How to Escalate a Financial Complaint, Step by Step

Escalating is moving one matter up one rung with everything it already has. The process is seven steps, and the substance of it is that almost nothing should change on the way up: the same request, the same reason, the same documents, plus the record of what the rung below said.

Carrying the matter unchanged is the whole of the craft, and it is the opposite of what almost everybody does. The levels themselves are covered separately: an entity's own grievance process, an ombudsman arrangement, the statutory consumer redressal commissions. So is the order they sit in. Moving a live matter from one of those levels to the next without damaging it on the way is a small, unglamorous act, and that act has a craft to it that can be written down in seven steps.

One invented household, the Bhosales, works through all seven steps. Meghna Bhosale works on a payslip, Ashok Bhosale runs a tailoring counter, and in June a health claim on their floater policy settled in a way they did not expect. The insurer paid Rs 91,440/- against a bill of Rs 1,42,000/-, so the household paid Rs 50,560/-, or 35.6 per cent of the bill. On 20 June they wrote and asked for Rs 24,000/- of it back. On 4 July an answer came declining. On 11 July they moved the matter up. The figures settle to the rupee. How a health claim is settled in the first place is a separate subject from moving one up a rung.

The instinct on escalating is to make the case stronger, and the instinct is wrong. A household that has spent three weeks on a matter, been declined once, and now faces a second round of paperwork feels the pull of making the second attempt count for more than the first. So it adds. In go the other things that went wrong, the tone of the reply, the amount it did not contest the first time. And a matter that grows on the way up stops looking like the same matter. The rung above then cannot work on it at all.

What does it actually mean to escalate, and why does so little change?

To escalateTo move one matter up one level, carrying with it what it already has, rather than starting a fresh matter somewhere else. is to move one matter up one level. Escalating is not starting again somewhere better, and it is not opening a second front. The level above is meant to receive the matter as decided below, recognisably so. Only then can the level above do its one job: look at what happened when the entity was asked directly, and form a view about it.

Consider a complaint at a school. A daughter's marks on one paper look wrong, so a parent speaks to the class teacher, and the class teacher explains the marking scheme and declines to change anything. Going next to the head teacher, what the parent carries is the paper, the marking scheme, what was asked of the class teacher, and what the class teacher said. The parent does not arrive at the head teacher's office with the marks, plus a complaint about the bus timings, plus something about last year's sports day. Not because those things are illegitimate. Because the head teacher's first job is to work out what the class teacher decided and why, and a bundle containing four different grievances makes that job impossible before it makes it easy.

A financial escalation is that same act with the stakes raised and the informality removed, and the discipline that makes it work is subtraction rather than addition. Each rungOne level of the structure. Every level above the first assumes the level below has already been asked and has already answered. above the first is built on the assumption that the rung below was asked and answered. The assumption is load bearing. The moment the matter arriving above cannot be matched to the matter decided below, the assumption fails, and the first thing the rung above must do is send back a request for clarification.

Step one: is the rung below genuinely finished?

ReadinessThe condition of the level below being genuinely finished, evidenced by something dated, which is what step one establishes. is step one, and it is the step people skip. Readiness feels like a formality and is not one. A matter is ready to move up when the level below has been asked and has answered, and when there is something dated to show both halves of that. Asked, and answered. Not asked. Not asked and still waiting, unless waiting has itself become the answer in a way the route above recognises, a threshold each forum sets for itself.

Why does this carry so much weight? Because the level above is not a fresh pair of eyes on the household's problem. The level above begins its work with what happened below. Strip that away and there is nothing for it to begin on. Nobody has looked at the merits, so a matter that arrives without evidence of a finished rung below does not get refused on its merits. The matter gets returned, and the household reads a routing message as a defeat.

The Bhosale household passes this test on 11 July for two reasons, and both are pieces of paper. The letter of 20 June asked, in writing, for Rs 24,000/-. The reply of 4 July answered, declining, and gave a reason. Two dates, two documents, one asking and one answering. The pair of dated documents is what readiness looks like, and a household holding that pair holds the only thing step one requires.

Now the honest part. Plenty of households do not hold that pair, and not because they were careless. Some asked at a counter and nobody wrote anything down, some telephoned and got an answer that left no trace, some wrote and heard nothing. Dealing with large organisations while also working has exactly that texture, and the practical response is not self reproach but the same step done once more in a form that leaves a record. If the rung below has not answered in writing, ask it again in writing. Asking again in writing is not a delay in the process. Asking again is the process.

Step two: what belongs in the record, and what does not?

The recordEverything submitted and everything received, each with the date it carries, assembled in the order it happened. is everything that was submitted and everything that came back, each with the date it carries, laid out in the order it happened. The record is not a case and not an argument. A chronology with documents attached is the whole of it, and its only job is to let somebody who has never seen this matter reconstruct it in five minutes without asking a single question.

The test for whether a thing belongs in the record is mechanical and worth memorising: does it have a date, and did the rung below see it? If both answers are yes, it goes in. If either answer is no, it stays out. The two part test does more work than any amount of judgement about what is important, and it keeps working while the household is angry. Anger is exactly when judgement is least available.

The exclusions are the point, so notice what the test excludes. It excludes the household's view of the reply, which has no date and was never put below. The test excludes the other things that went wrong with the same entity, never put below either. An undated thing cannot sit in a chronology, and a chronology is the entire deliverable, so the test excludes anything undated however true. None of that says the excluded thing is untrue or unimportant. The exclusion says only what this particular parcel is for.

Here is what the Bhosale household assembles on 11 July. The claim documents as filed. The schedule of the floater policy, held by the household from the start. The settlement statement showing Rs 91,440/- paid against a bill of Rs 1,42,000/-. The letter of 20 June asking for Rs 24,000/-. The reply of 4 July declining and giving its reason. Five items, four of them dated on their face, all five already seen by the rung below. The parcel is thin, and thinness is a feature.

Step three: what is the request, and why does it not move?

The request is the sum asked for and the reason it is asked for, in one sentence, and at step three it is copied rather than written. An unchanged requestThe same sum asked for the same reason as at the level below, which is what lets the level above match the two matters to each other. lets the rung above lay its own matter beside the matter the rung below decided. The two are visibly one matter. Change the sum and that placement becomes an exercise in inference.

On 20 June the Bhosale household asked for Rs 24,000/-. On 11 July it asks for Rs 24,000/-. Not Rs 50,560/-, the whole sum the household paid. Not some figure in between chosen because three weeks have passed and a letter has come back declining. The same number, for the same stated reason, in the same words if possible.

Holding the number still is genuinely hard, so be precise about why it does not grow. The reason is not that the household decided the other Rs 26,560/- was fair, and it is not politeness. Of the Rs 50,560/- paid, three components totalling Rs 26,560/- follow from terms printed in the schedule the household holds: non payable items at Rs 8,400/-, room charge above the limit at Rs 8,000/-, and co-pay at Rs 10,160/-. The fourth, the proportionate deduction of Rs 24,000/-, turns on a mechanism the schedule does not describe. The undescribed mechanism is why the request was Rs 24,000/- on 20 June, and nothing about the schedule changed before 11 July.

The reason that fixed the request at Rs 24,000/- on 20 June still holds on 11 July, and the level of annoyance in between is not one of the inputs. That single component is 47.5 per cent of what the household paid, which is a substantial thing to be asking about and needs no inflation to be worth asking. A request that grows between rungs invites a question the household cannot answer well: which of these two numbers was the real one, and why should the rung above believe the second rather than the first?

Step four: how is what the rung below said stated?

Step four is one paragraph and it is the shortest paragraph in the whole submission. The paragraph states what the rung below said. Not what it meant, not what it was really doing, not how it came across. The words it used, in its own terms, close enough to the original that anybody holding the reply can lay the two side by side and see they match.

CharacterisingDescribing what a reply was like rather than stating what it said. Characterising adds no new information and gives the reader something to disagree with first. is the alternative, and it is where a great many otherwise sound escalations lose their footing in the first thirty seconds of being read. Characterising is writing that the reply was evasive, or dismissive, or a standard form letter that nobody read properly. Every one of those may be an honest impression. The reply is in the parcel and the reader can form an impression from it, so none of them is information the reader lacks.

Stating is checkable and costs nothing; characterising is arguable and costs the first thing the reader does with the submission. A characterisation in the opening paragraph makes the reader's first mental act a decision about whether to agree with the household concerning the tone of a letter the reader is about to read anyway. If the reader does not agree, the submission has opened by being wrong about something, and it has done so before reaching the part where it is right. Opening by being wrong is an expensive trade for a word.

The Bhosale household writes that the reply of 4 July stated that the terms in the schedule had been applied. That is all. The line is accurate, it is one line, it is checkable against a document in the parcel, and it moves the reader straight to the question the household wants asked: does the schedule describe the mechanism that produced Rs 24,000/-? The household's private view of the reply is not in the submission. The private view would not help, so it stays out, and that reason is better than politeness.

Step five: what exactly remains unresolved?

Step five is one sentence, and getting it to one sentence is most of the work of the whole submission. The one sentence carries what remains unresolvedThe single thing still open after the level below has answered, stated in one sentence and nothing more. , the single thing still open after the rung below has answered. Not the list of things the household is unhappy about. The one thing that the reply did not settle.

For the Bhosale household that sentence is: the mechanism producing the Rs 24,000/- deduction is not described in the schedule that the reply relies on. Read it again and notice how much it is doing. The sentence accepts that the schedule exists. The sentence accepts that the reply relied on the schedule. Nobody is accused of anything. One specific gap between the document and the arithmetic is identified, and that gap becomes the only question in the matter.

Why one sentence and not five? Because the sentence is what the rung above will end up deciding, and if the household does not write it, somebody else will write it for them from a longer document, and the version somebody else writes will not be the version the household would have chosen. Writing it is the household's only control over what question actually gets answered. The one sentence at step five fixes the subject of the matter, and after that sentence the decision belongs to somebody else.

A practical way to find that sentence: write the paragraph that wants writing, then ask of every clause whether the reply already dealt with it. A clause the reply dealt with is resolved, welcome or not, so out it comes. The clauses that survive are what remains. If nothing survives, the matter is finished, and a finished matter is a legitimate outcome rather than a failure. If four things survive, the matter probably contains four, and the honest response is to notice that rather than bundle them.

Step six: how is the matter submitted, and what is kept?

Step six is the least interesting step and the one that most often turns out to matter. The matter is submitted by whatever route the forum in question accepts, and the fact of the submission is then recorded: what went, when it went, by what route, and whatever acknowledgement or reference came back. The record of the sending is the one genuinely new document the entire escalation creates. Somebody may later ask when this was submitted, and the only satisfactory answer is a document rather than a memory.

The routes differ from each other and change, so every forum's own route and required form are read at that forum. One habit holds across all of them: whatever route was used, the evidence that it was used gets captured. A screenshot of the acknowledgement screen. An acknowledgement message. A postal receipt. A dated copy of what was sent. Each takes a minute at the time and is unreconstructible afterwards.

The copy kept is a copy of exactly what was sent, not of what was meant to be sent. That distinction sounds pedantic and is not, because submissions get edited at the last minute, attachments get dropped, and forms truncate long fields without saying so. The household's copy should match what arrived at the forum. If a form limited what could be typed, that fact is worth noting, along with what had to be left out. The note costs nothing and answers a question that is otherwise unanswerable months later.

Step seven: where does it stop, and when is that decided?

Step seven is decided before step six and it never appears in the submission at all. Step seven is the household settling, in advance, what it will do if the answer goes against it. Settling that is the whole step. Deciding in advanceSettling what will be done if the answer goes against the household, at a moment of calm rather than in the week the answer arrives. is not a formality and it is not pessimism. Deciding in advance is recognising that somebody has to make the decision at some point, that there are two available moments to make it in, and that the two are not remotely equal.

The first moment is now, before anything is sent. Nothing has been spent yet on this rung. No answer has arrived to be disappointed by. The household can look clearly at its own position: how many working days it can actually give to this, what those days cost in a tailoring counter that does not run itself, what else those days were needed for. Working days, earnings and competing demands are facts about the household, and all of them are knowable today.

The second moment is the week the answer arrives, if it goes against the household. By then there is sunk time behind the decision, there is a fresh disappointment sitting on top of it, and the household's own position has become very hard to see because it is being looked at through the answer. People in that week either abandon a matter they had good reason to continue or continue a matter that was costing more than they could afford, and both errors come from the same cause: the decision was made at the worst available moment.

The Bhosale household decides on 11 July, before sending, that this is the last rung it takes. The reason is arithmetic about itself rather than about the matter: the next route would cost days that a tailoring counter and a payslip cannot both spare in the same month. Stopping here is a decision about the household's own position and not a judgement that the matter lacks merit, and the two are genuinely separate questions that only look like one question in the week a disappointing letter arrives.

What must not change on the way up, and why does it matter?

Four things travel unchanged: the request, the reason, the documents and the shape of the matter. Two things are added: one sentence saying what remains unresolved, and a record that the submission happened. Four carried and two added is the complete list, and every other tempting addition makes the parcel heavier without making it better.

Carrying the matter unchanged is not merely tidy. The reason turns on what the rung above actually does first. The rung above is not starting from the household's grievance. The rung above starts from a decision that was already made, and its work begins with reconstructing what that decision was about. If the matter it holds and the matter decided below are the same shape, the reconstruction takes one reading. If they are different shapes, the reconstruction is impossible from the parcel alone, and the only thing the rung above can do is ask the household what the rung below actually considered.

The same thing is visible from the other side. A neighbour asks for help with a dispute with a shop and hands over the bill, what the neighbour wrote, and what the shop wrote back. The three fit together and a view forms in five minutes. Add to the same three four grievances the shop was never told about, and no view can form at all, not because the four are false, but because it is no longer clear which parts the shop has ever had a chance to answer.

The shape of the matter is the thing being carried, and the documents are only how it is carried. A household that understands this can make its own judgement calls without a rule for every case: the question is always whether the thing being added was in front of the rung below, and if it was not, adding it now converts an escalation into a new matter that happens to be arriving at a level built for escalations.

How does a request stay narrow when the matter has become personal?

By this point the matter is usually personal, and pretending otherwise is not useful. Three weeks of documents, a bill that was paid out of savings that were needed elsewhere, and then a letter that says no. The household is not being irrational when it starts to feel that the whole thing was unfair. Feeling that is being a household.

The protection against that is structural rather than emotional, and it is very simple: the request was written down before the matter became personal, and it is not reopened. Three of the four components follow from printed terms and one does not, so on 20 June the household worked out that Rs 24,000/- was the part it could evidence. The analysis was done calmly, on the documents, before any reply arrived. The June version is better than any analysis the household could do on 11 July, and the correct response to being annoyed is to trust the earlier version rather than to improve on it.

Feelings about a matter are legitimate and they are not evidence, and the request is exactly where the difference between the two becomes visible. The way to keep a request narrow is not to feel less. The way is to have decided the number at a moment when feeling less was possible, and then to refuse to reopen it. Anger is information about how much this cost the household. Anger is not information about which component of a settlement the schedule failed to describe, and the request is made only of the second kind.

There is a smaller habit that helps here too. Write the escalation, then leave it a day, then read it looking only for adjectives. Adjectives are where annoyance hides. In a well made escalation there are almost none, and the sentences that survive are dates, sums, what was asked, what was said, and what remains open. The surviving document is not colder than the angry one. The difference is that it gets read all the way through.

What happens at the rung above before anything is considered?

Whatever the forum, the first work done on an escalated matter is not evaluation. The first work is establishment: working out what was asked below, what was answered below, what is being asked now, and whether those three things line up. Only after that does anybody form a view about who is right, and a great deal of the time delay that households experience as indifference is actually this establishment stage taking longer than it needed to.

The parcel arrives. Somebody opens it and looks for four things. Which sum did this household ask the entity for? Which answer did the entity give? Which sum is this household asking the forum for? Is the second thing an answer to the first, and is the third thing the same as the first? When the parcel contains a dated letter, a dated reply and a request that matches the letter, those four questions are answered in one reading and the matter moves to the part that is actually about its merits.

When the parcel contains a request the entity was never asked for, or four grievances of which one was raised below, those questions cannot be answered from the parcel. The rung above cannot evaluate what it cannot first establish, so an unestablishable matter produces a request for clarification rather than a decision, and every day of that is a day nobody spent on the substance. The household reads the delay as being ignored. The delay is the predictable consequence of the parcel's shape.

The same reasoning answers whether a longer submission works better. Length is not the variable; whether the extra length establishes or complicates is. Length spent making the chronology unambiguous is well spent, and length spent adding matters nobody below considered buys delay at full price.

When does a household stop, and why is that a decision about itself?

Every route beyond the current one costs something, and what it costs is mostly not money. The cost is working days, and attention, and the specific kind of attention that is hard to give while also holding a job and a counter and a household together. Such costs land differently on different households, and where to stop therefore cannot be settled from outside.

Two questions get tangled together, and separating them is what helps. The first question is whether the matter has merit, and that question belongs to the forum. The second question is what this household can afford to spend on it, and that question belongs to the household and to nobody else. A household that decides in advance to take one rung and stop is answering the second question, and answering it correctly does not require any view at all about the first.

The Bhosale household answers it like this. Meghna Bhosale's leave is finite and is already committed. Ashok Bhosale's counter earns nothing on a day it is shut. The buffer the household holds covers well under a month of ordinary outgoings, so days spent are not neutral. Given that, the household decides that it will take this rung, put its best version of the matter into it, and accept whatever comes back, whatever it is. The household writes that down on 11 July and does not revisit it.

Deciding in advance is competence, and it is worth naming as such. Stopping feels like losing, so continuing indefinitely is the easy act. Deciding the limit of one's own capacity in advance, while calm, on facts about oneself that are actually known, is the harder and more disciplined one. Kahneman and Tversky described the tendency to treat a settled outcome as closed and rarely reopen it, and the same machinery works in the other direction: a matter still open exerts a pull that has nothing to do with whether continuing is a good use of the household's remaining days. Deciding in advance is how a household keeps that pull from doing the deciding.

And it must be said plainly: the escalation may produce nothing. A route being available says nothing about what any forum will decide. A complaint that does not succeed was not therefore a mistake to make. The household asked a specific, evidenced question about a mechanism the document did not describe, and asking it was reasonable whatever the answer turns out to be.

How does this household's escalation of 11 July actually read?

Here are the seven steps run end to end on the Bhosale household's own matter, with nothing added and nothing softened. No two matters produce the same sums or the same dates, so what follows is one instance of the procedure rather than a model to copy.

Look at what is not in that list. There is no argument about whether the deductions were fair in general. There is no mention of the Rs 26,560/- the household paid under terms the schedule does print, more than half of what it paid and not contested at all. There is nothing about how any of this felt. The 11 July matter is the 4 July matter, one sentence longer, and the whole procedure exists to protect that identity.

What goes wrong when a matter grows on the way up?

Growing the matterAdding grievances at the escalation that were never put to the level below, which turns one traceable matter into several untraceable ones. is the characteristic failure of the whole procedure, and why it happens deserves sympathy. Every rung costs more than the one below it. The household has already spent three weeks. More is about to be spent. The feeling that the second attempt should carry more than the first is not stupidity, it is a reasonable sounding piece of arithmetic about effort, and it is exactly backwards.

Which routes exist in India, and what can only be read at the forum?

The seven steps are craft rather than law, and they hold wherever a matter has to move from one level to the next. A jurisdiction fixes which routes exist, what each requires before it will look at a matter, and what periods apply. Requirements and periods are settled at each forum's own source, and the reason is not caution for its own sake.

A wrong period is worse than no period. A household that reads a stale number, believes it, and acts on it has been given something that looks like certainty and is not, and the cost of that lands entirely on the household. A household that is told to check at the source has been given one extra task and no false confidence. The second is the better trade, every time.