How SkyStatus Measures Flying Blue Member Data: Methodology
Every measured statistic on SkyStatus, from the headline figures in Flying Blue by the Numbers to the data boxes inside individual guides, is produced by one system with one set of rules. This page documents those rules: where the data comes from, who is in the sample, which privacy floors apply, how we guard against statistical noise, and how numbers stay current. If you quote our data, this is the page that tells you exactly what you are quoting.
The method in one paragraph
SkyStatus members import their official Flying Blue account statements. SkyStatus parses them into a structured ledger of flights, earn lines and redemptions, and reconciles every import against the statement's own totals. Statistics are aggregate database measurements over that ledger. A numeric claim requires at least 30 members. Groups of 10 to 29 get direction-only statements. Anything smaller is suppressed. Every published figure carries its sample size and measurement date, is re-measured weekly, and is stored append-only so past editions stay reproducible.
On this page
Where the data comes from
The primary source is the official Flying Blue activity statement, the transaction history Air France-KLM provides to every member. SkyStatus members download their statement and import it. A parser turns each line into structured data: flights with cabin, XP, UXP and miles, ground earn lines with their channel, redemptions with miles spent. Every import is reconciled against the statement's own balance arithmetic before it is accepted. An import that does not add up is reviewed by hand, not silently accepted.
Members can also log data the statement does not carry, such as ticket prices and the cash value of award bookings. Price-based statistics (cost per XP, realized value per mile) come from those member-logged fields, and the relevant statistic always names that base. The result is a ledger that currently holds 19,396 logged flight segments (scheduled included) across 594 members plus the earn and redemption lines behind them, with median 1.5 years of imported statement history per member (p25 0.5, p75 2.8).
Who is in the sample, and who is not
SkyStatus users are people who chose to install a Flying Blue tracker. That makes the sample self-selected and status-focused, and no statistic we publish pretends otherwise. What the sample measures well is how informed, engaged members behave. What it cannot measure is the average of all roughly 30 million Flying Blue members.
Because different statistics need different bases, every statistic names its exact denominator from this ladder: 670 tracked members, of whom 647 members with earning history, 547 members with completed flights and 509 members with imported statements (July 2026). A cabin-mix figure stands on flying members. A card statistic stands on earning members. Copy never swaps one denominator for another.
As a consistency check, we recompute key statistics on the import-sourced subset alone. Where a full-ledger value and its import-subset twin diverge materially, the statistic is flagged and the divergence is documented instead of published as settled.
Privacy floors
The dataset exists because members trust us with their statements, so the privacy rules are hard rules, enforced by the publishing system itself rather than by editorial discipline.
- Aggregates only. We publish medians, quartiles and shares. Never single accounts, never minimums or maximums, which in a small group would describe one identifiable person.
- The 30/10 ladder. A precise number requires at least 30 members behind it. Between 10 and 29 members we state direction only ("rare", "roughly double"), without the precise figure. Below 10 members nothing is published at all.
- Dominance guard. If one account contributes more than 30 percent of the underlying data behind a statistic, that statistic is blocked from numeric publication, whatever its sample size.
- Redaction at the source. The public statistics feed that renders these pages already excludes suppressed values and strips the numbers from direction-only statistics. There is no path from a published page back to a small group's exact value.
- Self-selection framing. Statistics describe "SkyStatus members", never "Flying Blue members". The sample is ours, not the program's.
Statistical guards
Privacy floors do not make a number reliable, so reliability has its own guards.
- Stability check. Key statistics are recomputed with the single largest contributor excluded. If the conclusion flips without one account, the wording is neutralized or the statistic is demoted. Example: the share of redemptions beating our planning value moves visibly with one heavy redeemer excluded, so the published wording stays direction-neutral on that share.
- Wording bounds. Where copy depends on a threshold ("about half", "fewer than 3 in 100", "roughly one in eight"), that dependency is stored as a machine-checked bound next to the statistic. If a re-measurement crosses the bound, publication fails and the wording is rewritten by a human. Claims cannot silently drift away from their numbers.
- Distribution over average. Wherever a distribution exists, we publish median plus quartiles rather than a bare average, because loyalty data is heavy-tailed and averages flatter the tail.
- Hypotheses labeled. When a measured pattern has a plausible but unproven explanation (premium economy XP medians beating business, for example), the measurement is published as fact and the explanation as hypothesis, explicitly.
Definitions that decide the numbers
Most disagreements about loyalty statistics are hidden definition disagreements, so here are ours.
- Realized value per redeemed mile: the cash value of the ticket the redemption replaced, minus surcharges actually paid, divided by miles actually spent, in euro cents. Measured on flown redemptions with a recorded cash value, by members with a EUR profile. It conditions on people who found availability and redeemed. It says nothing about unredeemed balances or failed award searches.
- Ground earning: every positive earn line that is not a flight, including cards, subscriptions, hotels, partners, shopping and incoming transfers. Flight earning is miles earned on flown segments.
- Completed flights: segments that were flown, excluding cancelled and future scheduled segments. Statistics name it when they include scheduled segments.
- Tracked history: the window a member's imported statements and logs cover, median 1.46 years for imported accounts. "Ever redeemed" always means "within tracked history", never a lifetime claim.
- Calendar-year statistics: XP per member and status pace are measured over one calendar year for members active in that year, which is not identical to a member's personal qualification year.
- Per-redemption medians: redemption-value medians weight each redemption equally, not each member or each mile, so frequent redeemers carry more weight than occasional ones.
- Shares of flight XP: UXP and SAF shares are shares of flight XP, not of all XP. Copy states this every time.
Freshness, editions and fail-closed publishing
No statistic on these pages is typed into the text. Each figure is a token that resolves from a central statistics registry at publish time, and every registered statistic is re-measured weekly by re-running its stored, audited query. When the member base grows, every placement updates automatically at the next publish.
Every measurement is stored append-only: re-measurements add a new version, they never overwrite history. Flying Blue by the Numbers carries a measurement stamp that follows the data, and because history is append-only, the measurement set behind any dated citation remains reconstructable after the numbers have moved on.
Publishing is fail-closed. If a re-measurement breaks a wording bound, drops below its privacy tier, or trips the dominance guard, the affected statistic stops publishing and every page carrying it fails its next build until a human rewrites the claim. A broken number cannot quietly stay live.
Independence, terms and citation
SkyStatus is an independent tool, not affiliated with Air France-KLM or Flying Blue. The measurements describe SkyStatus member accounts. Program rules quoted around them come from official Flying Blue sources. The use of anonymized, aggregated member data in our content is covered by the aggregated-data paragraph in the terms of service and the privacy policy.
Citation policy: quote any published figure with attribution to SkyStatus and a link to the page you quote it from, together with the sample size and measurement date the figure carries inline. Published data is licensed CC BY-NC 4.0: free to cite and reuse non-commercially with attribution. For a specific data cut or a question about a definition, contact details are on the about page.
Official References
- Flying Blue Status & XP: the official program rules our measurements are read against
- Flying Blue Terms & Conditions (PDF): full program rules
- CC BY-NC 4.0: the license our published data carries
Frequently asked questions
Can individual members be identified from your statistics?
The floors are designed to prevent exactly that. Everything published is an aggregate over at least 10 members, precise numbers require at least 30, minimums and maximums are never published, and a statistic dominated by one account is blocked. The public statistics feed strips exact values from small-group statistics before any page is rendered.
Why do your numbers differ from The Points Guy or other valuation tables?
Different object. Editorial valuation tables estimate what a mile could return if redeemed well. We measure what redeemed miles did return across real member accounts, published as a distribution. AwardWallet is the exception worth naming: its Flying Blue figure is also measured from user bookings, but published as a single average across hundreds of programs, where we publish the Flying Blue-specific distribution. A good planning valuation should sit near the middle of our measured distribution, and ours does.
How current is a number I see on a SkyStatus page?
Each figure carries its measurement month inline. Statistics are re-measured weekly, and pages re-publish from the registry, so a figure is at most days old at publish time. If a measurement invalidates a claim, the page fails its build rather than showing the stale claim.
This methodology page describes the measurement system behind all SkyStatus member-data statistics. The system itself enforces the floors and guards described here: they are code, not policy intentions. Last verified: 30 July 2026. SkyStatus is an independent tool and is not affiliated with Air France-KLM or Flying Blue.