How to hire a paralegal: a scoring rubric for small law firms

I placed people in law firms for years, and the most common hiring mistake I saw wasn't a bad candidate. It was a good candidate judged on the wrong thing: the name of the last firm, the size of the last caseload, the confidence in the interview. A bad hire in a small firm shows up months later as missed deadlines, and by then the firm has paid for it twice.

It's also a hire most firms make more often than they'd like. The Bureau of Labor Statistics projects about 39,300 openings a year for paralegals and legal assistants against essentially flat employment, so nearly every opening is a replacement for someone who left. The median pay is $61,010 a year as of May 2024 (BLS Occupational Outlook Handbook).

Write the rubric before you read a résumé

Most firms carry the hiring bar in the head of whoever is hiring that week. That's why the same firm hires very differently in March and in August. The fix is unglamorous: before the first résumé, write down four to six things the person must be able to do in their first months, and what strong evidence of each looks like.

The criteria come from the work, not from the job ad. For a litigation paralegal, think about the last three times something went wrong on a matter, and what the person in the seat would have needed to prevent it.

A worked paralegal rubric

Here is a rubric for a litigation paralegal in a small firm, with anchors for a 1, a 3 and a 5. Copy it, then change the criteria to fit your practice.

1. Deadline and calendar control

• 5: Owned the calendar for named attorneys, and describes the system: deadlines calculated from the court rules, a second check, a weekly docket review.
• 3: Lists calendaring among their duties, with no detail of how it was done.
• 1: Nothing about deadlines anywhere.

2. Catching errors

• 5: Describes a specific error they caught before it left the office, and what changed afterwards.
• 3: Mentions proofreading or quality checks in general terms.
• 1: Only self-description, such as "detail-oriented".

3. Knowing when to escalate

• 5: Gives an example of taking something to the attorney and an example of handling something without them, and can explain the line between the two.
• 3: Describes working closely with attorneys, without examples of judgment.
• 1: No sign of independent decisions in either direction.

4. Practice-area software

• 5: Has used practice-management or document-review software for the same kind of work your firm does.
• 3: Has used comparable tools in a different practice area.
• 1: No software named.

5. Client communication

• 5: Handled client updates directly, including difficult ones, and says how they kept the attorney informed.
• 3: Some client contact, mostly scheduling.
• 1: No client-facing work.

An example score, and a capped one

A score without a reason is an opinion. Write the evidence next to every score, and two things happen. Your scores get more consistent, because you can see when you're grading on vibes. And the next person to look at the shortlist can check your work instead of trusting it.

• Deadline and calendar control: 5. "Managed the litigation calendar for three trial attorneys, including court-rule deadline calculation and a weekly docket review."
• Catching errors: capped at 3, no evidence. The résumé says "detail-oriented team player" and nothing more. A strong résumé that never shows the skill hasn't shown you the skill, so it can't earn more than the middle score. The interview is where this candidate gets the chance to raise it.

Interview questions that test the rubric

Generic paralegal interview questions are written for candidates rehearsing answers. Write yours for the interviewer: one question per criterion, aimed at the evidence the résumé didn't give you.

• Deadlines: "A scheduling order comes in on Friday afternoon. Walk me through what you do with it, and who checks your work."
• Errors: "Tell me about a mistake you caught in someone else's work. What did you do next?"
• Escalation: "When did you last decide not to bother the attorney with something? Were you right?"
• Software: "Show me how you'd find every document a particular witness sent, in the last system you used."
• Clients: "A client calls upset that nothing has happened on their case. What do you say before you've spoken to the attorney?"

For accuracy, a short skills test beats any question. A one-page draft with three planted errors and a ten-minute limit tells you more than an hour of conversation.

What a mis-hire costs, in your own numbers

You'll find a lot of confident figures for the cost of a bad hire. The popular "30% of first-year salary" can't be traced to a current primary source, so don't plan around it. Do the arithmetic on your own figures instead:

• Salary for the weeks before you notice. At the BLS median, that's about $1,170 a week.
• Attorney hours spent re-checking work, at your billing rate.
• The second search: posting, screening, interviews, and the weeks the seat is empty again.
• Whatever slipped: a missed deadline, an unhappy client, a write-off.

Most firms that do this sum once never skip the rubric again.

Don't start every search from zero

Most firms have already met their next hire. They applied for a different role last year, or came second in the last search. Keep your past applicants, scored against the rubrics you used, and look there first.

A note on AI screening and the law

If you use any AI tool to screen applicants, know that the law is moving. Illinois amended its Human Rights Act, effective January 1, 2026, to require employers to give notice when they use AI in employment decisions. New York City's Local Law 144 requires a bias audit and notice for automated tools that substantially assist hiring decisions. One law firm's summary warns that keeping a human as the final decision-maker doesn't by itself take a tool outside these laws (Shipman & Goodwin). Rules differ by state and city, and they change. Ask employment counsel before you start, and treat any vendor's compliance claim, ours included, as something to verify.

How we handle it

This is the job Sara does in TalentOS. You set the role and the rubric. She scores every applicant 1 to 5 on each criterion, quoting the résumé line each score relies on. She checks each quote word for word against the résumé, and when no quote survives, the score is capped at 3. She explains why a candidate fits the role or doesn't, citing their profile.

She keeps a pipeline for every role, which you move people through, and rediscovers past applicants who fit the role you're filling now. She can write the role for you to post to the TalentOS job board in one click, where it goes live after a review. She proposes interview times or sends the invite once you approve it, and writes the interview assessment from an interview recorded in Chrome or Edge.

She doesn't decide who you interview or hire, and nothing goes to a candidate without your approval. Sara is one worker at $500 a month.

The test

Take the last three people you hired and score them against a rubric written today. If the scores surprise you, the rubric is doing its job.

See how Sara fits a law firm, what each worker does, and what it costs.