Benchmark Data: training AI matching with sample resumes

What Benchmark Data is in CVViZ

Benchmark Data in CVViZ is not market salary data or industry comparisons. It's a tab on the job edit form where you upload sample resumes that represent your ideal candidate. CVViZ uses these to improve the AI matching it runs against the job's incoming applicants.

The section is titled Benchmark Resumes, with the description: "Resumes already linked to this job. Remove any that no longer represent your ideal candidate."

Where to find it

Open the job and go to the Benchmark Data tab.

How it improves matching

CVViZ ranks every incoming candidate against the job using AI. Benchmark resumes give that model concrete examples of "what a good fit looks like" for this specific role — a stronger signal than the job description alone.

Typical use:

• For a senior backend role, upload CVs of two strong engineers you've already hired or interviewed and would have gladly hired.

• For a niche role with hard-to-describe attributes, upload CVs that capture the pattern.

Adding benchmark resumes

1. Open the job's Benchmark Data tab.

2. Upload sample CV files (the same formats supported elsewhere in CVViZ — PDF, DOCX, etc.).

3. Save.

Files are stored under your account at — you don't need to think about that path, but it's where they live.

Reviewing existing benchmarks

The Benchmark Resumes panel shows every linked benchmark file with:

• The file name (clickable — opens a viewer modal titled "Benchmark resume: {filename}" with a tooltip "View benchmark resume").

• A delete button per file with the tooltip "Remove benchmark resume".

Removing a benchmark

Click the delete icon next to the file and save. The Benchmark Data tab tracks the change in and updates the model on save.

When to update benchmarks

• After you hire — add the hired candidate's CV as a benchmark for similar future roles.

• When the bar shifts — if your standards for a role change, swap out old benchmarks.

• Periodically — review benchmarks every few months; remove anyone who no longer represents your ideal.

What this isn't

To avoid confusion: this tab does not show salary benchmarks, time-to-fill comparisons, or any external market data. CVViZ doesn't have that feature in this tab.

Tips

• Quality over quantity. 3 strong benchmarks beat 15 mediocre ones.

• Don't include CVs of clearly weak candidates "for contrast" — the model treats every benchmark as positive.

• Refresh after each successful hire to keep the signal current.