Where the association's money is, and what is leaking out right now

The association collected R$ 17,763,071 from 3,000 members over 38 months and grew for 34 of them without stumbling. Then two things broke at the same time. Sales fell 54% in the first half of this year against the same period in 2025, and in April the billing for 222 members stopped all at once, leaving R$ 288,271 overdue and R$ 97,239 a month out of the bank. In May the membership shrank for the first time. And there is no handful of large members to protect: the problem is the whole base.

Where you are

The membership grew hard and cleanly. Between January 2025 and March 2026 monthly revenue rose from R$ 454,201 to R$ 1,063,872, and the number of members paying rose from 1,123 to 2,572. The whole series climbs, in fact: from May 2023 to March 2026 that is 34 straight months up, without a single month down. Retention agrees: of the 2,601 members who have been around long enough to reach their sixth month, 93.7% got there; of the 1,758 who have been around long enough to reach their twelfth, 89.5% got there (counted only over people who had been members long enough to reach the milestone, otherwise anyone who joined this year would show up as a loss without having left). Hold one caveat about that number: it is a three-year average and it already hides a recent decline. Among the people who joined in the last quarter of 2025, only 84.3% reached a sixth payment. The third front of this report is about exactly that.

The living base today is 2,511 members, 2,289 up to date and 222 in arrears, carrying R$ 172,572,975 in protected vehicle value. Added up, the current fees of those 2,511 come to R$ 1,056,123 a month, of which R$ 958,884 comes from the ones who are up to date. That is all but identical to the R$ 959,051 that actually came in during June, which confirms the reading of the data matches the bank.

What is getting in the way

First: the small group you are probably picturing does not exist. The question behind this analysis was which few members account for most of the revenue. The honest answer is that no small group does. The 30 largest members (1% of the base) add up to R$ 1,618,439, only 9.1% of everything that came in. The largest 150 (5%) add up to R$ 4,595,961, or 25.9%. To gather 80% of the revenue you have to go all the way down to member number 1,307, nearly half the base.

That is not a problem, it is what a subscription business looks like: everybody pays about the same amount every month, so the revenue spreads out. But it changes the strategy. A dedicated service programme for thirty customers would protect 9.1% of the revenue and leave 90.9% uncovered. The money is in the whole base, and what threatens the whole base is the billing operation, not one large member walking away.

Second: sales halved, and it started before everything else. In October 2025, 160 new members joined, bringing R$ 66,857 in new monthly fees. In May 2026, 34 joined, bringing R$ 15,222. The fall is month on month, without a hiccup: 160, 153, 133, 112, 93, 66, 40, 34. Comparing half-year against half-year, the first half of 2025 brought 758 members and R$ 310,502 in new fees, the first half of 2026 brought 352 and R$ 138,444: 54% fewer people and 55% less new fee income.

That knocks out the arithmetic that was holding the growth up. While more than 100 came in every month and 27 left, the base rose on its own. Except the other side of that sum changed too: departures were 27 a month between October and December 2025 and became 34, 39, 48, 40 and 54 from January to May 2026. That is 215 departures in five months, R$ 75,947 a month in cancelled fees (the file carries no cancellation date, so a departure is counted in the month of the last payment of anyone recorded as cancelled; that same count is what the chart below shows). In March the net change was only 18 members, in April it was zero, and in May it went negative by 20 for the first time in the association's history.

Worth saying that June shows only 7 sign-ups and I checked whether the file was simply cut off half way: it is not. A member's first fee always falls in the month they join, in 2,994 cases out of 2,994, the last sign-up on record is 24 June, and the file runs to 18 July. June really is a closed month.

Third: billing for 222 members stopped all at once in April. March 2026 closed at R$ 1,063,872. April fell to R$ 967,495, a drop of 9.1%, and the months after it stayed at that level: R$ 971,393 in May, R$ 959,051 in June. Together, April, May and June delivered R$ 293,676 less than they would have at March's pace.

The group is easy to isolate: 217 of the 222 have their first unpaid instalment falling due in April 2026 and the other 5 in May. Of those 222, 214 had March as their last settled invoice, 2 settled the March invoice late, already inside April, and 6 have never paid a single fee since joining. That last group is a different conversation and is kept separate in the file. Not one of the 222 got as far as paying the April invoice. Today they each have three instalments outstanding (217 of the 222 have exactly three), adding up to R$ 288,271.

What stands out about these 222 is not that they stopped, it is how they stopped. They do not look like people who were on their way out: their average fee is R$ 438 against R$ 419 for those up to date, their time as members is practically the same (15.4 against 15.1 months), and they are spread across the five regions in nearly equal parts (48, 48, 44, 42 and 40). Half the story is right here: 100 of the 222 have never been a single day late in their whole time with the association, and the median of the group is 12 fees paid before the stop. A member who paid twelve months in a row without ever being late and then vanishes all at once, in the same month as 221 others, tends to be a billing failure — cards expiring in a batch, a payment file that never went out, a change of payment method — and not 222 individual decisions to cancel taken in the same week (high confidence that this is not normal churn, given the even spread and the clean history; low confidence about which specific failure it was, which only the billing system can answer).

If those 222 turn into cancellations, R$ 1,166,873 a year leaves the recurring revenue, on top of the R$ 288,271 already overdue.

The question

If there are no whales to pamper, where should the team's effort go over the next few weeks?

The answer: three fronts, in this order

If the money is not concentrated in a few names, it is concentrated in a few problems. The three fronts below attack the three holes in the order they pay back: the money that is already yours and sitting still, the accounts that carry almost a third of the fee income, and the sales engine that stalled. Worth knowing up front that the first two aim at lists which overlap. Adding the billing, top-fee and multi-vehicle files together would give 696 names and R$ 619,811 a month, but the unique names are 569 and the real fee income is R$ 426,240 a month, which is the number to use when someone asks how much is at stake.

1. Unblock the April billing — starts today

The first thing is to find out what happened in April before calling any member. Today, you or whoever runs the finance side needs to open the April 2026 billing file and compare it with March's: how many invoices went out, how many card charges were declined, whether there was a change of bank, payment processor or file format. That check takes a morning and determines everything that follows. If it is a technical failure, most of the R$ 288,271 comes back with a second attempt at collection and a notice message, with no negotiation needed at all.

In parallel, from tomorrow, the service team works the list cobranca_parada_abril.csv by phone, in the order the file already comes in. The order matters and it is not by amount: the file is grouped by the grupo_prioridade column, and first come the 100 who have never been a day late — R$ 108,894 outstanding and R$ 36,445 a month. If the reading that there was a billing failure is right, that group is exactly the one that comes back with a phone call and a fresh invoice, because there was never any intention to stop. Then come the 116 who already had a history of late payment (R$ 174,430 outstanding), which is a genuine negotiation conversation. Last are the 6 who have never paid a single fee since joining (R$ 4,946): those are not customer recovery, they are records that never became customers, and the right move is to check whether the sign-up was ever completed. With two agents making 25 calls a day each, the 222 are covered in five working days.

The script starts by asking, not by chasing: "I can see the April fee didn't come through and you'd never been late before — I want to check whether something went wrong on our side." Treating the first group of 100 as debtors is the fastest way to turn an operational failure into a real cancellation. Inside that group the order is alphabetical on purpose, because they all have exactly the same problem and the same urgency: 48 of the 100 are on the Leve plan, and sorting by amount outstanding would push most of them to the back of the queue precisely in the group where the association has the least reason to chase and the most reason to apologise.

Anyone who does not answer within three days goes into a second round by app message with a payment link for the three instalments, and anyone still not responding by the tenth day gets a call from the owner. That owner's call covers the 216 members in the first two groups, in the same queue order: the 100 who were never late first, the 116 with a history of lateness after. Within it, 15 members are also among the 126 highest fees in the base and on their own are worth R$ 25,041 a month and R$ 74,494 outstanding — 3 in the first group and 12 in the second. There is a sixteenth high-fee member among the 222, but he is one of the 6 who never paid anything: with him the conversation is about the record, not the billing.

On the Friday of the first week, the number that says whether it worked is how many of the 222 are paying again. Recovering 30% to 40% of the group is 67 to 89 members, and what that is worth depends on who is in the count. Following the recommended queue, the first 67 carry R$ 24,185 a month and R$ 72,235 outstanding, and the first 89 carry R$ 33,611 a month and R$ 100,394 outstanding — R$ 290,220 to R$ 403,332 over twelve months. If the recovery spreads evenly across the whole group instead of following the file's order, the same 30% to 40% is worth more, R$ 29,172 to R$ 38,896 a month, because the recommended queue deliberately starts with the smaller fees (medium confidence: the 30% to 40% range is a general benchmark for active collections, not your own track record, which the data sent does not record; if the cause really is technical, the actual recovery tends to run above that). If by Friday recovery is under 20%, stop calling and go back to the billing system: it means the problem is not being solved on the phone, and carrying on dialling only burns the patience of people who did nothing wrong.

2. Protect the 348 members with more than one vehicle — over the next two weeks

If there is anything resembling a critical group in this base, this is it. There are 348 members with two or more vehicles, 13.9% of the living base, who on their own account for R$ 302,414 a month, 28.6% of all current fee income. Their average fee is R$ 869 against R$ 348 for those with a single vehicle. And they are more exposed to April's problem: 40 of the 348 are in arrears, a rate of 11.5% against 8.4% for single-vehicle members.

Management takes this book from next week using contas_multiveiculo.csv, with one relationship call to each of the 348 over two weeks, roughly 35 calls a day split between two people. The call is not a sales call: it is to confirm the vehicles on record are still the right ones, that the contact details are current, and to ask whether any vehicle has joined or left the household or the fleet. Every unregistered vehicle that surfaces in that conversation is new fee income with no acquisition cost.

The target by the end of the second week is all 348 accounts contacted and at least 15 additional vehicles identified. It is worth insisting on full coverage rather than an 80% target: stopping at 280 accounts would cut exactly the 68 with the smallest fees, of which 66 are on the Leve plan and 2 on Essencial 60, and those are precisely the accounts nobody ever calls. If the team cannot manage all 348 in two weeks, stretch it to three rather than cutting the tail. Anyone who does not answer after two phone attempts gets a visit from the regional representative, since these accounts justify the trip by the monthly value they carry. Without that contact, the concrete risk is losing an account worth R$ 869 a month without having had a single conversation first, the equivalent of losing two and a half members at once.

If fewer than 200 accounts have actually been spoken to by the end of the two weeks, the bottleneck is missing phone numbers on the records and not the team's diary. In that case the priority becomes updating contact details, and the book goes back into next month's queue.

3. Find out why sales halved — starts in the third week

This is the most important front on a one-year horizon and the one the data sent can least explain on its own. The two spreadsheets show clearly that the fall exists and when it started, but they carry no advertising spend, no count of enquiries received, no proposals made, and no number of active representatives per month. In other words: it is possible to prove the engine lost power, it is not possible to say from inside the data whether it was budget cuts, a new competitor, or price.

What the data does rule out is a local explanation. Comparing March to May 2026 with the same months in 2025, every channel fell together: Google −70%, representatives −73%, referrals −64%, Meta −40%. So did all five regions, from −48% in Serra Gaúcha to −70% in Fronteira Oeste. Nothing collapsed on its own, and Meta actually gained share, from 20% to 32% of sign-ups in the period. That rules out "a representative left" and "one channel broke" and points at something hitting everything at once — total budget, price, competition — which is where the conversation with the sales side should start.

So the first task here is not a campaign, it is a diagnosis, and it is yours. In the third week, sit down with whoever runs sales and build a monthly table from January 2025 to May 2026 with four columns: how much was spent on advertising, how many enquiries arrived, how many proposals were made, and how many representatives were active. That is 17 rows and not seven on purpose: the deterioration started well before sign-ups fell, as the numbers just below show, and a table starting in November 2025 does not reach the origin. If each column can be broken down by channel and by region, break it down — that is what locates the fall fastest. Alongside it, put each month's sign-ups, which are in the last table of the appendix. The column that falls together with sign-ups is the answer, and that table takes two hours to build if the numbers exist anywhere. If they do not exist, that is the finding: the association is selling without any instrument to measure it, and creating that record becomes the task.

Meanwhile, one cheap thing can start straight away. Of the 489 members who cancelled, 139 left having paid three fees or fewer — R$ 56,038 a month that came in and went out before the association recovered the cost of bringing that person in. And the more recent intake is worse: of the 250 who joined in 2026 and have already had at least four months as members, 88.8% reached a third payment, against 97.0% of the 1,599 from 2025.

Except this did not start in 2026, and that is the part that changes where to look. By quarter of joining, the share reaching a third payment was 98.4% for people who joined in the last quarter of 2024, held at 97.6%, 98.1% and 97.5% through 2025, and dropped to 95.1% already in the last quarter of 2025, before the 88.8% of 2026. Cancellation within the first six months tells the same story more forcefully: 4.2% for the late-2024 intake, then 6.1%, 7.1% and 9.6% through 2025, and 15.2% for those who joined between October and December 2025. Those departures happened spread across each intake's early months, not bunched into April, so they are not an effect of the billing break. The quality of who was coming in had been getting worse three quarters before the volume fell, which is why the diagnostic table starts in January 2025 and not in November.

So whoever runs sales starts calling every new member on the seventh day after they join, confirming what the plan covers, what it does not, and the due date. That is about 13 sign-ups a week at the current pace, half an hour of work a day. Among those joining over the next three months, the share reaching a third payment should rise from 88.8% to 95%.

This is the cheapest action of the three fronts and it has measurable value. The 233 living members with less than six months carry R$ 95,463 a month, and at the early-loss rate of the last complete intake, 15.2% within six months, R$ 14.5 thousand a month leaves in that window. Bringing that rate back to the 4.2% of the late-2024 intake preserves around R$ 10.5 thousand a month, R$ 126 thousand over twelve months (medium confidence: the loss rate comes from the data, how much a welcome call recovers of it is my estimate). It is the same order of magnitude as the add-on finding left out further below, and it costs half an hour a day.

On the acquisition channel, the data suggests less than it first appears. Counting only people who have been members four months or more, Meta advertising loses 5.7% of members in the first three months (33 of 577) against 2.8% for representatives (10 of 359). But Meta is all but level with the other two channels — Google at 3.9% and referrals at 3.9% too — and the difference against them is far too small for the number of cases there are. Add to that the fact that representatives are the channel with the fewest cases of all, so the comparison most favourable to them is also the most fragile: 10 losses in 359 against 110 in 2,539 across the other three channels combined, and at that sample size a difference like that turns up by chance easily (one time in five, by the test that measures exactly this). It cannot be said that representatives bring people who stay longer, nor that Meta is the villain (low confidence, few cases and channels nearly level). Do not cut a channel's budget on the strength of this, and least of all now, with sign-ups already at a third of what they were.

The number that closes this front is the monthly net change in the base: joiners minus leavers. It was +133 in October, +18 in March, zero in April and −20 in May. The target for September is simply to be positive again. If after two weeks the diagnostic table has not been built because the numbers do not exist, do not wait: hire or assign someone to record enquiries and proposals from now on, because without that, next month arrives with the same unanswered question.

An extra finding that costs an afternoon of checking

Outside the three fronts, something turned up in the records that is worth checking and has nothing to do with billing. In seven of the eight plans, the value of the most expensive protected vehicle sits just under the number the plan is named after: Leve 25 goes up to R$ 24,982, Essencial 60 up to R$ 59,944, Plus 180 up to R$ 179,824. Everything points to that number being the coverage ceiling and to the system respecting it. The exception is Master 400: 18 members have a vehicle above R$ 400,000, reaching R$ 680,000, adding up to R$ 2,176,691 above the ceiling — and 17 of them have a single vehicle, so it is not a fleet total.

Whoever looks after the records should check those 18 cases (they are in veiculos_acima_do_teto.csv) this week, contract by contract: either there is special cover agreed that the data does not show, or there is a vehicle protected above what the plan covers. There is only one question to answer, and the contract itself answers it. If it is the second case, the risk shows up at the worst possible moment — when a total-loss claim is shared out, with the member believing they are covered for an amount the plan does not cover (medium confidence: the pattern across the other seven plans is strong and consistent, but reading the number in the name as the ceiling is an inference of mine from the values, not something written in the file — you can confirm it in a minute). If the check has not happened by Friday because the contracts are on paper or scattered, narrow the scope to the 10 cases above R$ 500,000, which on their own hold 82.6% of the excess, and leave the other 8 for next month.

How to read these numbers before acting

The files have no phone number and no email. The two spreadsheets sent carry name, tax ID, region and channel, but no contact column. Every file produced includes the tax ID precisely so you can match it against the contact records in your own system. Without that match, none of the actions above leave the page. It is the first practical step on Monday.

The mensalidade column in your records is the fee from when the person joined, not what they pay today. I found this by checking: each member's first payment matches that column in 2,994 cases out of 2,994, but the last payment only matches in 1,412 of those same 2,994. The other 1,582 have already been through a price rise and pay more. (Further on the text mentions 1,618 members having had a rise: that number is about the 3,000 on record and includes 36 members in arrears whose rise only shows on the unpaid invoice, which they never paid. The two pairs add up: 1,412 + 1,582 = 2,994 who have paid; 1,382 + 1,618 = 3,000 on record.) That is why every monthly figure in this report uses each member's most recent charge and not the column on record. The difference between the two methods is R$ 43 thousand a month across the living base, enough to change any calculation. If someone there pulls a fee report straight from the records, they will get a number lower than the real one; worth fixing at source.

"Who has paid the most" is not "who is worth the most today". Cumulative revenue rewards length of membership: someone who joined in 2023 has had 38 chances to pay, someone who joined in 2026 has had four. That is why the list by cumulative total (whales_top_5pct.csv) and the list by current fee (whales_por_mensalidade.csv) only agree on 78 of the 150 names. To decide who to serve better from here on, use the list by fee; the cumulative one is for recognising loyalty.

The data cut-off is 18 July 2026. Everything that refers to "today", "gone for more than 90 days" or "outstanding" is anchored to that date, not to the date you are reading. If weeks have passed, the 222 have more than three instalments open.

The annual price rise did not drive anyone away, but the data does not prove that on its own. Of the 1,618 members who went through a fee increase (from 5.5% to 16.5%, with 5.5% being the most common), 99 cancelled — 6.1%, well below the 28.2% of the 1,382 who never had one. The detail that blocks the easy conclusion: those who went through a rise have been members for 21.4 months on average, against 9.6 months for the others. Length of membership and surviving a price rise go together, so the number says that raising prices is safe for people who have already stayed, not that raising prices keeps people.

Sorting by amount in reais always leaves the same group for last, and the group has a name: the Leve plan. This is the bias that most affects the recommendations above, and it is not regional — the 222 are spread evenly across the five regions, so any regional quota would be decoration. The problem is by plan: of the 222 with billing stopped, 100 are on Leve, and if the queue went by amount outstanding they would sit on average at position 159 of 222, with half of them falling below position 171 and only 6 making the first 50. Except 48 of the 100 who have never been late are on Leve, and the aggregate arrears rate of the Leve family (9.2%) is higher than the Master family's (7.2%) — worth saying that comparing by family hides variation between plans, and at the finer cut Master 400 has the highest rate of all, 13.1% across 61 members. That is why the two fronts above were reordered: collections follow payment history rather than amount, and the multi-vehicle book covers all 348 accounts rather than stopping at the largest 280. If your team reorders by amount "to make it pay better", know that they are choosing to serve less of those who have less — and that the R$ 108,894 of the group that never fell behind is still money.

The file whales_at_risk.csv has nobody "at risk" in it — it has people already lost. It lists the 72 top-revenue members who have not paid for more than 90 days, which is a cut this analysis was asked for. Except none of the 72 is in an active state: 43 are the April arrears cases and 29 already show as cancelled. In this base the recorded status and the payment behaviour go together — there is not a single member marked as active who has quietly stopped paying. That is good news about the quality of your records, and it means the 43 are already covered by the first front and the 29 require a decision from you about attempting to win them back, which is a different kind of work.

Three things I looked at and decided not to turn into action. The first: selling an add-on (roadside assistance, glass, courtesy car, third-party cover) to those who do not have all of them. And the audience is not only the 468 members with no add-ons at all: the living base has 468 with zero, 1,069 with one, 758 with two, 199 with three and 17 with all four, which means 2,494 members still have a step ahead of them. Within the same plan each step is worth something similar: the median of the differences by plan is R$ 37 for the first add-on (from R$ 13 on Essencial 90 to R$ 41 on Essencial 60) and R$ 34 for the second. Those medians use only the five plans with at least 30 members on each side of the comparison — in the three smallest (Master 260, Master 400 and Plus 180, with 7 to 23 members holding no add-ons) the difference runs from −R$ 4 to +R$ 264, which is small-sample noise and not price. At a conversion of 15% to 20%, working only the 468 with no add-ons brings R$ 2,623 to R$ 3,497 a month; working all 2,494 brings, at R$ 35 a step, around R$ 13.1 thousand to R$ 17.5 thousand a month, R$ 157 thousand to R$ 210 thousand over the year (medium confidence: the price comes from the data, the conversion is my estimate, and the difference between those who have an add-on and those who do not is the list price, not proof that whoever is approached buys). Even in the larger scenario it is seven times smaller than April's problem and should not compete for attention now. The second: the 166 members on Master plans carry 26.0% of the association's total exposure and pay 20.4% of the fee income. That may well be correct — in a cost-sharing model an expensive vehicle usually does pay a smaller percentage — but whoever calculates your cost-sharing should check whether that proportion is the one designed or merely the one that happened (low confidence, it is a question to ask and not an error identified). The third: winning back those who have already left. There are 183 members who cancelled in the six months before the cut-off, R$ 63,750 a month in fees that used to belong to the association, with 10.5 months of membership on average; at 10% win-back that would be R$ 6.4 thousand a month (low confidence, there is no internal win-back history in the data sent). It was left out because it is a different kind of work — it needs an offer and an answer to why they left, which the data does not record — and because it would compete for the same team as the three fronts. Worth returning to once the April billing is resolved.

One alert worth setting up now: any month in which more than 50 members end up with an unpaid fee at the same time should trigger a check of the billing file the same day. April went three months without anyone pulling the handbrake, and that is what turned a one-month failure into R$ 288,271.

The next step

Monday morning, two things in parallel: someone in finance opens the April 2026 billing file and compares it with March's, and someone in records matches the tax IDs in cobranca_parada_abril.csv against the phone numbers in the system. On Tuesday the team starts calling. On Friday you look at one number: how many of the 222 are paying again.

Hold the second question for the following week, but do not lose sight of it: what changed between October and May to take sign-ups from 160 down to 34. The April billing is the fire, and putting out a fire is urgent. The fall in sales is the foundation, and it is what decides whether this association grows again.

Appendix

Files produced

Every file carries the same identifying columns: id, display_name, cpf, mensalidade_atual, mensalidade_adesao, monetary_total, frequency, last_seen, dias_sem_pagar, tenure_days, status, plano, veiculos, valor_fipe, extras, regiao, canal_aquisicao. The mensalidade_atual column holds the most recent charge and mensalidade_adesao the original amount on record; the difference between the two is the price rise already applied.

The lists overlap: 40 members appear in both the April billing list and the multi-vehicle accounts, 16 in the April billing list and the largest fees, and 80 in the largest fees and the multi-vehicle accounts — hence the 569 unique names and the R$ 426,240 a month already quoted in the body, instead of the raw sum of 696 names and R$ 619,811.

On names: the records hold 3,000 identifiers and 3,000 distinct tax IDs, but 2,948 distinct names — 52 names appear more than once. Since each has its own tax ID, they were treated as different people, with no merging of records. If your team knows that any of those cases is the same person with two records, that account's real fee is higher than it appears here.

What was read from the data

Cumulative revenue curve

Half the revenue comes from the first 483 members (16.1% of the base); 70% comes from the first 954 (31.8%); 80% comes from the first 1,307 (43.6%).

Current fee income of the living base, by band

Current means the most recent charge to each of the 2,511 living members, including the 222 who are not paying. What actually came in during June 2026 was R$ 959,051, and what comes from those up to date is R$ 958,884.

State of the base

The 222 with billing stopped, by priority group

Nominal plan ceiling and largest protected vehicle

Sign-ups and new fee income by month

The fall started in November 2025 and is continuous. The series runs from January 2025 on purpose: it is the window the diagnostic table in front 3 asks for, and it shows that up to October 2025 sign-ups only moved between 118 and 160 a month. June 2026 is a closed month: the last sign-up in the file is 24 June and the cut-off is 18 July.

In the first half of 2025, 758 members joined bringing R$ 310,502 in new fee income; in the first half of 2026, 352 joined bringing R$ 138,444.

* About this sample. This is a real report from the Distilo pipeline, run over synthetic data for a fictional company. That company is Brazilian: it is a vehicle-protection association, a cost-sharing arrangement common in Brazil where members contribute monthly and losses are shared out, so the amounts are in Brazilian reais and the plan names, regions and member names are Brazilian. It is published as it was produced rather than adapted, because the analysis is what we are showing you and the analysis does not change with the country. Your own report is written in your language, about your business, in your currency.