Cold email was already easy to ignore. A note that could have been sent to any company, with the same compliments and the same “I would love to pick your brain,” now has a second problem. Readers suspect a machine wrote it, because a machine could have. Greenhouse’s candidate report, announced 20 October 2026 from a U.S. survey of 1,200 people, says about 6 in 10 job seekers were rethinking how they write so they would not sound like AI, even while most used AI somewhere in the search. The headline figure is a behavior. People are editing in order to sound like a person. Your note has to survive that suspicion.
The same release has the surrounding weather. Eighty-one percent said they had been rejected without speaking to a person, and 38 percent of Gen Z respondents said that happens regularly. Forty-five percent of people who reached an interview said they were ghosted afterward. Forty percent said they had added skills they do not have, as a way to get through screening. Seventy-one percent had taken some extra step to be noticed. Nearly half of that group sought a referral, and 27 percent paid for a resume service. This post uses the writing figure and the fake-skill figure as the instructions. The rejection and ghosting figures explain why people are tempted to spray notes and stuff keywords. They do not justify either habit.
What a machine-shaped note looks like
On 15 October 2026, @GergelyOrosz described a student’s cold email and follow-up that both read as clearly machine-written, and said that pattern wrecks cold email for everyone else. The post is one recipient’s reaction, with a large audience. It is not the Greenhouse survey. It matches the survey’s premise. If a careful reader can tell the note was not written for them, the note is done, whether a person or a tool drafted the first pass.
A note that sounds like you has limits a template hates. It names one role. It names one fact the reader could check: a sentence on their team’s public page, a product behavior, a method in a paper. It names one fact about you that matches, from work you did. It asks for one thing: a referral, a ten-minute conversation, or a pointer to the right person. It fits in half a screen. If you remove the employer’s name and the note still works for fifty other employers, it is the kind of note the October post was complaining about.
You can use a tool to check grammar. You should not use a tool to invent the fact, the enthusiasm, or the skill. The 40 percent who added skills they do not have are the caution on the resume and in the note at once. A skill in the first paragraph that you cannot explain in the second is the same failure as a keyword you cannot defend in an interview. Greenhouse’s respondents were describing a screening tactic. The cost shows up later, when a person asks and the skill is missing.
A structure you can reuse without sounding reused
Four sentences.
Who you are, in one line: year, major, and the role’s title.
Why this reader: the specific public sentence you read, quoted briefly so they know you are not guessing.
Why you: one project or job task, with the result in their vocabulary only where the vocabulary is true.
The ask, and an easy no. “If you are the wrong person, I am glad to be pointed elsewhere.” Then stop. Do not add a paragraph about your passion for the industry. Passion is the word templates use when they have no fact.
Wait a week if you hear nothing. One follow-up can restate the ask in three lines. A second follow-up needs a new reason, such as a decision date you actually have. After that, close the row. The 81 percent rejection-without-a-conversation figure is what a silent note often is. Another generated follow-up will not beat it. A different role, with a real fact, might.
If you are writing to a student or a recent graduate instead of a hiring manager, the same four sentences work, with a smaller ask. People your age can tell a template even faster, because they use the same tools. The Greenhouse survey is of job seekers. Students were not the whole sample. The editing behavior it describes is already in the rooms you are writing to.
What not to copy from the 71 percent
Taking an extra step is common in this survey. A referral from someone who has seen your work is an extra step that matches the nearly-half figure without the damage of a fake skill. Paying for a resume service is a fact about 27 percent of the people who took an extra step. A service that writes experiences you did not have is the 40 percent problem with an invoice. A service that fixes layout is a purchase, and you still have to check every line.
Do not add a skill because 62 percent of respondents tried tactics to get through screening. The survey is describing what people did under pressure. It is not a recommendation. Your test is the interview you may not get, and the interview you may get. In both cases the skill has to be yours.
Keep the October 2026 date on the figure. A later survey can move the 6 in 10. The writing rule will not move. One role, one checkable fact, one true line about you, one ask.
How Flightpath helps
A draft in Flightpath is built from facts you confirmed and from the posting you pinned. You are supposed to edit it until it sounds like the four sentences above. If the draft could be sent to a different employer unchanged, it is not ready, and you are the person who can tell. You approve the exact text before anything is sent. The product does not mail the note on its own.
Gmail is where the approved text goes out, from you. The tracker records the role and the date so the one follow-up has a clock. People suggestions come from public contacts or a search link. A role you cannot take stays labeled and is never ranked above one you can take. Greenhouse’s shares stay in the October 2026 release. The fact in the second sentence stays something you actually read.