Spam costs a second, a message, or your savings
One a day is 365 interruptions. A wanted mail in Junk costs what it was for. A phishing mail in the inbox cost two French households their savings.
Our evidence page on Apple Mail ends on one line: spam in the inbox costs a second, wanted mail in Junk costs whatever the message was for. It stops one rung short. The third cost is the account, and two public French cases from this summer put a figure on it.
The second, 365 times a year
A spam mail you see and delete costs a few seconds and a change of focus. One a day is 365 interruptions a year. Gloria Mark’s team at UC Irvine measured about 23 minutes to get back to a task after an interruption (CHI 2008). On an 8-hour day that is 17 workdays a year, and the chart scales it.
Workdays a year spent refocusing
Spam mails a day, turned into interruptions a year, at 23 minutes to refocus after each one and an 8-hour workday.
| Spam a day | Interruptions a year | Workdays a year |
|---|---|---|
| 1 | 365 | 17 |
| 3 | 1,095 | 52 |
| 5 | 1,825 | 87 |
| 10 | 3,650 | 175 |
Arithmetic only: count × 365 × 23 minutes ÷ 480. The 23 minutes are Mark, Gudith and Klocke (CHI 2008); the count is yours.
The arithmetic stops at days. Turning them into money needs a wage, and we do not know yours.
The message, whatever it was for
A wanted mail filed in Junk does not cost the minute to move it back, because nobody checks Junk daily. It costs whatever the message was for. On our evidence page, Apple Mail users report losing recovery mail, verification codes, Apple’s own billing notices, and a theatre’s email about a rescheduled show. One freelancer’s client now texts to confirm every email arrived.
The account: two households, one summer
The third cost ends with money leaving. Two French cases from 2026, both public, both told by the people it happened to, and both beginning with an email.
| Murielle and Jean-Marc | Anna-Belle | |
|---|---|---|
| First contact | An email, they say on camera | An “Amazon” mail in her main inbox, 22 June |
| Then | A caller who knew their IBAN, their adviser’s name and their children’s birthdays | A card page copying Amazon. Four days later, a caller citing that mail: over three hours, WhatsApp, screen sharing |
| Loss | Over 6,000 euros, all their savings | 8,640 euros established, about 10,000 by her count |
| Source | Cash Investigation, France 2, 10 September 2026 | Her 21-page written account |
“22 juin 2026 : je reçois un mail “amazon”, dans ma boite principale, qui m’indique la fin d’essai de mon abonnement Amazon Prime.” That is the first page of Anna-Belle’s account, written for her bank’s fraud service and the investigators. The page it links to asks for card details, “je trouve cela bizarre mais réalise l’action, le site est semblable à celui d’amazon”. Four days later a caller introduces himself as her bank’s anti-fraud service and cites “le mail reçu le 22 juin”. Everything after the first message is a voice on the phone, and no filter reaches it.
$3.05 billion is what businesses and individuals reported losing to business email compromise in 2025, to the FBI’s Internet Crime Complaint Center. A total across complaints, in one country, for one fraud category: not your expected loss, not a claim about any filter, and not a rate. It says how large the third cost is once added up.
Which one are you paying?
Three costs
Three costs, three different responses, and only one of them is a filter.
The last message that was not where it should have been: what was it?
The interruption, the cheapest of the three and the only one you can see.
Count them for a week. One a day is 365 a year, and a filter that files them before you see them gives that attention back. This is the cost a spam filter is for.
What this page does not prove
Two stories and one national figure do not measure your inbox. Four limits:
- Television and Instagram select the dramatic cases. No denominator comes with them.
- Both cases are manipulation fraud. The email is the first link; every later step is a phone call, a fake page, a person on the line. A filter reaches the first link only. For the couple, the email is in what they say on camera, not in the caption we reproduce; for Anna-Belle, it is the first page of her account.
- The IC3 figure is reported losses, mostly from businesses, in the United States, for one fraud category. A total predicts nothing about any one inbox.
- We build a spam filter, and our own engine delivered a reconstruction of the Amazon lure until a fix the same day. Nothing on this page claims we would have caught it.
Where Klar fits
Klar reads each message as it arrives on your Mac and files spam and phishing before you see it, with a language model rather than a word list, on your machine. It works on the first link of the chain and no other. For the later links, the scam dictionary shows the pressure patterns and where to report; for companies, the handling-cost calculator keeps the time cost and the fraud risk as two separate arguments. The evidence for the second cost is on the Apple Mail page.
Every source on this page
In the order they appear. Counts are as read on 11 September 2026.
- The cost of interrupted work: more speed and stress Gloria Mark, Daniela Gudith and Ulrich Klocke, CHI 2008, the source of the 23-minute figure on our homepage
- What people say about Apple Mail spam filtering klar.im, 10 September 2026, the sourced complaints behind the second cost
- « On a tout perdu. » france.tv on Instagram, reel for Cash Investigation, 10 September 2026, 2,916 likes and 196 comments
- Cyber-arnaques : État et entreprises nous protègent-ils vraiment ? Cash Investigation, France 2, 10 September 2026, replay on franceinfo
- Le témoignage de ma mésaventure est dans la bio Instagram, @annabelle_explore, 11K likes and 472 comments
- Fraude au faux conseiller bancaire, reconstitution chronologique des faits Anna-Belle Alegre Gonthier, 21-page PDF linked from her Instagram profile; the quotes above are its first page
- 2025 Internet Crime Report FBI Internet Crime Complaint Center, business email compromise: $3,046,598,558 in reported losses
