram. Separated, the shape can be kept while only the cut gets replaced.

The diagram was not wrong. Dividing into stages and counting is useful, and that does not change. What moved is the single point of what the stages get cut by. Cut by involvement, what each stage should hand over becomes settled, and whether what was handed over worked becomes readable in the substance of the other party’s judgement rather than in the headcount at the next stage.

What order your own funnel is presently arranged in can be read by the three ways of reading the cut of the stages. Even if what you read is the order of selling, that is not a failure. Given that the diagram in general circulation is arranged that way, starting there is the ordinary case.

One further note: what this design work is actually rebuilding, at the level of the business as a whole, is set out in structural autonomy.