CourseToJob guide · content date . Tool instructions reflect the CourseToJob service. Editorial method.
Quick answer
Take repeated tools, qualifications and duties from the advert and use them as words in true sentences. Do not stuff. A low overlap score often means you should skip the vacancy, not that you should lie. This checker is a gap list, not a hiring oracle.
Other situations: Skip keyword theatre if you are emailing a founder who will read 20 CVs by hand. Write clearly; do not optimise for a database that is not in the loop.
How do you extract keywords from a job description?
Read the advert twice. Highlight tools, qualifications, domain nouns and verbs that appear more than once. Ignore padding (“passionate”, “fast-paced”). You should end with six to twelve terms. If you cannot map four of them to real work, the job is a mismatch.
Paste the advert into the builder’s matcher for a second pass. Treat the output as homework, not as a score to game. Open the builder.
Where do keywords belong on the CV?
In the profile’s first line (job title as they wrote it), in the skills list, and inside evidence bullets. Not in white text, not in a keyword cloud at the footer, not twelve times. Parsers and humans both see dumping.
ATS context: ATS-friendly CV.
When should a low keyword match make you skip the job?
When the missing terms are licences, years of a named tool, or a clearance you do not have. When the gaps are tone words, ignore them. Honesty beats a 90% overlap built on fiction — interviews test the fiction immediately.
This is the “say no” the big CV mills will not print: some adverts are not for you. Saving the application time is a win.
What does extracting keywords from one advert look like?
Take a fictional UK analyst advert that repeats SQL, stakeholder reporting, Excel, and “NHS datasets”. Your keep-list is those four, plus any registration they named. Your drop-list is “passionate”, “fast-paced”, and tools you have not used. Map each keep-term to one bullet. If SQL is not yours, skip the vacancy rather than adding the word.
| Advert phrase | Keep? | Where it goes if true |
|---|---|---|
| SQL (mentioned four times) | Only if you wrote queries | Skills list + one bullet with table size or report name |
| Stakeholder reporting | If you actually sent the pack | Profile or latest role; name the audience |
| Passionate about data | No — padding | Nowhere |
| Tableau | Only if you built dashboards | Skills; do not list it from a two-hour tutorial |
Then paste the advert into the builder matcher as a second pass. Gaps that are licences or years of a named tool are knockout. Gaps that are tone words are noise.
What does keyword stuffing look like, and why does it fail?
Stuffing is the same phrase in the profile, every bullet, a footer cloud, and sometimes white text. Humans see it immediately. Some parsers ignore tiny or hidden text; some highlight it as a reason to discard you. Once per skill, in a true sentence, is enough. If the file reads like the advert concatenated, you overshot.
White text, zero-point type and keyword walls are old tricks. They are also easy to detect. We will not teach you how to hide strings. We will tell you to delete them. ATS context: ATS-friendly CV.