Three outlets cited Kpler on Hormuz in sixteen days and told readers three incompatible things
Flows “topped pre-war levels,” hit a “two-month low,” and ran “43% below pre-conflict levels” — all sourced to the same commercial shipping tracker, none labelled with the metric being counted.
On 5 October, Oilprice.com reported that Kpler data showed crude flows out of the Strait of Hormuz topped pre-war levels in late September, with Brent nonetheless above $100. On 8 October, CNN reported that record tanker attacks had pushed crossings to a two-month low, the fewest since 23 July, per preliminary Kpler figures. Around 16 October, the analysis site DiscoveryAlert put the seven-day average flow at roughly 43% below pre-conflict levels, again citing Kpler, alongside Brent at $104.51 and a $12 Brent–WTI spread. DiscoveryAlert is not a news publisher and that 43% should be treated as unverified until confirmed with Kpler or against a wire. The three figures almost certainly measure different things — export volume against total transits, a seven-day average against a monthly one — but they reach readers as the same fact.
CNN was the only one of the three to tell readers the data is preliminary and systematically undercounts, because ships switch off transponders and figures are often revised upward later. That caveat sits in CNN’s body text, not its headline. The same window shows the attribution problem in miniature: Euronews on 14 October headlined “Oil prices rise after reported Iranian attacks on ships in Strait of Hormuz,” the word “reported” carrying an unestablished attribution. Meanwhile Iranian Foreign Minister Abbas Araghchi told Iranian media on Sunday 18 October that Tehran has not withdrawn its offer to reopen the Strait within seven days if Washington accepts its conditions — an offer Trump rejected on or about 26 September, and which this desk could find no Western outlet treating as still standing.
Assessment: This is an editorial failure rather than a statistical one. Kpler publishes several distinct series; the newsroom convention of writing “per Kpler data” without naming the metric or the window converts a vendor feed into a scoreboard nobody can audit. The fix is cheap — one email settles whether these are one dataset or three — and nobody has sent it. Until someone does, treat any Hormuz percentage lacking a stated metric and date range as decoration. CNN deserves credit for disclosing the undercount and criticism for burying it where headline readers, who are most readers, will never reach it.