# The Resume Tailoring Teardown, One Posting Mapped to a Sent Draft

*You can take your master resume and one posting, rank the posting's terms, rewrite the lines that move the match, and know when to stop editing.*

- Canonical URL: https://www.refolk.ai/candidates/guides/resume-tailoring-teardown
- Pillar: Applying at volume
- Format: Teardown
- Published: 2026-08-14
- Last reviewed: 2026-08-14
- Reading time: 16 min

You have your master resume open and one specific posting beside it, and you need to tailor before you hit submit. This guide carries a single case all the way through: the term extraction, the ranked map, the baseline score, each bullet rewrite, the recomputed match, and the two edits I abandon because they do not move the signal. It sits beside the per-posting edit framework, which scores whether an edit is worth the time; here I do the work line by line so you can follow along on the posting in front of you.

The worked case is a US Product Manager role. I chose it because Refolk's index gives me real numbers to reason with: how common each skill actually is, how often the literal title even appears, and therefore which edits earn their place. Your posting will differ, but the moves are the same.

## What "done" looks like before you start editing

Done is a draft that lands in the 75 to 85 percent keyword-match band, reads like a person wrote it, and took roughly 15 minutes. That band is the whole target, and naming it first stops you from over-editing.

There is no universal pass score, because each employer configures its own cutoff. Public scanners cluster the bar between 60 and 80 percent: 60 percent or higher is generally treated as good, and below 40 percent a resume is likely filtered before a recruiter sees it. Analysis of screening thresholds puts human referral at roughly 70 to 80 percent coverage of the required qualifications. So the floor is real and the ceiling is a trap.

| Source | Cited pass bar | Note |
|---|---|---|
| loopcv | 60%+ good | below 40% likely filtered |
| Jobalytics | 70%+ | aim for 70% or higher |
| airesume.guru | 75-85% | sweet spot, natural and readable |
| ResumeFry | 70-80% | 80% comfortably above threshold |
| CVCraft | 55-70 cutoff | avg ~60 |

Read the table as a range, not a contradiction. The lowest credible cutoff is around 55 to 60; the highest human-review threshold is around 80. Landing at 75 to 85 clears all of them with margin. Going higher costs disproportionate effort for shrinking gain: moving from 80 to 90 costs far more than moving from 70 to 80, for a smaller benefit, and a 100 percent match often indicates stuffing that a human reviewer penalizes.

**75-85% - The match band to stop editing in**

High enough to clear most filters, low enough to still read naturally to a recruiter.

## Extract the posting's terms field by field

Copy the whole posting first, then pull concrete terms out of each field. The requirements section is the keyword goldmine; the responsibilities section yields action-verb-plus-skill combinations; and any tool or platform named anywhere is high priority.

For the worked case, here is the posting I am tailoring against, compressed to its load-bearing text:

> Product Manager. You will own the roadmap for our payments product, run experimentation and A/B testing, and partner with engineering. Required: 4+ years in product management, strong SQL, experience shipping B2B products, comfort with experimentation. Preferred: Salesforce familiarity, fintech background, experience at a company under 200 people.

I read it field by field and write down only concrete terms, discarding filler like "fast-paced" or "team player" that no scanner weights and no recruiter remembers. The raw list:

- Title: Product Manager
- Required: product management, SQL, B2B products, experimentation, A/B testing, roadmap
- Preferred: Salesforce, fintech, company under 200 people
- Responsibility verbs: own, run, partner, ship

That is 13 concrete terms, comfortably inside the documented 15-to-25 range that most postings carry. If your list runs short, you are probably discarding real terms as filler; if it runs long, you are keeping filler as terms.

> **Tip:** Grab the literal strings, not your paraphrase
>
> Write down the posting's exact wording, including capitalization and acronyms. You will need the literal string later, because many scanners will not connect your paraphrase back to the term they are searching for.

## Rank the terms into two tiers

Rank every extracted term into Tier 1, the non-negotiables, and Tier 2, the nice-to-haves. Tier 1 is the exact job title, everything in the required section, and anything repeated anywhere; Tier 2 is the preferred section.

Required and must-have terms carry more weight in scoring; preferred terms still add points but less. Two signals sharpen the ranking beyond section labels. First, repetition: prioritize any term mentioned more than once, because the posting is telling you what it cares about. Second, rarity: a term that repeats across many postings is an industry-wide standard, while a term unique to this posting is a job-specific priority worth more.

Here the sources genuinely disagree, and it is worth knowing where. Harvard Business Review's framework treats top-of-list position as highest priority. Other practitioners weight job-specific rarity over list position. I resolve it by using position as the tiebreaker, not the primary sort: required beats preferred, repeated beats mentioned-once, and only then does list order decide.

Applied to the case:

- **Tier 1:** Product Manager (title), SQL, experimentation, A/B testing, product management, B2B products, roadmap.
- **Tier 2:** Salesforce, fintech, company under 200 people.

Now rarity tells me which Tier 1 and Tier 2 terms actually differentiate me. This is where Refolk's index earns its keep. Among US professionals titled Product Manager, here is how common two of the posting's tools are:

| Skill term | Profiles listing it | % of US PM base | Rarity vs SQL |
|---|---|---|---|
| SQL | 6,409 | 9.6% | 1.0x (base) |
| Salesforce | 199 | 0.30% | 32x rarer |

Read that carefully. SQL is required and listed by fewer than one in ten US PMs, so matching it moves me above most of the field. Salesforce is only preferred, but it is 32 times rarer among PMs, so if I genuinely have it, it differentiates me hard. The lesson generalizes: a term nearly everyone lists barely moves your ranking, while a rare one signals fit. Match the rarer required terms first, and treat a rare preferred term as a bonus you claim only if it is true.

> A term everyone lists barely moves your ranking; a rare one you can actually prove is where the leverage lives.

## Baseline the match before you touch a word

Run your unedited master resume against the posting first, so you have a pre-edit percentage and a list of flagged gaps. Without a baseline you cannot tell which edits moved the number and which just cost you time.

In the worked case my master resume scores **62 percent** against this posting. That is inside "good" by loopcv's floor but below the human-review threshold, so it is worth editing. The scanner flags four missing terms: the exact string "Product Manager" is not my current title (my resume says "Senior Product Owner"), "A/B testing" appears nowhere, "B2B" is implied but never written, and "SQL" sits in a skills list but never in a bullet.

That flag list is the map. Everything after this step is closing named gaps and rescoring, not guessing.

#### The posting narrows to the edits that matter

| Stage | Figure | Note |
| --- | --- | --- |
| Terms extracted | 13 | every concrete term in the posting |
| Tier 1 priorities | 7 | required plus repeated |
| Flagged as missing | 4 | absent from the master resume |
| Rewritten to close | 4 | title, summary, skills, top bullets |

*Thirteen extracted terms collapse to the four flagged gaps that actually move a 62 percent baseline.*

## Cross-check every Tier 1 term against real evidence

Before you rewrite anything, prove each Tier 1 term. For every term you plan to include, name the exact bullet or achievement where it is demonstrated, or drop the term. This one check is the line between optimization and stuffing.

Listing a skill you genuinely used is optimization. Listing one you never touched because it appears in the posting is stuffing, and it falls apart in the first interview question. The scanner cannot tell the difference; the interviewer can, in about one sentence.

Here is the cross-check for the case:

| Tier 1 term | Proof in my history | Verdict |
|---|---|---|
| Product Manager | Same role, different title label | Rewrite title |
| SQL | Wrote queries for cohort analysis | Keep, move into a bullet |
| A/B testing | Ran pricing experiments | Keep, name it explicitly |
| B2B products | Two B2B roles | Keep, make literal |
| Roadmap | Owned quarterly roadmap | Already present |

Every Tier 1 term survives because I can point at real work. If "A/B testing" had no proof, I would drop it and accept the lower score rather than plant a claim I cannot defend. Do the same on your posting: a proof bullet or a deletion, never a bare claim.

> **Rule:** A term with no proof bullet does not go on the resume
>
> If you cannot point to the exact achievement that demonstrates a skill, remove it. A scanner score you earned by claiming a skill you lack is a score that collapses in the first interview question.

## The step-by-step, start to finish

Here is the full procedure as a checklist you run in order. Steps five and seven are the ones people skip and then wonder why the draft either fails the filter or fails the interview.

#### Tailor one resume to one posting

1. **Assemble inputs** - Open your master resume beside one full posting and copy the entire posting: title, overview, required, preferred, and responsibilities.
2. **Extract terms field by field** - Pull skills, tools, certifications, and the exact title from each field, ignoring filler like fast-paced environment.
3. **Rank into tiers** - Split into Tier 1 (title plus required plus anything repeated) and Tier 2 (preferred and nice-to-have).
4. **Baseline the match** - Run the master resume plus posting through a scanner for a pre-edit percentage and a flagged missing-terms list.
5. **Cross-check each term against evidence** - For every Tier 1 term, name the bullet that proves it, or drop the term.
6. **Rewrite the load-bearing lines** - Put the exact title in the summary, mirror skills, and rewrite the top two or three bullets per role to carry Tier 1 terms.
7. **Recompute and stop at the threshold** - Rescore, confirm the 75 to 85 band and a natural read, and stop.

Now the rewrites, shown before and after, so you can see exactly which words moved.

**Title.** This is the single highest-leverage edit. Postings are scored against their own literal terms, and the title is the most-searched one. An exact-title match is cited at roughly 10.6 times interview likelihood. In Refolk's index only about 84 percent of a US PM sample even used the literal string "Product Manager"; the rest wrote variants like "Founder & Head of Product." So copying the posting's exact wording is a cheap gain that many candidates leave on the table.

- Before: `Senior Product Owner`
- After: `Product Manager`

**~84% - US PMs whose current title is the literal string "Product Manager"**

In a 25-profile sample from Refolk's index; the rest used variants, so exact-title matching is a real, common gain.

**Summary.** Put the exact title and one or two primary keywords in the first line.

- Before: `Product leader with 5 years building tools users love.`
- After: `Product Manager with 5 years shipping B2B payments products, from roadmap to experimentation.`

**Skills.** Mirror the posting's literal strings, and pair every acronym with its spelled-out form so a scanner that does not resolve abbreviations still registers both.

**Skills line, literal-mirrored**

```
Product management, SQL, A/B testing, experimentation, B2B product strategy, roadmap ownership, Search Engine Optimization (SEO)
```

*Include only terms you proved in the cross-check. Pair acronyms with spellouts.*

**Bullets.** Rewrite the top two or three under each role so Tier 1 terms appear in context, not just in the skills block. Use each high-priority term two to three times across zones.

- Before: `Analyzed user data to inform pricing decisions.`
- After: `Ran A/B testing on pricing across two B2B products, writing the SQL cohort queries that set the winning tier.`

That single rewritten bullet lands SQL, A/B testing, and B2B, all three flagged terms, in one line of true work.

## Recompute, and the two edits I abandoned

Rescore after the rewrites, confirm you are in the 75 to 85 band, and stop. In the worked case the draft moves from 62 to **79 percent**, which sits comfortably in the sweet spot: high enough to clear most filters, natural enough to survive a human read.

The instructive part is the two edits I did not keep, because knowing when to stop is half the skill.

**Abandoned edit one: chasing Salesforce.** Salesforce is a preferred term and 32 times rarer than SQL among PMs, so it looked tempting. But my cross-check found no proof bullet: I have never used it. Adding it would have lifted the machine count and collapsed in the interview. I dropped it and accepted 79 over a hollow 84.

**Abandoned edit two: rewriting a five-year-old bullet.** I spent two minutes polishing a bullet in my oldest role to squeeze in "roadmap." The rescore did not move, because "roadmap" was already covered in my current role, and a recruiter's six-to-seven-second scan never reaches that far down. The edit changed neither the score nor the read, so I reverted it. That is the rule: if an edit moves neither the number nor the top-of-page read, it is polish, and polish costs applications.

> **Watch out:** A perfect score is a red flag, not a finish line
>
> A 100 percent match usually indicates keyword stuffing, and a cited 2025 survey found 76 percent of recruiters prefer resumes that use keywords naturally. The same edit that maxes the scanner can sink the human read. Stop in the 75 to 85 band.

Time matters because it bounds how many jobs you can reach. A professional resume writer puts good tailoring at 10 to 15 minutes once you have a rhythm; Indeed cites 10 to 20 minutes once a master resume exists. Deep manual customization runs 30 to 45 minutes and naturally means applying to fewer jobs. Yet skipping tailoring is cited as forcing 3 to 5 times more applications for the same number of interviews. So the target is the middle: about 15 minutes, spent on the load-bearing lines, edits abandoned the moment they stop moving the signal. This is exactly the friction [Refolk](/candidates) removes: it writes the resume from your own history, tailors it to each posting, and scores the fit, so the 15 minutes goes into judgment rather than retyping.

## How this goes wrong, and how each failure surfaces

Most tailoring failures are false positives: an edit feels like progress while quietly hurting the draft. Each has a specific tell and a specific check. Learn the tells and you stop shipping drafts that pass the machine and fail the human, or vice versa.

| Failure mode | The false positive | The check |
|---|---|---|
| Chasing score past the band | 95%+ feels like winning | Does it still read like a person? Stop at 75-85 |
| Trusting one scanner | 92 on one engine reads as done | Run more than one scanner if the role matters |
| Synonym drift | You covered it in your own words | Search the draft for the posting's literal string |
| Claiming a skill you lack | The score jumps | Point to a proof bullet, or drop the term |
| Hidden-text tricks | Invisible keywords lift the machine read | Paste as plain text and see what surfaces |
| Over-editing weak lines | An hour of polish feels thorough | Did it move the score or the top-of-page read? |

**Synonym drift** is the most common silent failure. Many ATS do not reliably resolve synonyms or abbreviations, so a semantically correct paraphrase scores zero on that term. If the job says "project management," writing only "program coordination" may not register. The fix is literal mirroring plus acronym-and-spellout pairs, like "Search Engine Optimization (SEO)."

**Trusting one scanner** hides real risk because engines disagree. One vendor reported the same resume scoring 84, 71, 92, 68, and 79 across five systems. A resume scoring in the 80s across several engines is more reliably filter-friendly than one scoring 95 on one and 60 on another. Chase consistency, not a single peak.

**Hidden-text tricks** deserve a flat no. Many systems detect text whose color matches the background or fonts below a readable threshold, and flag the document. The gain is imaginary and the flag is real.

#### Is this edit worth keeping?

Horizontal axis runs from You cannot prove it to You can prove it. Vertical axis runs from Does not move the signal to Moves the signal.

| Quadrant | What it means |
| --- | --- |
| Unprovable, no lift | Delete without hesitation |
| Provable, no lift | Revert, it is only polish |
| Unprovable, lifts score | Stuffing, drop it before the interview |
| Provable, lifts signal | Keep it, this is the work |

*Keep an edit only when it moves the signal and you can defend it; abandon everything else.*

The bottom-left quadrant, provable but no lift, is where honest candidates waste their time. The top-right, a lift you cannot prove, is where the score-chasers get caught. The only edit worth keeping is provable and moves the signal.

## Before you submit, and how to keep this repeatable

Run this checklist against your draft before you hit submit. It is the same seven-step method compressed to what you verify.

#### Submit-ready check

- [ ] The posting's exact job title appears in the title line or summary
- [ ] Every Tier 1 required term is present as a literal string, not a paraphrase
- [ ] Every claimed skill points to a real proof bullet
- [ ] Each acronym is paired with its spelled-out form at least once
- [ ] The top two or three bullets per role carry Tier 1 terms in context
- [ ] The match score sits in the 75 to 85 band on at least one scanner
- [ ] A plain-text paste surfaces no hidden or stuffed keywords
- [ ] Total edit time stayed near 15 minutes and every edit moved the score or the top-of-page read

To keep this repeatable across many postings, save your master resume as the source of truth and treat each tailored draft as disposable. Do not tailor a tailored resume; always start from the master, or your terms drift toward the last posting and away from this one. When you notice the same required term appearing across most of your target postings, that is an industry-wide standard worth building into the master itself; a term unique to one posting stays in that posting's draft only.

When the volume climbs past a handful of applications a week, the 15-minute manual pass becomes the bottleneck. That is the point to let Refolk carry the mechanical part: extracting and ranking each posting's terms, mirroring them against your real history, and scoring the fit, so you spend your minutes on the two judgment calls that no tool makes for you, which claims you can defend and which edits you abandon.

Ask me this: `Product managers in the US who list both Salesforce and SQL and worked at a company under 200 people.` - [run the search](https://www.refolk.ai/start?q=Product%20managers%20in%20the%20US%20who%20list%20both%20Salesforce%20and%20SQL%20and%20worked%20at%20a%20company%20under%20200%20people.).

*Returns the small set of PMs who actually pair a rare tool with a common one at small companies, which is the profile the worked posting was screening for.*

## Frequently asked questions

### What resume match score should I aim for before I submit?

Aim for 75 to 85 percent keyword coverage of the required qualifications. Below 40 percent a resume is likely filtered before a recruiter sees it, and published cutoffs cluster between 60 and 80 percent. Above 85 percent returns diminish sharply and a 100 percent score usually signals keyword stuffing, which backfires when a human reviews. The 75 to 85 band clears most filters while still reading like a person wrote it.

### How many keywords should I match from one job posting?

There is no fixed number, but most postings contain 15 to 25 key terms worth matching, and most tailored resumes carry 10 to 20 of them. Cover at least 70 to 80 percent of the required skills and tools, not every keyword but the priority ones. Use each high-priority term two to three times: once in your skills section and once or twice in context inside a bullet.

### Which resume bullets should I rewrite for a specific posting?

Rewrite the load-bearing lines: the summary, the skills section, and the top two or three bullets under each role. Those are the zones a scanner weights and a recruiter reads in the first six to seven seconds. Leave lower bullets and older roles alone unless a Tier 1 term has no home anywhere else. Editing low-leverage lines burns time without moving the score or the human read.

### How do I rank a job posting's keywords by importance?

Use a two-tier read. Tier 1 is the exact job title plus everything in the required or must-have qualifications, plus any term repeated across the posting. Tier 2 is the preferred and nice-to-have section. Within a tier, weight terms that repeat and terms that are rare across postings, since a rare required tool differentiates you more than a term nearly everyone lists.

### Why did my resume score high on one scanner but low on another?

Different engines parse and weight terms differently, so one number is unreliable on its own. One vendor reported the same resume scoring 84, 71, 92, 68, and 79 across five systems. A resume scoring in the 80s across several engines is more reliably filter-friendly than one scoring 95 on one and 60 on another. If the role matters, run more than one scanner and treat the lower numbers as the truth.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/resume-tailoring-teardown*