This is the actual spec driving a bitsARK feature, published as-is and versioned in git: the problem, the evidence, the requirements with acceptance criteria, the metrics, and every decision along the way.
Brazilians moved more than R$ 500 billion in crypto in 2025 and most of it is stablecoins (~71% of declared annual volume, up to 90% of transactions in some months), yet spending that value in the real economy remains expensive and opaque: five channels compete with fee anatomies that cannot be compared side by side anywhere neutral.
An evidence-based research report ranking the five spending channels available to Brazilians, plus an interactive comparator that computes the real total cost of spending R$ X through each channel, with every assumption visible, editable and dated.
bitsARK covers the full crypto lifecycle for Brazil (acquire, hold, receive, and now spend), captures the underserved "gastar cripto" search cluster, and gives the existing personas pages a natural next step.
Brazil is the fifth-largest crypto market in the world 1 and its on-chain economy is overwhelmingly denominated in stablecoins: about 71% of declared annual volume, rising to as much as 90% of reported transactions in some months 23. bitsARK's own persona research found that the largest user group by headcount is retail savers holding digital dollars as a store of value 5 - value that eventually needs to buy groceries, pay bills, or fund a purchase in BRL.
The moment a holder tries to spend, the market fragments into five channels with completely different cost anatomies: selling on an exchange and withdrawing via Pix, buying gift cards with crypto, loading a crypto debit card, paying merchants directly, and P2P. Each hides its true cost in a different place: trading fees, spreads versus the reference rate, fixed withdrawal fees, card FX spreads, IOF, or gift-card premiums.
Independent, Brazil-specific comparisons do not exist. Search results for spending-crypto queries are dominated by affiliate content that recommends whatever pays the highest commission, and adoption research keeps flagging the same paradox: stablecoin ownership is near-universal among Brazilian crypto users, but everyday spending lags far behind 4.
bitsARK already answers how to choose an exchange, how stablecoin flows behave, and how to receive money from abroad. Spending was the missing quadrant, and the one I found hardest to scope: unlike an exchange or a stablecoin flow, "spending" isn't one product with one fee schedule. It's five different mental models that happen to share a word.
Findings that motivated this feature, in decreasing order of weight. Full citations in the sources section.
Market size: Brazil ranks 5th in the 2025 Chainalysis Global Crypto Adoption Index 1, with US$ 318.8 billion received on-chain - nearly a third of all Latin American activity 16. Receita Federal data shows R$ 505.5 billion moved in declared operations in 2025 (up 21.5% over 2024), averaging ~4.7 million individual taxpayers reporting crypto operations per quarter 2.
Stablecoin dominance: stablecoins are ~71% of Brazil's declared annual crypto volume (USDT alone ~65%), reaching up to 90% of reported transactions in some months 23. Digital dollars are a means of holding value, and holding always ends somewhere - which makes the spend/off-ramp moment universal and recurring across the entire holder base, on a long enough timeline.
The spending gap: adoption studies place stablecoin ownership at 91.8% of Brazilian crypto users while everyday spending remains marginal 4 - the friction sits downstream in the spending layer, past the adoption question entirely.
Internal persona research: of the seven personas mapped in bitsARK's stablecoin study, at least four (global professionals paid from abroad, digital-dollar savers, remittance recipients, and digital-economy consumers) face a recurring convert-to-spend decision 5.
Channel supply is expanding but unmapped: Binance and Mastercard relaunched a crypto card for Brazil in October 2025 6, gift-card platforms operate Brazil-specific catalogs, and P2P remains a R$ trillions-capacity market concentrated on three platforms 10. No neutral source compares their costs.
Regulatory stakes are rising: spending crypto is a taxable disposal that counts toward the R$ 35,000 monthly exemption 7, IOF on international card spending was unified at 3.5% 8, and DeCripto monthly reporting starts in July 2026 9. Cost-aware spending now has compliance consequences.
Persona profiles come from the published bitsARK stablecoin research. This feature serves the four with recurring spend decisions:
Paid in USDT/USDC from abroad; convert part of every invoice into BRL living costs. The receive side is already covered by the PJ guide - this covers the spend side.
View persona profileThe largest persona by headcount. Hold stablecoins as savings; spend episodically (an emergency, a large purchase, monthly top-ups).
View persona profileReceive family transfers in crypto; nearly 100% of the received value must become spendable BRL quickly and cheaply.
View persona profileGaming, betting and digital services users who already pay digital merchants; most likely to use gift cards and direct crypto payments.
View persona profileEach story is written from a persona above and maps to at least one functional requirement.
Acceptance criteria in given/when/then form. A requirement ships only when its criteria pass on both the EN and PT pages.
Lighthouse mobile score of 95 or higher, matching the site-wide CI budget. The comparator ships as an inline vanilla-JS island; no framework payload, no chart library.
All interactive controls keyboard-operable; outputs announced via aria-live; every data table carries a caption; heading outline valid with no skipped levels.
No analytics, no cookies, no PII. Nothing the user types leaves the page - all computation is client-side.
Static-only, consistent with the site: data baked at build time from committed JSON and the existing PTAX pipeline. No new runtime endpoints.
No affiliate links anywhere in the feature. Gift-card platforms, card issuers and exchanges all run referral programs, and slotting one in would have been easy - a single parameter on the exchange dropdown. It stays out because a recommendation with a commission behind it is a different product than the one this page claims to be.
Targets reviewed 90 days after ship. This site runs no analytics by design, so metrics lean on search-side signals.
| Metric | Target (90 days) | Measured via |
|---|---|---|
| Search impressions, "spend crypto Brazil" query cluster (PT + EN) | Page enters the cluster's top-10 impressions for the site's country profile | Google Search Console, query and page filters |
| Indexation health | All four new URLs indexed with valid hreflang cluster, zero Search Console errors | GSC page inspection + sitemap report |
| Internal discovery | Research page linked from at least 4 existing pages (personas, PJ guide, exchanges, home) | Build-time link audit (check-links) |
| Qualitative signal | At least 3 substantive feedback submissions or corrections on comparator assumptions | Existing on-site feedback form |
Cuts keep v1 shippable by one person while covering the complete decision. Each cut is a candidate for a future iteration, promoted only if the metrics above justify it.
Catalog prices change daily and scraping breaks the moment a vendor tweaks their HTML. v1 ships dated, sourced premium ranges instead - a number that is a few weeks stale beats a "live" number that silently returns zeros for a week before anyone notices.
Each integration is a maintenance liability on a solo-maintained site. The editable-parameters model gives users the same power without the fragility.
Requires state serialization and testing surface disproportionate to v1 value. Revisit if feedback asks for it.
The site's moat is Brazil-specific regulatory and fee depth. Generalizing would dilute exactly what makes the comparison trustworthy.
Spending crypto is a taxable disposal and the page must say so, but computing capital gains requires cost-basis data this site should never collect. Link to guidance instead.
Dated, append-only. New decisions are added on top; superseded ones are struck through, never deleted.
Ship the research as a new top-level section using the Portuguese slug for both locales.
Spending covers BTC, gift cards and cards - broader than stablecoins. The site already shares PT slugs across locales (stablecoins-brasil, receber-exterior-pj), and the PT slug targets the query language of the underserved cluster.
Fee defaults ship in a versioned data file, each carrying a last-verified date and source links, rendered in the UI.
A dated, sourced, slightly stale number the user can inspect and edit beats a scraped number that silently breaks. I looked at a nightly scraper early on; the exchanges' own fee pages disagree with their APIs often enough that automating it would have shipped wrong numbers with a confident timestamp attached - worse than an honest 'last checked two weeks ago.'
No framework, no runtime dependencies; inline script with data passed via attributes, following the proven pattern from the PJ cost simulator.
Keeps the Lighthouse >= 95 budget intact and the maintenance surface near zero. The interaction model (inputs, recompute, re-rank) does not need a framework.
The feature will never carry affiliate parameters or sponsored placements.
Search results for this entire query cluster are already dominated by affiliate content recommending whatever pays the highest commission - that is the problem statement, section one. Adding this page to that pile, even quietly, would make its ranking exactly as untrustworthy as the ones it is trying to replace.
This document ships in its own commit, before any feature code, and is updated by the commits that build the feature.
Makes the process auditable end-to-end, and it forces the spec to be complete enough to actually build from - a vague section is obvious once it is public, instead of sitting quietly in a draft only I ever read.
| Version | Date | Status | Notes |
|---|---|---|---|
| v1.0 | 2026-07-08 | Shipped | Launched. Navigation and site-wide footer links, homepage product card, OG images (EN + PT), llms.txt entry and internal cross-links from related pages are live in both locales. All requirements R1-R8 met and verified; the feature is discoverable and indexed. Status moved from Draft to Shipped. |
| v0.9 | 2026-07-07 | Draft | Interactive comparator shipped: editable per-channel fees, amount presets, exchange picker fed from the live dataset, live re-ranking summary, and per-parameter verification dates and sources (R2, R3, R4, R5, R6, R7 met). Parameters live in src/data/spend-channels.json; defaults also render server-side for the no-JS baseline (R1). Covered by 11 Playwright tests. Pending: launch wiring (nav, homepage, OG, cross-links) before v1.0. |
| v0.2 | 2026-07-07 | Draft | Discovery complete, research shipped: the five-channel ranking, executive summary, channel profiles, tax layer, methodology and full citations are live at /gastar-cripto/. The comparator is present as a static ranked baseline; the interactive version (R2, R3, R5, R7) is still pending. |
| v0.1 | 2026-07-07 | Draft | Initial public draft: problem statement, discovery evidence, personas, user stories, requirements R1-R8, NFRs, metrics, scope cuts and founding decisions. Published before any feature code. |
The comparator applies the same skeleton to every channel: percentage components apply to the amount being spent; fixed components are flat BRL values. Channel parameters and their sources ship in the public data file (src/data/spend-channels.json in the site repository).
X = amount to spend, in BRL
netBrl = max(0, X - sum(pctComponents) * X / 100 - sum(fixedComponentsBrl))
totalCostBrl = X - netBrl
totalCostPct = totalCostBrl / X * 100
Channel components:
exchange-pix = tradingFeePct + spreadVsPtaxPct + pixWithdrawFeePct [+ pixWithdrawFeeBrl fixed]
gift-card = premiumPct (may be negative) + fxSpreadPct [+ networkFeeBrl fixed]
crypto-debit-card = topUpFeePct + fxSpreadPct + iofPct
merchant-direct = conversionSpreadPct + processorFeePct
p2p = priceDiscountPct + platformFeePctFixed fees dominate at small amounts, percentage fees dominate at large ones, so the cheapest channel flips somewhere in between X = 0 and X = 10,000. Exactly where it flips depends on which exchange and which card issuer you plug in, which is why these parameters are editable instead of baked into a static table.
Citations referenced with [n] throughout the document.
Sources last reviewed:
Related
The persona study below is discovery evidence item [5] - the published starting point for everything in this document.