---
title: Write the bar down before you read the first CV
description: A shortlisting rubric written at intake is faster to apply, survives a second recruiter, and is the only version of your judgement that still exists later.
---

[Echo Recruit blog](https://www.echorecruit.com/blog)

# [Write the bar down before you read the first CV](https://www.echorecruit.com/blog/write-the-bar-down-before-you-read-the-first-cv)

 Written by [Echo Recruit](https://www.echorecruit.com/blog/author/echo-recruit) | Sep 21, 2026, 6:43:35 AM

*The bar exists on every role you work. The only question is whether it is written down or living in somebody’s head, moving quietly as the week gets harder.*

In short

- Unwritten bars drift: the first twenty CVs get judged harder than the last twenty, or softer, depending on the deadline.
- Rules come in three kinds — knockout, scored and manual — and mixing them up is the most expensive mistake in screening.
- A requirement the CV cannot prove must be excluded from the arithmetic, not scored as zero. Otherwise you are penalising candidates for what they left off the page.
- Two roles at the same client need two rubrics. Version them, because the bar will move in week three.

Ask a recruiter what the bar is on a role and you will get a good answer. Ask them again three weeks later and you will get a different good answer. Neither is dishonest. The bar was never written down, so it has been quietly renegotiating itself against the number of CVs in the inbox and the number of days left.

This matters for three separate reasons, and they are worth separating because they point at different fixes.

- **Speed.** An unwritten bar has to be re-derived for every CV. That is the part of screening that feels like work.
- **Consistency.** Two recruiters on the same role are running two different processes, and the candidate’s outcome depends on who opened the file.
- **Defensibility.** Six months from now, somebody will ask why a particular candidate was rejected. “They were not a fit” is a sentence, not a record.

## Get it out of the hiring manager at intake

The rubric is not written by the recruiter guessing. It is the output of the intake call, and it is the reason the intake call is worth an hour.

The question that produces it is not “what are the requirements?” — that gets you the job description, which is a list of everything anyone could possibly want. The question is comparative:

> “Two candidates arrive. They are identical except that one has eight years and the other has four but shipped the exact thing you are building. Which one do you want to see first — and if I could only send you one, which?”

Ask that four or five times, varying the pair, and you have the weights. You will also find out which “must-haves” are real: a manager who trades away a must-have in order to keep something else has just told you it was a preference.

## The three kinds of rule

Every rule you write is one of three things, and the whole discipline is in not confusing them.

### Knockout

Fail it and the candidate is out, whatever else is true. Reserve these for things that genuinely end the conversation: the right to work in the country, a licence or certification the job legally requires, a hard location constraint on a role that cannot be done remotely.

The mistake is promoting ordinary preferences to knockouts. “Must have 7 years” as a knockout throws away the person with six who would have got the job, and it does it silently, before anyone sees them. A knockout should be a rule you would defend out loud to the hiring manager after they have met the candidate you rejected.

### Scored

Contributes weight toward a percentage. This is where most requirements belong: specific skills, depth of experience, domain, the shape of the last two roles. Weights should total 100 so that a score means something absolute rather than relative to whatever rules happen to be present.

### Manual

The rule nobody can check from a CV. Compensation expectation. Notice period. Appetite for on-call. Visa status where you cannot infer it. Willingness to be on site three days a week.

**These must be excluded from the arithmetic entirely** — from the numerator and the denominator — and turned into screening questions instead.

This is the rule that gets broken most often, and it breaks quietly. If “salary expectation within band” is worth 10 points and the CV does not mention salary, a naive scorer gives 0 out of 10. The candidate has now been marked down for not printing their salary expectation on their CV, which nobody does. Do that across five manual rules and the people who get through are the ones who over-share, not the ones who can do the job.

## A worked rubric

A senior controls engineer for a robotics client. Salary band agreed, on-site four days, starting within eight weeks.

| Rule | Kind | Weight |
| --- | --- | --- |
| Right to work in the country, without sponsorship | Knockout | — |
| Within commuting distance of the site, or relocating | Knockout | — |
| Motion control or servo systems in production | Scored | 30 |
| Five or more years in controls engineering | Scored | 25 |
| C or C++ on embedded targets | Scored | 20 |
| Has owned a system from prototype to production | Scored | 15 |
| Robotics or automation domain | Scored | 10 |
| Compensation expectation within band | Manual | Question |
| Can start within eight weeks | Manual | Question |
| Comfortable on site four days a week | Manual | Question |

Five scored rules totalling 100, two knockouts, three questions for the screening call. Now set two thresholds rather than one: a score at or above which the candidate is shortlisted, and a lower one above which they are worth a human look. The band between them is where your judgement is actually needed, and it should be the only pile you read carefully.

## Five habits that keep a rubric honest

1. **1One rubric per requisition, never per client**The same client’s controls engineer and firmware engineer are different jobs with different bars. Sharing a rubric across a client account means the second role is screened against the first role’s priorities, and the candidate who is perfect for one scores badly on the other for no reason anyone can explain.
2. **2Write the rule as something checkable**“Strong communicator” is not a rule; it is a feeling with a weight attached. If you cannot say what evidence would satisfy it, it belongs in the interview scorecard, not the screening rubric.
3. **3Version it, and record which version judged whom**The bar moves in week three — it always does, usually because the market answered. That is fine as long as re-scoring adds a new record rather than overwriting the old one. Otherwise you have quietly rewritten the reason a candidate was rejected in week one.
4. **4Score the existing pool before sourcing anything new**You already have the rubric and you already have a database. Run one against the other first. The cheapest candidate you will ever place is one you sourced for something else eighteen months ago.
5. **5Show the breakdown, not just the number**A candidate at 72% tells you nothing. 72% made of full marks on three rules and zero on the one the client cares most about is a rejection; 72% made of near-misses everywhere is a phone call. If your tooling only shows the total, it is hiding the part you need.

## What a rubric is not

It is not a decision-maker. A rubric ranks and flags; a person decides. The moment a score is allowed to reject somebody automatically you have built something quite different, with obligations attached to it in a growing number of jurisdictions — and you have also made your process worse, because the interesting candidate is almost always the one the rules did not anticipate.

It is also not a substitute for reading. It is a way of deciding *what* to read carefully, so that the two hours you have go to the twelve CVs in the band where judgement changes the outcome instead of the ninety where it does not.

In Echo Recruit

Rubrics, per requisition and versioned

Every job carries its own sourcing brief and its own shortlisting rubric — nothing is shared across roles at the same client by accident. Rules are knockout, scored or manual, and manual rules are excluded from both the numerator and the denominator and surfaced as screening questions instead of counting as failures.

Rubrics are versioned and every evaluation records the version that produced it, so changing the bar in week three never rewrites what was decided in week one. A criteria builder drafts a full rubric from what you tell it about the role, with a rationale per rule, which you edit before it applies.

[View full post](https://www.echorecruit.com/blog/write-the-bar-down-before-you-read-the-first-cv)

```json
{
  "@context" : "http://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "Echo Recruit"
  },
  "dateModified" : "2026-09-21T06:46:45.327Z",
  "datePublished" : "2026-09-21T06:43:35Z",
  "headline" : "Write the bar down before you read the first CV",
  "image" : {
    "@type" : "ImageObject",
    "height" : 60,
    "url" : "/hs/hsstatic/content_shared_assets/static-1.4092/img/default-amp-logo.png",
    "width" : 60
  },
  "mainEntityOfPage" : "https://www.echorecruit.com/blog/write-the-bar-down-before-you-read-the-first-cv",
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject",
      "height" : 60.0,
      "url" : "https://www.echorecruit.com/hubfs/echo-recruit-avatar-green-400.png",
      "width" : 60.000004
    },
    "name" : "Echo Recruit blog"
  }
}
```