Updated October 2026
Candidate ranking scores each applicant against your open job and puts the strongest matches at the top, so you know whose resume to read first when applications pile up. In Curriculo, every candidate gets an Impact Score from 0 to 100 with written reasoning. You still decide who advances.
What is the most effective way to rank candidates at scale?
Score every applicant on the same evidence against the same job requirements, then read the list from the top. A consistent rubric gives the 200th application the same review standard as the first. Use the score to decide whom to read first, while a person decides whom to interview, advance or reject.
SmartRecruiters Recruitment Benchmarks 2025 reports an average of 73 applicants per role, with 3 interviewed and 1 receiving an offer, based on 89 million applications from September 2023 to August 2024.
Some roles draw far more. On our calls, a UK recruiter described "about 350, if not 400 for one job". A solo agency recruiter running job ads said a role can bring "anywhere from like 250 to 1000 applicants".
A startup recruiter told us their current tool makes you "go one by one" with no bulk view. Their biggest fear with AI was rejecting "an otherwise great candidate". Curriculo keeps every applicant visible and gives each one a score and written explanation. A person still makes every hiring decision.
How do algorithms rank top candidates?
Ranking systems compare resume evidence with job criteria and assign scores based on the product's rules or model. Curriculo's Resume screening agent provides a score with written reasoning for every candidate. Recruiters review that reasoning, correct mistakes, and make the final decision about who moves forward.
Candidate ranking systems compare each application with the job, assign a score and sort the pool by score. Some methods count exact terms. Others compare meaning or assess evidence against stated qualifications.
A global talent acquisition leader asked us the question every ranking page should answer: "How do you come up with the 80% match? So, like, what's the criteria?" The percentage was the speaker's example, but the question applies to every vendor. Buyers should know what data enters a score, what the system returns and what remains a human decision.
Ranking method comparison
| Method | What it uses | What it misses | How explainable it is | Common failure |
|---|---|---|---|---|
| Keyword filter | Exact words and phrases | Related terms expressed differently | High when rules are visible | Qualified applicants can be hidden because they used different wording |
| Boolean search | Recruiter written AND, OR and NOT rules | Evidence outside the chosen terms | High for people who can read the query | A narrow query excludes relevant resumes |
| Knockout questions | Applicant answers to fixed questions | Context and evidence elsewhere in the application | High | A rigid answer can remove someone who needs human review |
| Embedding similarity | Mathematical similarity between job and resume text | The reason a particular requirement was met or missed | Low to medium | Similar wording can rank above stronger evidence |
| Curriculo language model scoring | Full resume, job details, qualifications and applicant submitted information | Information the applicant did not provide | Written summary and requirement checklist | An incorrect job description produces scores against the wrong criteria |
Curriculo ATS scores every candidate from 0 to 100 against your job, with a written AI summary of the fit.
The score is one AI judgment per candidate. The checklist marks each must-have and preferred qualification as met, partial or missing, with a written reason.
What data do AI candidate scoring tools use to rank applicants?
Curriculo reads the full parsed resume, the job details and what the applicant submitted. Job details include the title, description, must-have qualifications, preferred qualifications and location. Applicant information can include a cover letter, expected salary, stated years of experience and preferred locations.
Location is weighed on purpose. The score reads the whole resume, so it is not blind and does not exclude names or other resume content before scoring. The AI provides advice. A person decides what happens next, and candidates receive an AI notice and can ask for a human-only review.
Before you trust a ranking score
Ask each vendor:
- Can you read a written reason for the score?
- Does the score use the job's actual must-have and preferred requirements?
- Are names and photos excluded, or does the model read the full resume?
- Can the system reject anyone without a person reviewing the decision?
- Can you recalculate scores after editing the job?
- Has the specific system been tested for bias, and can the vendor provide the method and results?
Curriculo does not claim names or photos are excluded, that bias testing has passed or that a score should replace human review.
Ranking versus filtering
Filtering applies rules that can hide or remove applicants before a person reads them. Ranking keeps the pool visible and changes its order.
Curriculo scores each candidate against the job, then places higher scores nearer the top. You can open any application, read its summary and inspect how each qualification was marked. Ranking does not draw the interview cutoff for you.
Applicants beyond a plan's ranking limit remain stored in the ATS. They stay locked and unscored until you upgrade, at which point they are scored. Starter includes 1,000 ranked candidates with no time window. Pro includes 10,000.
What ranked results look like
Illustrative example
| Rank | Candidate | Score | Written summary | Qualification checklist |
|---|---|---|---|---|
| 1 | Candidate A | 91 | Strong evidence for the required role scope and preferred industry experience. | Met: required experience and core skill. Partial: preferred certification. |
| 2 | Candidate B | 83 | Meets the main experience requirements, with limited evidence for one preferred qualification. | Met: required experience. Partial: preferred industry experience. Missing: preferred certification. |
| 3 | Candidate C | 68 | Relevant background is present, but the resume does not establish the required scope. | Met: core skill. Partial: required experience. Missing: preferred industry experience. |
These labels and scores explain the interface rather than representing real candidates or customer results. Your list uses the evidence in each submitted application and the requirements entered for your job.
How do I rank resumes automatically?
- Create a job with a title, description, must-have qualifications, preferred qualifications and location.
- Publish it on your hosted careers page or receive applications through the Email agent.
- Let the Resume screening agent score each application as it arrives.
- Open the job and review candidates in score order, including each written summary and qualification checklist.
- Advance, interview or reject candidates after human review.
You can also rank candidates from any MCP client, including Claude or ChatGPT. On a demo with an agency team that works in ChatGPT, we showed the ATS through MCP inside ChatGPT.
Editing a job description in the app recalculates scores in the background, one candidate after another. Edits through the API or MCP do not start recalculation. Once a Starter job has candidates, you can edit its title and description twice.
How do you know if you are a top applicant?
On the employer side, top applicants provide clear evidence for the job's main requirements. Relevant responsibilities, results with numbers and comparable scope can help a reviewer understand the fit. A ranking score still depends on the specific job and the information submitted.
If you are applying for jobs, reflect the posting's requirements honestly in your summary and experience bullets. These resume summary examples show ways to state your experience clearly without copying claims you cannot support.
Plans for candidate ranking
| Plan | Price | Ranked candidates | Team members |
|---|---|---|---|
| Starter | Free | 1,000 | Unlimited |
| Pro | $50 per month at the early-bird price, 50% off $100 | 10,000 | Unlimited |
| Enterprise | Custom pricing | Talk to us | Talk to us |
Starter requires no credit card. You can sign in with Google or Microsoft. There are no per-seat fees.
Start free on Starter
Rank your first 1,000 applicants free.
Related: [ai resume screening](/ai-resume-screening/), [candidate ranking](/glossary/candidate-ranking/), and [best free ats](/best-free-ats/).FAQ
What is a candidate ranking system?
Software that orders applicants by how well they fit a job, usually with a numeric score and a short written reason for it.
What is an applicant ranking system?
The same idea under another common name. It compares each person with the role itself, not with a keyword list alone.
How is ranking different from keyword filters?
A keyword filter checks whether certain words appear and hides resumes that miss them. Ranking orders every applicant against the whole job, so a strong candidate who phrased a skill differently still shows up. In Curriculo, every profile stays available whatever its score.
Do recruiters still decide after ranking?
Yes. Scores tell you where to start reading. Hiring decisions stay with people, and you can open any profile at any score.