In Building a Persona the Audience Is Not an Input but an Output of Settling the Criterion of Value

Give a fictional person an age, an occupation, an income, a family, a way of spending weekends. The more fields get filled, the more that person acquires an outline and begins to look usable as a basis for judgement.

How dense the outline becomes and whether that person exists move in opposite directions. Stack independent attributes and the number of people satisfying all of them at once falls away rapidly. This is not a figure of speech; it is a relation that can be written as a product. Only the sense that deciding in more detail makes things more certain points the other way from the arithmetic.

And what actually gets consulted in the judgements of writing is not the attributes you lined up. Settling a heading, selecting an example, deciding how far to explain — what the hand touches is what that person feels to be important.

Let me raise three questions that receive answers. First, what the work of adding attributes is doing as narrowing. Second, what problem this method was originally built to solve, and what it is used for now. Third, which settles first, the audience or the criterion of value.

One thing cannot be decided here. Whether the audience narrowed by a criterion exists in sufficient numbers in the market cannot be confirmed from inside this work. It is not known until you deliver. How detailed a portrait to write, and how far the audience may be narrowed, do not come out either, since they vary with the territory and the unit price.

Place the audience as an input or as an output and what you are obliged to settle swaps over. As an input, what you settle is how to investigate: find out who wants what and build to match. As an output, what you settle is what you yourself count as value, and the audience remains behind it. In the former no criterion is required on your side; in the latter a criterion is required first.

Where the mix-up actually bites is in the form of a portrait that gets consulted not once in the judgements that follow. A design document that goes unconsulted stays there quietly. Anything found wrong can be fixed, but what is unused never has its errors exposed. The time spent settles as cost, and the next period runs the same procedure again.

And going unconsulted does not surface as a problem of the quality of judgement, since what was written did get written. What surfaces is that the articles fail to mesh with one another. Separate articles can assume separate readers and each will still hold up when read on its own.

The property of stacking attributes, the origin of this method, and the order by which the audience settles only after the criterion are examined below, in that order.

How Is Building a Persona Taught as a Procedure

A persona is a portrait of the typical customer of your product or service, set out in detail as though a real single human being. It is described as a method of narrowing the audience through a concrete person rather than a statistical band.

The fields to set get shown. Age, sex, place of residence, occupation, position, income, family. On top of these, values, the shape of daily life, how information is gathered, the difficulties being carried. Including the inner life and not attributes alone is said to be the point.

The procedure gets shown as well. First, research: interviews with existing customers, questionnaires, records of use, information from the sales side. Second, organising the information and extracting what is common. Third, integration into one person. Fourth, entry into a template and sharing.

The effect described is that the axis of judgement is aligned. Where several people are involved, a divergence over who is being addressed sends each measure in a different direction. Share a portrait and this dispersion falls, runs the account.

Points to watch come attached too. The failure most easily fallen into is building an ideal portrait convenient to yourselves, runs the caution. Base it on the customer’s reality rather than the company’s ideal, written over and over. Criticism of the method itself gets introduced as well: that dividing people by attributes has stopped matching reality.

That a divergence in recognition produces dispersion in measures is a fact, and drifting toward an ideal portrait happens too. As a record of the procedure there is no item missing.

One premise is shared across the procedures lined up here. The persona is handled as an input to the design. Build it first, and settle what to supply and how on that basis. The order is fixed.

It functions as an input in one situation: where what is supplied is already fixed and only the question of whom to deliver it to is open. Here, identifying the audience becomes an independent problem.

What happens once it is organised entirely as an input is that the direction gets fixed: the substance of what is supplied gets derived from the audience. Find out who wants what, and build to match. Proceed in this direction and no criterion is required on your own side at all. You are supplying what is asked for, so the judgement sits in the market. And a judgement sitting in the market can be read off by anybody else as well.

Being readable means the same research yields the same portrait. Several writers in the same field, working independently, tend to arrive at a similar person. If the arrival point is the same, what gets derived from it resembles as well. This convergence is invisible from inside each piece of work, because the portrait emerged as the result of one’s own research.

Stack Attributes and the Probability That the Assembled Portrait Exists Falls

Detailing has a consequence visible from the side of number.

Chapman and Milham raised a point about the validity of personas (Chapman & Milham, 2006, Proceedings of the Human Factors and Ergonomics Society Annual Meeting, 50(5), 634–636). A person satisfying many attributes at once becomes rare rapidly as the number of attributes grows.

The arithmetic is simple. Say three in ten satisfy one attribute. Stack two independent attributes and you have nine per cent; three, two point seven; five, zero point two four. With ten attributes it falls to around zero point zero zero zero six per cent. Each setting looks natural on its own, and at the moment of combination that person all but disappears.

The point was later checked against real data. Chapman and colleagues counted matching respondents in a national survey dataset while increasing the number of attributes (Chapman, Love, Milham, ElRif & Alford, 2008, Proceedings of the Human Factors and Ergonomics Society Annual Meeting, 52(16), 1107–1111). Combinations for which not a single matching person could be found appear after only a handful of attributes have been stacked.

What both papers handle is how many people in a population match a set combination of attributes. Neither measured the results of writing produced with such a portrait. What they showed is the direction: the range within which detailing works as narrowing is narrow.

And a written portrait reads as though it exists. There is a name, a photograph, a description of daily life. A sense of existence can be built independently of the probability of existence.

You can read this as meaning a representative need not exist. For a representative, an average will do. A person assembled from the average of each field, however, is unlikely to resemble a real person. Somebody of average height, average weight and average income is a point in a table, not somebody with a life.

Attributes do not thereby become unnecessary. Where the form of supply depends on an attribute, the attribute genuinely operates as a constraint: a trade limited to one area, a scheme only a certain age group can use. What jams is lining up attributes that are not constraints.

The arithmetic has one further implication. Where attributes are not independent — occupation and income, for instance — the probability falls more gently. That is, lining up related attributes has not advanced the narrowing. Line up independent attributes and the narrowing advances rapidly while moving away from existence. Either way, detailing has a ceiling.

A detailed setting gives the person who built it reassurance: having decided this concretely, it cannot be far off. Concreteness and accuracy are different things, and a setting that is concrete and wrong gets corrected less readily than one that is abstract and wrong, because the more detail there is, the harder it becomes to see which part should be corrected.

Trace the Origin and the Problem Being Solved Was Not Narrowing

Look at what this method was built for and the question of order comes into view.

The persona as a method is best known from the form Cooper presented in his 1999 book (Alan Cooper, The Inmates Are Running the Asylum, Sams, 1999). It was a tool for solving a problem in software design.

The problem being solved was not the identification of an audience. It was that when designers use the word user, that user deforms conveniently from scene to scene.

When a feature is to be added, the user becomes somebody knowledgeable. When an explanation is to be dropped, the user becomes somebody quick on the uptake. The same word points to a different person each time.

The persona was placed to stop this deformation. Fix one person with a name and they can no longer be changed to suit convenience. The purpose was fixing, not narrowing.

The difference is not small. If fixing is the purpose, what is required is consistency rather than accuracy. Whether that person exists matters less than that they do not change midway.

You can take the position that whatever the origin, it is fine if it is now used for identifying an audience. It can be. But when the purpose changes, the necessary conditions change. If fixing is the purpose, more detail is better; if identification is the purpose, more detail moves further from existence.

The same procedure receives opposite evaluations depending on the purpose. Execute the procedure without settling which purpose you are using it for and you cannot adjudicate whether adding detail is a good thing or a bad one. Unable to adjudicate, adding detail becomes the default direction.

Where one person runs the business, the problem of fixing looks unlikely to arise. There are no multiple designers, so no divergence in recognition occurs, and the original benefit thins out.

The elastic user, however, occurs when writing alone as well, in the form of the writer changing the image of the reader to suit the content they want to write. A knowledgeable reader when the material is difficult, a first-time reader when it is to be written plainly. This deformation occurs inside a single article too. Where the reader assumed in the first half differs from the one assumed in the second, the grain of explanation fails to line up. It reaches the reader as an unexplained difficulty.

Execute the procedure without confirming the purpose and the use of the product is unclear. The state in which a built portrait is never once consulted in later judgements is born along this route.

What the Judgements of Writing Consult Is Not the Attributes

When writing, which part of the setting actually gets used?

Settling a heading. Selecting an example. Deciding how far to explain. What gets consulted in these judgements is what that person feels to be important.

Age is not consulted directly in the judgements of writing. Nor are income, family, or how weekends are spent.

The reason attributes get lined up is that they are used as a proxy. Late thirties, children, income at a ceiling, therefore probably interested in the use of time. The estimate is sometimes right. A proxy, however, is not the thing itself. Where the thing itself can be written directly, the proxy is unnecessary.

Set two side by side. Thirty-eight, five years independent, income six million and flat. And: feels the problem to be less the size of the income than the inability to settle next month’s schedule. Only the second can be used in the judgements of writing.

The second cannot be derived from the first, because among people with the same attributes there are those who feel the size of the income to be the problem. And the writing addressed to those two people comes out entirely different.

The point that you have trouble searching without attributes holds. Where you use a mechanism that specifies an audience by attribute, as in ad delivery, they are required. That is a technical requirement of delivery, not a requirement of the judgements of writing. Mix the two and the judgements of writing get made by attribute as well.

Estimating importance from attributes is not without value either. At a stage with no other foothold, there are situations where it is the only place to begin. The estimate, however, is a hypothesis, and a hypothesis is something to confirm.

Confirming is easy. Look at the finished article and raise which of the attributes you set actually bore on a judgement. Attributes that do not get raised are fields you need not write next time. What gets raised may be less an attribute than the structure the person is placed in. Income is proportional to their own working hours. They do not hold their own contact list. These are descriptions of state rather than attributes, and they can be used directly in the judgements of writing.

Being able to write only in attributes means the criterion of value has not been put into words on your side. With what you count as important settled inside you, it can be written directly. The detail of the attributes is frequently used to fill in for the absence of a criterion. And while it appears filled, the absence of a criterion never rises as a problem.

The Audience Can Be Narrowed Only After the Criterion of Value Is Settled

What settles first is what you yourself count as value. The audience remains behind it.

The reason is simple. To narrow an audience by what they feel to be important, that criterion must be present on your side. Without it, neither agreement nor disagreement in importance can be adjudicated.

Say you hold the criterion place sovereignty over your time above the size of the income. With this criterion, the audience narrows automatically: people who hold the same criterion, or in whom something moves when the criterion is touched.

Try to narrow the audience alone without a criterion and the axis of narrowing will not settle. So you narrow by attribute. Attributes are observable by anybody, so they can be used with no criterion in hand.

This order bears on the side of what is supplied as well. With the criterion first, what to build settles from it too. With the audience first, you build what the audience asks for, and this becomes the form in which the conclusion of research gets implemented directly.

The worry that deciding by your own criterion turns solipsistic holds. It turns on whether that criterion can account for the reader’s situation. If the criterion names a state the reader already carries, it is not solipsistic. The person named feels their own state put into words. If it names nothing, it is the writer’s preference rather than a criterion.

A criterion is not necessarily in words from the start. It may appear first, over the course of working, as I do not want to take this or something about this is off. Words come later. Because of this order, the stage at which a criterion appears absent and the stage at which it is not yet in words need separating.

Whether it has become words has a confirmable form. Write your criterion in one sentence and see whether a position opposing it holds. If opposition does not hold, it is a generality rather than a criterion.

I want to deliver something good admits no opposition. Place being able to settle your own schedule above the size of the income admits opposition: the position that the amount matters more holds perfectly well. Only statements that admit opposition divide an audience.

Seeing whether opposition holds has a secondary effect. Raise a criterion that admits opposition and the people who leave appear first. Look at who left and you learn what the criterion is actually dividing. If nobody leaves, that criterion is not operating as a boundary.

Put the criterion first and the audience comes out narrower than expected. An audience narrowed by criterion tends to be smaller than one narrowed by attribute. How narrowness bears varies with the form of the business, and is separated by unit price and duration.

Begin From Research and the Present State Gets Reproduced

Research itself is effective, and what needs adjudicating is where it is placed.

Ask existing customers and the portrait of existing customers comes out. This looks like a tautology and tends to be overlooked in practice. Existing customers are the set that remains after selection by what is currently supplied. So a portrait built from research reproduces the present state. Where you want to change it, it cannot be used.

And extract what is common across several people and the uncommon parts drop away. In the dropped parts sits, sometimes, the reason that person is that person. What remains is the part that applies to anybody.

It can also be read as meaning that taking what is common is precisely what makes it reach many people. It reaches. It reaches shallowly. Writing composed only of what is common is denied by nobody and recognised as their own by nobody.

What reaches deeply contains the uncommon part. Those to whom that part does not apply leave, and to those to whom it applies it reaches strongly. This asymmetry changes meaning with the scale of the business. In a trade addressing many, the number leaving is larger. In a trade continuing long with few, the depth of reach is what bears.

A constraint enters on the side of asking as well. Ask what are you having trouble with and what comes back is trouble the other party has already put into words. A state not yet in words cannot be extracted in the form of a question. And part of the value a business handles sits on the unworded side, because if it were in words the other party would already be looking for a solution themselves. How to remove this constraint is handled in how to conduct a customer interview.

The accurate statement is that there are things research can take and things it cannot. For confirming facts it can be used. For confirming, after a criterion has been settled, that people matching that criterion exist, it is indispensable. What it is unsuited to is discovering the side of value.

In order, research comes second. Placed at the start, it yields a portrait that reproduces the present state. Placed at the point of confirmation, it tells you whether the criterion you settled hits reality. The same work does different work depending on where it is placed.

Placing it at the start has one further consequence. With research results as an external ground, there is no longer a need to decide for yourself. What was never decided will not be found later either, when something goes wrong, because no record of a decision remains. What remains is the record of the research, and no one’s judgement is in it.

What Can Be Placed as an Example Is a Real Person, Not a Fictional One

As an example of building a persona, there is the form of placing a real person instead of assembling a fictional one.

There are two conditions. First, somebody you have actually dealt with. Second, somebody in whom there was a reaction when your criterion was touched.

With even one person satisfying both, you can begin there. Collecting attributes is unnecessary. What that person feels to be a problem, and what they moved on hearing, have already been observed.

The difference from a fictional person sits on the side of probability. A real person exists with probability one. The combination cannot become unnatural, and they cannot deform to suit convenience.

And concretes can be drawn from memory. The words that person used, the order of their questions, the place where they were not convinced. None of these comes out of a fictional setting.

The point that one person skews things holds. It skews. The advantage is that the direction of the skew is known. The skew of a fictional person points in the direction of the builder’s assumptions, and that direction is invisible to the builder. With a real person you can identify that they are unusual in this particular respect, and having identified it, you can discount that part in use.

At some stages you have dealt with nobody: the period just after starting. Here you end up using your own past state. Your past self existed, and what you had trouble with remains in memory.

Using your past self requires one caution. You remember what you had trouble with, and how you understood it at the time has been overwritten by your present understanding. What can be drawn from memory is the tidied form that came after the resolution.

Separate what can be drawn from what cannot. Concrete events, words actually spoken, the conditions of life at the time can be drawn. What I thought the problem was is a field with a high probability of having been overwritten. Where a record from the time survives, that takes precedence. The mechanism of the overwriting is handled in how to write from primary information.

Using a real person strengthens the sense of writing to that person. This works as an advantage, and their individual circumstances can also pull the range too narrow. Swap in a different person periodically and reread, and this narrowing becomes detectable.

The swap has one further use. If the judgements of writing do not change when you place a second person, those two sit in the same position. If they change, what changed is the boundary of the audience. The boundary comes out faster from setting two people side by side than from raising people who do not apply.

A Narrow Audience Works For and Against You Depending on the Form of the Business

The audience comes out narrow because the criterion of value settles before it.

In a trade of low unit price that requires numbers, narrowness works against you, because there is a possibility of not reaching the necessary count. Here you take the form of increasing points of contact while holding the criterion. Loosen the criterion to widen the audience and you stand, to that extent, in the same place as other writers.

In a trade of high unit price continuing over a long period, narrowness works for you, because the depth of reach bears more than the number of people leaving. In a form of continuing long with few, writing composed only of what is common does not function.

Narrowness itself, therefore, cannot be adjudicated as good or bad. What can be adjudicated is whether that narrowness meshes with the form of your own business. Where it does not mesh, what moves is the form of the business or the criterion of value, not the audience.

Narrowness also carries a problem of measurement. An audience narrowed by criterion, unlike one narrowed by attribute, cannot be counted from outside. How many people place sovereignty over their time above the size of their income does not exist as a survey field. Uncountable, it cannot be adjudicated in advance whether the narrowing has gone too far.

Being unable to adjudicate has consequences in two directions. One is proceeding a long way while too narrow. The other is assuming it is too narrow and loosening the criterion before confirming. The latter happens more readily and is harder to notice in yourself, because the loosened side leaves behind the observation that reactions rose a little.

The confirmable form sits only on the side of the delivered result. If the reactions touched by the criterion are consistent in kind even when small in number, the audience exists. If the reactions scatter with nothing in common, either the criterion is not reaching or the criterion is not dividing the audience.

And this confirmation takes a period. In a form addressing few, the base is small, so short-period figures swing. Move the criterion on the ground of swinging figures and the reason for the move becomes unrecoverable afterwards. The judgement to change a criterion gets made after writing down the criterion as it stood before the change.

Writing it down takes one line: the date, the criterion in one sentence, and the reason for the change. With those three surviving, you can judge six months later whether to return. Without them, you cannot explain how the present criterion came to be selected. A criterion you cannot explain will move again the next time reactions fall. With every move the audience swaps over, and what you have written stops accumulating.

Discerning What the Portrait in Your Hands Was Built For

The audience you set can be checked for function in three ways. All can be carried out on your own.

The first confirmation is to see whether you can raise three kinds of people the portrait does not apply to.

If you cannot, it is not narrowed. To settle an audience is also to settle whom you are not addressing. Raise three, and if they are a real segment, a boundary has been drawn.

The second confirmation is to delete the fields you set one at a time and see whether the judgements of writing change.

Delete age and if the judgements do not change, age is not being used. Only fields that produce a change are the fields actually functioning. Sometimes only one or two remain. What remains is the real axis of narrowing in that business.

The third confirmation is to see whether another person in the same field displays the same portrait.

If they do, that portrait is the market average rather than something out of your own criterion. If they do not, it comes from your side. This can be confirmed by searching alone.

The detail of the setting goes unexamined by all three. Detail is not evidence of being narrowed.

There is one line the three cannot cross. None of them can confirm whether that audience exists. Whether an audience narrowed by criterion exists in sufficient numbers in the market is not known until you deliver. In order, you settle the criterion first and defer the confirmation of existence. Reverse it and you settle the criterion to fit an audience whose existence has been confirmed, and the criterion stops coming from your side.

How many fields to set, how far the audience may be narrowed, how many people suffice for confirmation. No numbers are attached to any of these. They vary with the territory handled, the unit price of what is supplied, and how the business is to be continued. Hand over a figure and the judgement gets made for you without those conditions being looked at.

A denser outline and a firmer probability of existence point opposite ways because attributes were not built as a tool for narrowing. This method was born as a tool for fixing one person against deformation and came to be used as a tool for identifying an audience. The same procedure serves two purposes, and in one detail is an ally while in the other detail is an enemy.

A portrait built in order not to change may be written in detail. A portrait built in order to narrow is working through the criterion rather than the detail. Which of the two you are holding can be told from three confirmations: naming people it does not fit, deleting fields, and comparing it with portraits in the same field. Which of the two you keep using is settled by the business you are running.

How the criterion of value itself gets put into words is handled in how to build a worldview and how to design a concept. The method of taking material from one real person is in how to conduct a customer interview, and the stage of settling what to write from it is taken up by what to do when you do not know what to write. The order by which settling the criterion first makes the audience come out as an output shows up in the same shape in the work of putting things into words. That pressing on brings out what was never settled is in what verbalisation is.