Tailoring
Tailoring is not swapping the job title at the top and adding three keywords. It is deciding which parts of your history the reader sees at all, in what order, and in whose words. Here is the method, an example worked line by line, and an honest answer on whether it pays to do it every time.
The method
The strongest edit is a cut. A posting that talks about scaling reads worse, not better, when your migration work is buried under two lines about a hackathon. Tailoring is mostly deletion, and deletion is the part people skip because it feels like losing something.
A recruiter reads the top third and skims the rest. If the thing they asked for is your fourth bullet in your second job, on this posting it is invisible. Nothing about your history changed; the ranking did.
You built an event pipeline. They call it streaming ingestion. Both are true, and only one of them matches what the person searching is typing. Use theirs when it is honest, keep yours when theirs would be a lie.
Dates, employers, scope, titles, numbers. If tailoring is editing those, it is not tailoring. It is also the one kind of mistake that ends a process at the reference call rather than at the screen.
Worked example
The posting is a composite, and the bullets are written for this page rather than lifted from a customer: what a person hands us is their document, not our teaching material. The edits are the ones the pipeline actually makes.
What the posting asked for
"You will own the reliability of a payments service handling several thousand requests per second, and lead the on-call practice for a team of six."
Before
Worked on backend services in Go and Python for a fintech platform.
After
Owned three Go services on the payments path, 4k req/s at peak, 99.95% availability over the last year.
Why
Same job, same year. The first sentence describes a role; the second answers the question the posting asked. The numbers were always in the profile, and the untailored version had no reason to spend a line on them.
Before
Participated in the on-call rotation and helped improve alerting.
After
Ran the on-call rotation for six engineers. Rewrote the alert set after a quarter of 40+ pages a week, ending at 9.
Why
"Participated" and "helped" hide who did it. The posting asks for someone to lead the practice, so the bullet has to say whether you led one, and what it looked like before and after.
Before
Experience with Kubernetes, Terraform, Kafka, Redis, gRPC, and CI/CD.
After
Cut deploy time from 40 minutes to 6 by moving the release path onto Terraform and reusable CI templates.
Why
A tool list is a claim nobody can check and everybody makes. One sentence about what a tool did is a claim a person can ask about in an interview, which is the point of a CV.
Nothing in the right column is new information. It was all in the profile, and the posting decided which parts were worth the space.
The line
Keyword stuffing is what people do when they cannot tell the difference between a term and the evidence behind it. Pasting "Kubernetes" into a skills block does not make you a person who has run a cluster in anger, and the interview finds out in about four minutes.
The rule that works: use the posting's word when you can defend it in a sentence. If they say observability and your bullet can say what you instrumented and what it caught, take their word. If they say observability and your honest answer is that you read the dashboards somebody else built, keep your own wording and let the bullet be smaller.
This matters more than it used to. A recruiter's search over their own applicant pool is a keyword search, and an untailored CV is invisible to it. That is an argument for using the right words, not for using all of them.
Paste a job description into our keyword finder and it splits what the posting demands from what it merely prefers, then lists the terms your CV does not mention. It runs in your browser, needs no account, and costs nothing.
Pull the terms out of a postingThe arithmetic
The honest answer starts with your own number, not ours. Time yourself doing it properly once: reading the posting, picking the evidence, re-stating six bullets. Whatever that comes to, multiply it by the number of applications a real search takes, which for most people is dozens and not five.
Then the other side. A CV written for one posting returns an interview rate of 6 to 9% across our customers' accounts. Twenty of those is one or two conversations. If you would rather have those twenty documents than the thirty hours, that is the trade, and it is a trade about your time before it is a trade about money.
There is a third option people forget: apply to fewer jobs. Twenty applications you actually wanted beats a hundred you did not, and it is the only version of this where tailoring by hand stays sane.
Read the posting twice and write down what it demands as opposed to what it prefers. Cut the bullets that answer neither. Re-order what is left so the demands are answered in the top third. Then re-state each surviving bullet with the posting's own noun where that is honest, and add the number you already know. Do not touch dates, titles or employers.
There is no count that helps. Every term you use should have a sentence behind it that you would be happy to be asked about. A CV with eight defensible terms beats one with thirty that collapse in the screen call.
No. Your title is a fact about your employment, and it gets checked. If your real title undersells you, keep it and let the bullets carry the scope, which is what a reader is actually judging.
Not usually by itself, and anyone who tells you exactly how a given system scores you is guessing about software they cannot see. What is not a guess is the recruiter's own search across their applicant pool, which is a keyword search. Missing their words means not being in the result set, and not being in the result set means nobody formed an opinion about you at all.
For anything you want, yes. Volume is a strategy for jobs that are interchangeable, and if the jobs are interchangeable then the interviews are too. The reason we sell one CV per posting rather than a subscription is that we do not think the hundredth application is worth writing.
It can select and re-state evidence, which is most of the work, and it is fast enough to do it per posting rather than once a quarter. It cannot invent the evidence, and a model left alone will happily write a confident sentence about something you never did. That is why a draft here is read by 25 independent reviewers with the profile in front of them before it becomes a PDF.
Both are on the table. This page is the method, and it works with a text editor and no account. If you would rather not do it by hand, that is what we sell: you approve a posting and the document comes back written for it.
Free account, free market watching, one payment on the day a posting is worth a document.