How to Tailor a Cover Letter to a Job Description

Quick Answer

• The distinction: the requirements list is written to filter; the responsibilities section is written by the person you would work for
• Tailoring is: naming the problem the posting describes, then naming a comparable thing you have actually done
• Tailoring is not: swapping the company name and returning the requirements as adjectives
• The evidence: tailored letters beat no letter in a large pre-AI field experiment; newer work finds alignment alone is a much weaker signal
• The consequence: matching a posting is cheap now, so the only part of your letter carrying information is the part the posting cannot supply

What the Evidence Actually Shows

The best-known number in this argument is also the oldest. ResumeGo ran a field experiment between 15 July 2019 and 10 January 2020: 7,287 fictitious applications to real openings on ZipRecruiter, Glassdoor and Indeed, split three ways — no cover letter, a competently written generic one, and one tailored to the specific job. Applications carrying a tailored letter came back with a 53% higher callback rate than applications with no letter at all.

Source: ResumeGo cover letter research, accessed September 2026 — 7,287 fictitious applications via ZipRecruiter, Glassdoor and Indeed, 15 July 2019 to 10 January 2020. Caveats: the applicants were fabricated, not real jobseekers; ResumeGo is a resume-writing service and so has a commercial interest in application documents mattering; and the per-group rates are published only as chart images, so just the relative figure is quoted.

That is pre-generative-AI fieldwork, and anyone quoting it today without saying so is selling you a 2019 conclusion. The better question is what happened to the effect once writing a matched letter stopped costing an evening.

There is research on exactly that. Jingyi Cui, Gabriel Dias and Justin Ye studied a large online labour platform that introduced an AI cover-letter writing tool, with data tracking its use application by application. Access increased the textual alignment between letters and job posts, and it raised callback rates — this is not a debunking. But after the tool arrived, the correlation between a letter's textual alignment and whether it got a callback fell by 51%, which is what theory predicts if the technology has reduced how much a matched letter tells anyone. Employers moved toward other signals, including workers' prior work histories. One more finding is worth keeping: time spent editing the AI-generated draft was positively correlated with hiring success.

Source: Jingyi Cui, Gabriel Dias and Justin Ye, Signaling in the Age of AI: Evidence from Cover Letters, arXiv:2509.25054, first posted 29 September 2025 and revised 12 November 2025. It is a working paper rather than peer-reviewed research, it covers one online labour platform rather than corporate hiring, and the editing result is a correlation the authors do not present as cause.

The Two Lists in Every Posting

Open any posting and you find two blocks of bullets that look alike and are not. They are usually produced by different people for different purposes, and they reward completely different treatment.

That split is structural rather than a research finding, and plenty of postings are written end to end by one person. But the tell is reliable enough to act on: requirements read like a specification, responsibilities read like someone describing their week. When a bullet contains currently, rebuild, establish or first, you are reading a complaint rather than a duty — written down by the person who has to live with it. Our breakdown of what job descriptions actually ask goes further into the competency underneath those lines.

One Posting, Read Properly

Everything below runs on one realistic posting. Read it twice: once as a candidate deciding whether to apply, once as someone working out what is going wrong at this company.

The requirements tell you almost nothing about the job. Hundreds of applicants match them and several hundred will say so. The responsibilities are a different document:

• "Own the pipeline number reported to the board" — the hard part is defending a figure senior people look at, not producing it.
• "A dashboard the sales team does not trust" — a credibility problem between two functions, not a tooling problem. Someone sat through that meeting recently.
• "The weekly lead handover review with SDR leadership" — a standing negotiation about lead quality, every week, forever.
• "One contractor and one agency relationship" — not a people-management role. One where your output depends on people you do not employ.

Four lines, and you know what the hiring manager worries about on a Sunday. That is the material, and the same reading is how you prepare answers as well as letters — which is why our guide to preparing from the job description starts in the same place.

The Letter Most People Send

Here is a competent letter for that posting — not a straw man. It was written by someone who read the posting and consciously tailored to it.

Every phrase in that second paragraph lands on a requirements bullet: HubSpot, Google Analytics, Excel, the degree, the communication line, the fast-paced line. The letter is tailored — to the list a recruiter assembled in order to filter people. That is why it reads as interchangeable. It is.

The Same Letter, Rewritten

Same candidate, same facts, same length. The only change is which half of the posting it answers.

Nothing in draft two is more impressive. The candidate acquired no new skill between versions. Every changed phrase traces to a specific line of the posting — and, with one deliberate exception, they all come from the responsibilities.

The last row is the one people get wrong. The requirements did not disappear, they were demoted: one clause states that the candidate clears the bar, and the paragraph draft one spent proving eligibility is spent in draft two on the work.

Now look at what each rewritten sentence is made of. "A dashboard the sales team does not trust" came from the posting; "I rewrote the attribution rules with our RevOps lead first" could not have. Half of every strong sentence comes from the job description, and the other half from somewhere the job description cannot reach.

The Part a Posting Cannot Supply

Run the exercise above with only the posting in front of you and you get most of the way — to a letter about the job, not a letter about you. No amount of reading Northvale's bullets produces the attribution rules being fixed before the dashboard, or the rejection-reason list. That material exists in one place, and it is not on the careers page.

This is what the fall in signal value means in practice. When matching a posting took an evening, a matched letter told an employer something. Now that it is nearly free, a letter that only matches tells them nothing they did not already have — they wrote the posting. The letters still saying something carry facts nobody could have produced from the advertisement.

It also gives the editing finding a sensible reading. That editing time correlates with hiring success is not evidence that reworded text performs better; the duller explanation is likelier — editing is where a person puts in what the machine could not know. Worth being blunt about, because none of this is about disguising that you used a tool or making text look less machine-written. If an employer asks, tell them; we work through what they can and cannot detect in can employers tell if a cover letter was written by AI.

A Method You Can Repeat

The rewrite above is a process, not an inspiration, and it survives being done at speed on your ninth application of the week.

The reuse rule falls out of that order. Steps four and five are yours and barely change; the rest belong to the employer and change every time. You need a fresh argument, not a fresh set of achievements — the real answer to the question we unpack in do you need a different cover letter for every job.

What Does Not Count as Tailoring

Most of what gets called tailoring is find-and-replace with extra steps.

• Swapping the company name and job title. Proves only that you noticed which posting you were on.
• Returning the requirements as adjectives. "Detail-oriented, collaborative, results-driven" is a mirror, and nobody learns anything from their own reflection.
• Quoting the "about us" paragraph back at them. They wrote it. Admiring it is not information.
• Stuffing in keywords from the requirements. The paragraph you spend on keywords is the paragraph you did not spend on the responsibilities.
• Declaring passion instead of demonstrating relevance. Enthusiasm is the one claim every applicant makes and none can evidence.

A letter aimed at the requirements list is a wasted page, and you get one page. If you are applying steadily and hearing nothing, the letter is rarely the whole story — our diagnosis of why you are not getting interviews covers the rest. And when you clear the responsibilities but not the requirements, the logic does not change; it just needs one extra sentence, the case we handle in writing a cover letter when you do not meet the requirements.

Frequently Asked Questions

Next Steps

It reduces to one line: matching a posting no longer distinguishes you, so the specifics have to be real. The posting supplies the problem; you supply the evidence. A letter with only one of those halves is either interchangeable or irrelevant, and the two failures look nothing alike on the page.