Client Service Boundaries: Where Service Stops and Advice Begins
Service is telling somebody what the work shows. Advice is telling them what to do about it. The boundary sits exactly there, and a question crosses it the moment the answer would be a course of action rather than a finding. The reply names what the work covers, what it does not, and offers to run the case the person is actually asking about.
One fact makes this harder than it sounds. Almost nobody who walks an analyst towards this line is trying to. In the usual picture of the problem, somebody leans across a table and presses for a view that is not theirs to give. If crossings happened that way the whole subject would be easy. Pressure like that is visible from the other end of the corridor. The truth is warmer and much quieter. Somebody asks the question they genuinely have. Wanting to know what to do next is the only reason anybody comes looking, so the question they genuinely have is almost always about exactly that. In the ordinary case there is no bad intent to detect, so the boundary cannot be held by detecting it. A method that depends on spotting a difficult person will fail on every pleasant one.
The check therefore moves off them and onto the person answering. Nobody can inspect somebody else's motive in the second and a half available before an answer is expected. The sentence about to be spoken can be inspected. The sentence is already half formed, it belongs to the speaker, and looking at its shape rather than its content takes no time at all. Is the thing about to be said something the work shows, or something somebody should do? The whole test is that one question. Everything else is the reasoning underneath it and the wording that follows from it.
Sharada Iyer is seven months into a role at the Kavery research desk inside Kavery Capital Services Private Limited, an invented firm, and produced a credit review on a line to Meenakshi Tubes Private Limited. Prakash Nadar reviewed it. Latha Menon chaired the committee that met on 12 March, took the decision and dated the record: the line falls from Rs 40,00,000 to Rs 25,00,000, a reduction of Rs 15,00,000, and comes back for review in 90 days. The finding underneath it was that the operating margin had gone from 12.0 per cent to 10.8 per cent against an internal floor of 11 per cent.
Client Service vs Investment Advice: What Is the One Sentence That Separates Them?
Client serviceTelling somebody what the work shows, in a form they can use. It is the whole of what a research desk sells. is telling somebody what the work shows. AdviceTelling somebody what to do about what the work shows. A different activity, with different duties attached to it. is telling them what to do about it. The definition is complete at that, not a simplification of a longer one. The difference can be stated that briefly because it sits in the grammar of the sentence rather than in the subject matter, the seniority of the person asking or the setting.
Outside finance the shape is completely familiar. A surveyor who walks a house and reports that the beam over the kitchen has moved by a certain amount and that the movement is recent has said what the survey found. A surveyor who says not to buy the house has said what to do. Both sentences might be said by the same person on the same afternoon, and only one of them is the thing the survey was paid for. The doctor who says a blood report shows this reading against this range is doing the first. The pharmacist at the counter who says take two of these is doing something else entirely, and both of them know precisely which one they are doing at every moment.
Now put it back. A findingSomething the work shows, with a source, a date and a workpaper behind it. It can be checked by somebody who was not there. is something the work shows, and it can be traced: there is an item number, a document and a date sitting behind it, and somebody who was not in the room can rebuild it. A course of actionWhat somebody should do next. It is a decision, and it belongs to whoever holds the decision rather than to whoever produced the analysis. is what somebody should do next, and it traces to nothing at all. No analysis has ever contained a decision. Decisions are made by the people who hold them, using findings among other things, and on this assignment the person holding it was Latha Menon, chairing on 12 March, with five other people in the room.
The sentence that separates the two is short enough to hold under pressure: a finding describes, a course of action instructs, and only the first one was produced by the work. Two things are absent from that sentence. How confident the speaker is does not appear, and nor does whether they happen to be right. Both are irrelevant, and the second one is the trap that catches careful people. Rightness is not what makes a sentence theirs to say. A person can be entirely right about what somebody should do and still have crossed.
The shape of an advice answer is what has to be recognised as it arrives, and the shape is short, unsourced and grammatically pointed at somebody else's decision. The sentence itself is not the thing worth carrying away.
Two sentences about the same reported year. Which one is a finding?
What Is a Professional Boundary, and Who Holds It?
A professional boundaryThe limit of what a person will speak to, set by what their work supports. It is held by that person; nobody hands it over and nobody checks it. is the limit of what a person will speak to, and who sets that limit is the first surprise. Not a regulator, not a manager, not the terms of an engagement letter. Regulators, managers and engagement letters all exist and all matter, and several of them are covered separately. A professional boundary is set by one thing only: what the work actually supports. The boundary moves when the work moves. On an assignment with a forward case built on stated assumptions, the boundary sits further out than it does on an assignment where only one year of filed accounts has been read, and it sits in a different place again on the same assignment a month later when three more items have gone into the source log.
A limit set that way is called professional rather than legal. Nobody hands it over at induction, nobody checks it, and on most days nobody would ever know if it moved. The boundary is a property of the work rather than of the person. A person working alone, with no team and nobody senior in the room, holds exactly the same boundary as the head of the desk. Professional practice rarely works this way. Being seven months in costs nothing at all: what a body of work supports has the same answer whoever is asking.
Because the boundary is set by the work rather than by a rule, the only person who can move it unnoticed is the person holding it. The test therefore has to be something run rather than something looked up. A rule can be consulted after the fact. A boundary held on the shape of one's own sentences has to be checked in the moment, and the moment is about a second and a half long.
How Is the Test Run in the Half Second Before Speaking?
The sequence is genuinely this small. The question lands. The answer forms, as it almost always does. The difficulty is that the answer is usually known and usually a good one. Before it leaves, it gets looked at once, and the question is which of two kinds of thing it is. Does it describe something, or does it instruct somebody? If it describes, it goes. If it instructs, it stops, and the three-part answer set out below takes its place.
Two properties make this workable where every other version of the check fails. The first is that it operates on material already to hand. The half-formed sentence is right there; nobody's motives are being reconstructed and no policy is being recalled. The second is that it produces a binary. There is no scale from service to advice with a grey zone in the middle where judgement is required. In a live conversation with somebody watching, a scale is the same as no test at all. Describes or instructs. Two boxes, and every sentence ever said falls into one of them.
Why Is the Test About the Answer Rather Than About the Question?
Because questions are unreliable evidence and answers are not. The same eight words, asked by two people in two moods, can be a request for information, a request for reassurance, a request for cover, or an actual request for a decision, and none of that can be told from the outside. Worse, the person asking often does not know either. The relationship manager who asked about the December order was not conducting a test. The manager had a customer they had held for four years, a decision that had gone against that customer on 12 March, and a genuine wish to know whether the situation might change. Underneath sat a perfectly honest interest in keeping the account. An interest like that makes the manager a professional with a different job rather than an obstacle.
The answer, though, is fully knowable to the person giving it. The answer has one grammar and one referent. The answer either points at the work or it points at somebody's decision. Moving the check from their question to one's own answer converts an unanswerable question about another person's mind into a trivial question about the next eight words, and no other move in the subject does as much.
There is a second reason and it is worth having. A boundary held by reading intent is held unevenly. Intent-reading holds the boundary carefully with people who feel like they are pushing and loosely with people who feel friendly. The weighting is precisely backwards. The friendly conversation at the end of a good meeting is where almost every crossing actually happens. A test that runs on one's own sentence runs the same way in both conversations. Somebody who cannot see into anybody has no other kind of consistency available.
Somebody asks the analyst: on this, would you keep the line where it is? The test applies.
Which Question Shapes Cross the Boundary, and How Does Each One Sound?
Three shapes account for nearly all of it, and they are worth knowing by ear because the ear works faster than the test does. Knowing them does not replace the test; it means the test has already started running before the question has finished arriving.
The first is the direct one: what would you do. The direct shape is honest, it is common, and it is the easiest of the three: nobody can mistake what is being asked for. The second is the disguised one, and it is the most frequent by a distance: what does this mean for us. The disguised shape is not automatically a crossing, and its slipperiness is exactly that. Answered as a pure finding, it says what the numbers show about the current position; answered as a plan, it says something else entirely, and the two answers begin with the same four words. The third is the hypothetical, and it is the most dangerous of the three because it sounds like analysis rather than like a request. If the order repeats, should the line go up. The hypothetical arrives dressed in conditionals and arithmetic, so it feels like a modelling question. The word sitting in the middle of it is should.
Against all three, one control case: how did you get to 10.8 per cent is not a boundary question at all, it is a request to see the working, and answering it fully is the most straightforwardly useful thing anybody does all day. Keeping that fourth shape clearly in view matters. Somebody who becomes nervous about every question starts hedging explanations, and a hedged explanation is its own kind of failure.
Which of these crosses the boundary?
What Is Said Instead, When Refusing and Crossing Are Both Wrong?
The answer has three parts and it takes about twenty seconds to deliver. Part one: what the work covers. On this assignment, the year just reported, on the filed accounts, with the margin at 10.8 per cent and the December order of Rs 2,96,80,000 sitting at 14 per cent of revenue inside that figure. Part two: what it does not cover. Any forward year at all, and any view on the line beyond the 90 day review that was agreed on 12 March. Part three, and this is the one that decides whether the whole thing works: the offerProposing to run the case the person is actually asking about, with the assumption written down. It converts a boundary from a refusal into work that can be done. to run the case they are actually asking about, with what it would need and roughly when it could be with them.
Part three does something mechanical. Parts one and two, on their own, are a complete and accurate reply, and they leave the person exactly where they started, holding a question and now also holding the knowledge that no help is coming with it. Part three converts the boundary from a wall into work that can be done. The offer says, in effect: the question being asked is real, it is answerable, the document on the table does not answer it, and here is what would. The boundary stops being about what will not be said and becomes about what would be needed in order to say something, and that is a completely different conversation to be standing in.
None of the three parts requires any standing. Nothing is declined, no policy is invoked, and nobody is told what their job is. The three parts describe the edges of one's own work. A person seven months into a role can do that with total confidence. Nobody in the building knows those edges better than the person who did the work.
The answer gives what the work covers and what it does not, and stops there. What is missing?
Why Is Refusing Flatly Also a Failure, and What Does It Cost?
There is a version of boundary-holding that is really just self-protection with better vocabulary. The flat refusal sounds correct, it cannot be criticised in any meeting, and it ends every conversation of this kind in about four seconds: that is not something I can comment on. Nothing about that sentence is false. The refusal is also, in practice, a failure. The reasoning for that is not sentiment, and it is worth being exact about.
The question was real. Somebody has a decision to make and a gap in what they know, and the gap did not close when the question went unanswered. The decision still gets made. Only the source of their input has changed, and the alternatives are worse in a specific way: they will get an answer with less evidence behind it, from somebody with less of the work in their head, delivered as an unhedged sentence with no conditions attached. People who have not done the work are considerably more comfortable being brief about it. A flat refusal does not remove a bad answer from the world; it removes the good one and leaves the field to one that carries nothing.
There is a cost to the analyst as well, and it compounds. The person who was declined once does not come back a second time. Over a year that means the questions stop arriving, and the questions arriving is how anybody finds out what the desk actually needed to know. Set the two twenty-second replies side by side. One of them ends the exchange and protects the person answering. The other ends the exchange, leaves the person asking better off, keeps a job in the queue, and costs exactly the same twenty seconds. Only one of them is the professional version.
Why is a flat refusal also a failure?
How to Communicate a Finance Recommendation Without Overclaiming: Which Words Do Most of It?
OverclaimingStating a finding as firmer, broader or further-reaching than the work behind it actually supports. It is usually done by a word rather than by a claim. is the quieter half of this subject, and it catches more people than the boundary itself. The reason is that overclaiming happens inside sentences that are otherwise entirely legitimate. Nothing steps across into somebody else's decision. The speaker stays on their own side and describes what they found as firmer, broader or further-reaching than the work behind it goes. Overclaiming almost never happens through a false statement. Four words do it, and they are hard to stop noticing once seen.
Will is the first. Four letters convert a computed figure into a claim about a period the work never touched. Clearly is the second, and it is doing something sly: the word asserts that everybody present already agrees. Disagreeing now requires somebody to be the person who missed the obvious. Should is the third, and it is the one that actually crosses the boundary rather than merely stretching it. Should is an instruction with a polite face on. Obviously is the fourth, and it substitutes a word for a reason: wherever it appears, a because has quietly gone missing.
Every one of the four can be deleted without weakening any statement that was true. Deletion is therefore the test that settles the matter: if the sentence gets weaker when the word comes out, the word was carrying the claim rather than describing it. Run against any finished memo, the sentences that survive deletion were findings. The ones that deflate were something else, held up by adverbs.
Which version has the overclaiming taken out of: the margin will clearly stay under the floor next year?
How Client-Service Boundaries Protect Professional Finance Work: Who Is Actually Being Protected?
Asked why the boundary exists, most people answer with some version of staying out of trouble. Staying out of trouble is why holding the boundary feels defensive, and feeling defensive is what makes people hold it badly. Too hard produces the flat refusal. Not at all is the other failure, and it happens because being defensive in front of a friendly colleague is uncomfortable and the discomfort wins.
The consequence matters more than the feeling. Suppose a sentence gets said that should not have been said, and suppose somebody relies on it. The person who said it gets an awkward conversation and a correcting email, and on an unlucky day a note in a file. The person who relied on it took a decision on the strength of a sentence with nothing behind it. The person relying on it could not see the emptiness. The sentence came from the person who did the work, and that is exactly what made it credible. Checking was not available to them. Only one person in that exchange knew the sentence was unsupported. The duty to say so therefore sat with the speaker and not with the listener.
The exposure runs the opposite way to the anxiety: the person asking carries the consequence and the person answering carries only the embarrassment. The boundary is therefore mainly protecting the person asking. The reversal is not a rhetorical flourish; it changes how the boundary is held. A boundary held to protect the person holding it is held nervously and inconsistently. A boundary held because the other person cannot see what the analyst can see is held calmly, and it is much easier to explain out loud. The true thing can simply be said: nobody should be making a call on a sentence that has nothing underneath it.
Who does the boundary mainly protect?
What Did the Whole of This Look Like on One Assignment?
A method like this is worth nothing if it does not survive contact with an actual afternoon. Here is the exchange in full. The committee met on 12 March, the decision was taken and the record was dated the same day: the line falls from Rs 40,00,000 to Rs 25,00,000 and comes back in 90 days. As the room emptied, the relationship manager, who has held the account for four years and who had just heard a decision go against a customer they will now have to call, asked whether the line should be increased if the December order repeats.
The manager's question is the boundary, exactly and completely. The question takes the hypothetical shape, it arrives from somebody with an honest interest, and it comes at the end of a conversation that has gone as well as those conversations go. Every condition for a crossing is present at once.
Sharada Iyer gave the three-part answer, and it took about twenty seconds. Part one, the covered ground: the year just reported, on the filed accounts, with the margin at 10.8 per cent and the December order of Rs 2,96,80,000 sitting at 14 per cent of that revenue, already inside the figure rather than added on top of it. Part two, the uncovered ground: any forward year, and any view on the line beyond the 90 day review the committee had just agreed. And the offer: the repeat case can be run as a stated assumptionA condition written down before the arithmetic runs, so a reader can see exactly what the answer depends on and change it if they disagree., it needs the order terms and one line on whether a repeat is committed, and it can be back by a named day.
Nothing was refused and nothing was crossed, and the relationship manager walked away with more than a yes would have given them. A yes would have carried no conditions and the three-part answer carried three. There is a detail in it that is easy to miss and does most of the work. The first part of the answer contains something genuinely useful: the order is inside the 10.8 per cent rather than on top of it. One fact reframes what the relationship manager was actually holding, and it is a finding, traceable to item 5 of the source log. Crossing the line was not needed in order to be useful to them. Answering the covered part properly was enough.
| The three parts | What was said on 12 March | Why it is safe to say |
|---|---|---|
| What the work covers | The reported year on the filed accounts. Margin 10.8 per cent against the 11 per cent floor. The December order of Rs 2,96,80,000 at 14 per cent of revenue, inside that figure | Every clause traces to an item number and a date in the record |
| What it does not cover | Any forward year. Any view on the line beyond the 90 day review agreed the same day | Naming the edge of the work is itself a finding about the work |
| The offer | The repeat case can be run on a stated assumption. The case needs the order terms and one line on whether a repeat is committed, and it can be back by a named day | The offer proposes work rather than a conclusion, and the assumption is written down before anything is computed |
Predict: a warm, unwritten yes at the end of a good meeting. How long before it matters?
The failure: the helpful yes, and the six weeks after it
Change one thing about that corridor and nothing else. The three-part answer is given, correctly and in full. Then the conversation is pleasant, the other person looks relieved, and one more sentence gets added on the end. It is short. It is warm. The extra sentence names a course of action about the line and attaches no condition to any of it. Most people have heard one and most people have said one. The sentence does not need writing out for its sound to be recognisable.
Nothing happens. The absence of consequence is the first thing to understand about this failure: no moment of reckoning, no raised eyebrow, nobody writes anything down. The sentence was not in the memo, it was not in the pre-read circulated on 11 March, and it is nowhere in the decision record dated 12 March. The sentence exists in exactly one place, one other person's memory, and memory does something specific with material like that. Memory keeps the sentence and drops the conditions. The conditions were in a document and the sentence was in a voice.
Six weeks later the position has moved and the line has not, and somebody asks why. Back comes not the memo, not the 10.8 per cent, not the floor and not the 90 day review, but one remembered sentence from the person who did the work, now standing on its own with nothing around it and doing a job it was never built for. Probably no rule was broken, so the cost is not that. The cost is that considered work with a stated boundary was replaced, in the only place that mattered, by an offhand remark carrying no evidence and no conditions, and the remark is now what gets relied on.
The cruelty of it is the delay. On the day, the extra sentence was the friendliest thing in the conversation and it improved the relationship measurably. Six weeks is long enough that nobody connects the two, including the person who said it.
What Happens After the Line Has Already Been Crossed?
Everybody crosses it. Not often, and probably not badly, but it happens, most likely at the end of a good conversation with somebody the analyst likes. The recovery matters far more than the crossing itself, and it runs in a fixed order. The order matters because the first step is the one everybody skips.
The crossing gets written down, the same day. Not after the weekend, not once there has been time to think about whether it was really that bad. Write what was actually said, in the words it was said in. The sanitised version reconstructed on Monday is not the version sitting in the other person's head. Then state what the work actually supports, with the item numbers behind it. Then say plainly which of the two should be relied on, and say that it is the second one. Then send it to whoever heard it and to whoever holds the record of the decision.
The correction is cheap on the day and expensive at every point after that. The moment somebody acts on the remark it stops being a correction and becomes a contradiction of something already relied on. There is nothing brave about the same-day version; it is a slightly uncomfortable email that takes four minutes. People avoid it because writing it down feels like making the remark real, and the remark was already real the moment it was heard. All the email does is give the remark somewhere it can be checked, and somewhere it can be checked is exactly what the sentence was missing.
The crossing becomes obvious on the way back to the desk. What now?
Where Does Any of This Show Up Outside a Research Desk?
The same line runs through a great many jobs, and seeing it in four or five settings is what stops it looking like a rule invented for finance meetings. In every one of them the pattern is identical: somebody has done the work, somebody else has a decision to take, and the useful thing to hand over is what the work shows plus an honest statement of where it stops.
| Who is speaking | The question as it arrives | What the covered answer looks like |
|---|---|---|
| A relationship manager repeating a desk finding to a customer | So what does your analyst think we should do | What the review covered and what it did not, then the offer to have the forward case run |
| An analyst asked a question after a results call | Is this a good level to be at | What changed in the period the note covers, and where the note stops |
| A credit committee member reading a paper | Does this cover the renewal or only the current year | The stated coverage line at the front of the paper, put there for exactly this |
| An accountant asked a question at a wedding | Quick one, what should I do about this | What a general answer can cover, what it cannot without the papers, and come to the office |
| A household reading a surveyor's report on a house | So should we go ahead | What the survey inspected, what it could not see, and what a second inspection would need |
In every row the boundary is held by describing the work rather than by declining the question, and in every row the third part of the answer is work being offered rather than a door being closed. Notice also that the surveyor and the accountant hold the line under exactly the same social pressure the relationship manager applied in the corridor: a friendly setting, a person who genuinely wants help, and a question that would take four words to answer badly.
One last connection is worth making. The three-part answer puts the covered part first. Barbara Minto set out the same ordering rule in 1978: the substance goes at the front, and the qualifications hang off it. A boundary stated at the end of a long preamble sounds like an excuse being constructed in real time. The same boundary stated in the first eight seconds, followed by an offer, sounds like somebody who knows exactly what their work does.
Who may hold themselves out as an adviser is a separate question, and a regulated one
Everything above is a professional boundary held because of what the work supports, and that reasoning is the same in every market. Separately from it, there is in India a registration regime administered by the Securities and Exchange Board of India at sebi.gov.in governing who may hold themselves out as giving investment advice and in what circumstances. Such a regime exists, and anybody whose situation might touch it reads the current position at the source. Professional conduct expectations for members of the Institute of Chartered Accountants of India at icai.org, and the cross-border conduct principles published by the International Organization of Securities Commissions at iosco.org, sit alongside it. The distinction to carry away is that a registration category settles what a person may do; the boundary described here settles what their work supports, and the second one binds on days when the first one does not apply at all.
References
| Source | Document | Where |
|---|---|---|
| Securities and Exchange Board of India | The registration regime governing who may hold themselves out as giving investment advice in India | sebi.gov.in |
| Institute of Chartered Accountants of India | Professional conduct standards expecting a member to state what their work supports and not overstate it | icai.org |
| International Organization of Securities Commissions | Published conduct principles that several national regimes draw on, covering communications that do not overstate what is behind them | iosco.org |
| Barbara Minto | The Pyramid Principle, 1978, for the ordering rule that the substance goes at the front | Published work |
The Kavery research desk, Kavery Capital Services Private Limited, Meenakshi Tubes Private Limited, Sharada Iyer, Prakash Nadar and Latha Menon are invented.
Educational material. Not advice on any investment, tax, budget or market position.