The desk · reader data points

Two things we can't compute without you.

When a transfer actually lands, and what a bank offers to keep you. No issuer publishes either. We publish what readers report only in aggregate, only above a floor, always labelled as a claim — never as a source.

Both timestamps, to the minute, with the zone you were in. We store the stamps and derive the hours — a stored duration is a second truth that can disagree with the clock it came from.

The month, not the date — nobody remembers which Tuesday they rang a bank, and a false precision is invented data. What they offered is a fixed list, on purpose: this page publishes numbers and categories, never a stranger's prose.

Filing needs an account — not to gate anything, but because every published count prints its contributors beside its reports, and that sentence needs distinguishable people. Reading needs nothing: everything below is public, signed out, with no cookie. Delete your account and your reports go with it.

What readers have reported

Loading what's on file…

reader-reported Readers told us this. Nobody has verified it, and it is not a source. It is not a source, it never moves a ₹ figure on this site, and it never enters our data files.

The rules — what we will and will not print
  • Below 3 reports, nothing is summarised. The same floor the desk's own stopwatch uses (3), and for the same arithmetic: the summary is a median, and three is the smallest n at which a median means anything. Below it the group does not leave the database at all — "we hold two reports on this card" is the disclosure a floor exists to prevent.
  • And below 2 distinct contributors, nothing either. Three reports from one account is one person's experience wearing a statistic's clothes. It is also what one motivated stranger can produce in a minute.
  • Reports and contributors are always both printed. "7 reports · 4 contributors" — at this scale the realistic failure is one enthusiast filing ten, and that sentence makes it visible instead of impressive.
  • A reader report never becomes a desk observation. Our own transfer-time table is built from transfers we ran and timed ourselves; these sit beside it, in their own tier, and no arithmetic mixes them. Nothing here is shrunk toward an issuer's published window either — that estimator prices the bank's word against our stopwatch, and a reader claim is neither.
  • Everything published is a number or a fixed category. The free-text box on both forms is read by the desk and by nobody else. That is what lets this run without a moderator: a closed vocabulary cannot carry a link, a phone number or an insult, and an integer cannot either.
  • What we refuse to collect at all: approval odds. They are PII-adjacent, unverifiable in principle, and the one typed-form data class that demonstrably kills the sites that try it. There is also no fuzzing anywhere here — both floors above are the real numbers, published, not tuning constants we withhold.
The wanted list — 12 transfers worth the most

Where a stopwatch buys the most: our answer today is weak — a community grid, a ceiling wide enough to be useless, or nothing at all — and it is a route a lot of readers hold. The desk is running these itself; if you get there first, file it.

  1. axis-edge-miles → Singapore Airlines KrisFlyer The richest route Atlas has and, per Axis's own per-partner table, among the slowest — "up to 10 working days", which is a fortnight in wall-clock time. Nothing tells a reader whether that ceiling is the real number or a lawyer's. The single most valuable stopwatch on the board.
  2. hdfc-reward-points → Singapore Airlines KrisFlyer HDFC publishes no transfer window anywhere a reader can reach — every one of the 22 HDFC rows rests on a community grid. This is the most-held premium card in the country transferring into the most-wanted programme, and our answer is somebody else's table.
  3. hdfc-reward-points → Air Canada Aeroplan Same blackout, and the community figure here is "24 hours" — the exact shape of claim that is either true and worth saying loudly or false and costing readers a booking.
  4. sbi-travel-credits → Air Canada Aeroplan SBI's T&C §2(m) says "up to 21 days" and that clause covers all 26 SBI routes identically. A ceiling that wide is honest and nearly useless; one real transfer converts it into a number.
  5. hsbc-reward-points → Singapore Airlines KrisFlyer HSBC is the only Indian issuer whose own page says "instantly or within 1 working day". If that holds, it is the fastest route in the country and should be famous; if it does not, we are repeating a marketing sentence.
  6. amex-membership-rewards → Singapore Airlines KrisFlyer The only `mixed` block on the board — Amex's FAQ publishes a 3–5-working-day SLA while every community report says 1–3 days. Both are in the string because we cannot choose between them. A stopwatch chooses.
  7. amex-membership-rewards → Marriott Bonvoy The 1:1 hotel route, and the one Amex row whose community figure ("24–48 hours") is materially faster than the airline rows. Worth confirming that the hotel side really is the quick one.
  8. icici-reward-points → Air India Maharaja Club Air India's own page publishes 5 working days, but this is the only airline exit ICICI has and it is open to just two cards — a slow surprise here has nowhere else to go.
  9. kotak-solitaire-air-miles → Etihad Guest Kotak publishes no window on any route — checked across all three ratio PDFs and the card page. A single observation here would be the only number that exists anywhere for this issuer.
  10. rbl-nova-points → Singapore Airlines KrisFlyer Retracted to "not published" on 2026-08-29 when the desk's own extrapolation was caught. The honest blank is better than the invented seven days, and a stopwatch is the only thing that fills it.
  11. aurum-reward-points → The British Airways Club (Avios) AURUM's transfer terms live behind a login on an Angular portal that serves an empty shell to every fetcher. The document is unreadable; the transfer is not.
  12. marriott-bonvoy → Air Canada Aeroplan The busiest hop on the multi-hop router and the on-ramp for a dozen chains. "~5 days" is an aggregator's word, and every chain that runs through it inherits that uncertainty end to end.