Overview

Lecture URL: https://skill.re/learn/recruiting/data-minimization-collecting-only-whats-necessary.php

TRANSCRIPT: Data Minimization: Collecting Only What's Necessary

Course: AI for Recruiters - Professional Credential

Module: Level 3: Independent Practice

Section: Chapter 15 -- Privacy Discipline And Data Handling

Theme: privacy-discipline-and-data-handling

Lecture: 15.2

Duration: 75 min

Format: Workshop + Case Studies

Audience: Experienced recruiters applying AI independently

Prerequisites: L2 Certification

What you will learn: Apply data minimization principles to collect only information truly necessary for recruiting decisions, reducing privacy risk and improving candidate experience.

Data minimization is simple in principle: collect only what you need. In practice, it's tempting to collect more. "We might use this data sometime." "Better to have it than not." "Other companies collect this."

But over-collection creates burden. You have privacy obligations for every data point. You have to track, secure, and eventually delete it. Candidates worry about what you might do with their data. And you're still making decisions based on a small fraction of what you collected.

This session focuses on practical data minimization: identifying what you actually need, eliminating what you don't, and building recruiting processes around necessity rather than convenience.

• *The Data Inventory Process**

Start by listing everything you currently collect:

• Resume (required)
  - Cover letter (optional but requested)
  - Work samples or portfolio (sometimes)
  - Background check (employment, criminal, education verification)
  - References (usually 2-3)
  - Assessment results (coding test, behavioral assessment, etc.)
  - Interview notes
  - Social media information (you look at LinkedIn, maybe Facebook)
  - Candidate communication (emails, chat messages)
  - Salary history or expectations
  - Demographic information (if you collect it)

Now ask, for each: *Do I actually need this to make a hiring decision?*

Resume: Yes. Essential. You need to know their experience.

Cover letter: Maybe. Some roles benefit from writing samples. Others don't. If you're not actually reading cover letters, stop asking for them.

Work samples: Depends. If the role requires certain skills, work samples are valuable. A designer should have a portfolio. An accountant should have work samples. A marketer should have writing samples. For roles where you're not evaluating specific outputs, work samples are less critical.

Background check: Depends on the role. For finance or security roles, background checks are standard. For most tech or creative roles, are criminal background checks actually necessary? Does employment verification predict performance? Does education verification matter? You might keep criminal checks but eliminate education verification.

References: Probably not as you're using them. Most references are self-selected people who will say positive things. You're spending time on this when the predictive value is low. Consider: do you actually call references, or do you have a policy you don't follow?

Assessment results: Only if they predict job performance. Before giving anyone an assessment, ask: "What am I trying to measure? Does this assessment measure it accurately? Does it predict job performance?" Many assessments fail the third question.

Interview notes: Yes. You need to document your evaluation for legal reasons.

Social media research: Probably not. What are you looking for? If you're checking to verify their resume matches their LinkedIn, okay. If you're trying to assess "culture fit" by looking at their Twitter? That's fishing for demographic information.

Salary history: Maybe not. In many jurisdictions, asking for salary history is illegal or restricted. Even where it's legal, does it actually help you make a hiring decision? Or does it just anchor your offer to their previous employer's underpayment?

Demographic information: Only collect if you're legally required to (for reporting purposes) and only if you're actually using it for non-discriminatory analysis.

• *The Elimination Decision Framework**

For each data category, use this framework:

1. Necessity: Is this data necessary for my hiring decision? (Not "nice to have." Necessary.)
2. Predictiveness: Does this data predict job performance? (Not "might be relevant." Actually predict.)
3. Risk: What's the privacy and compliance risk of collecting this? (Salary history might expose pay inequity concerns. Social media might expose bias risk.)
4. Burden: What's the candidate burden of providing this? (Cover letters require writing time. Assessments require time. References require outreach.)
5. Decision: Given necessity, predictiveness, risk, and burden, should I collect this?

Apply it to salary history: Is it necessary? No--you can set compensation based on the role. Does it predict performance? No. What's the risk? High--it can expose your own salary inequity and might be illegal. What's the candidate burden? Moderate--they have to look it up. Verdict: Don't collect.

Apply it to portfolio: Is it necessary (for design role)? Yes. Does it predict performance? Yes. Risk? Low. Burden? Moderate--they need to assemble it. Verdict: Collect.

• *Reducing Assessment Load**

Assessments have become ubiquitous. Coding assessments, personality assessments, skills assessments, culture-fit assessments. But ask: are you actually using all of these?

Many organizations give multiple assessments (coding + personality + communication assessment) without asking whether they're adding value. A candidate spends 3 hours on assessments for a role that might take 30 minutes to interview for.

Better approach: which single assessment (if any) is actually predictive of performance? Use that one. Eliminate the rest.

For technical roles, coding assessments make sense. They measure actual ability. For non-technical roles, coding assessments don't make sense. For roles requiring communication, a conversation (interview) is often more predictive than an assessment.

• *Data Retention Minimization**

Data minimization extends to retention. How long should you keep data?

Resume and basic information: Until the role is filled and there's no legal requirement to keep it. Typically 6-12 months.

Assessment data: Until the hire is made. Typically weeks or months. Assessment data for rejected candidates doesn't need to be kept.

Interview notes: Longer. You might need these for legal reasons if someone claims discrimination. Keep for at least 1-2 years. After that, delete.

References: Delete after hire. Reference notes aren't legally required after the decision is made.

Background check results: Keep as required by law. In most places, you keep employment verification for at least 1-2 years. Criminal records might be kept longer. Check your jurisdiction.

Social media information: Delete immediately after hire decision. There's no reason to keep this.

Rejected candidate data: Most jurisdictions don't require you to keep this. If you delete after the role is filled, be consistent.

• *Building Minimization Into Process**

To make minimization a practice:

1. Make it explicit. Document what data you collect and why. Review quarterly.
2. Get leadership buy-in. Some leaders worry that less data means worse decisions. Show them evidence. Good decisions are based on relevant data, not volume.
3. Train your team. Make sure everyone knows what data you're collecting and why.
4. Audit your tools. Many recruiting tools collect and store more data than you realize. Understand what your ATS is storing. Understand what your assessment vendors are keeping.
5. Communicate to candidates. Tell candidates what data you're collecting. This holds you accountable to minimization principles.

ANTI-PATTERNS

• *Anti-Pattern 1: The Comprehensive Backup Strategy**
  - Description:* Collecting data "just in case" you need it later, without clear use cases. *Why:* It feels prudent. You don't know what you might need. *What goes wrong:* You create privacy and storage burden. Candidates worry about what you're doing with their data. *How to avoid:* Collect only what you know you'll use. If you don't have a specific use case, don't collect it.
  - *Anti-Pattern 2: The Assessment Pile-On**
  - Description:* Giving multiple assessments (coding + personality + communication) when a single assessment (or none) would suffice. *Why:* You want comprehensive evaluation. Each assessment seems valuable. *What goes wrong:* Candidates get fatigued. You don't actually use all the data. Decision quality doesn't improve. *How to avoid:* For each assessment, ask: "Is this predictive of performance?" Keep only those that are.
  - *Anti-Pattern 3: The Proxy Collection**
  - Description:* Collecting data as a proxy for what you actually want to assess. Salary history as proxy for value. Social media as proxy for culture fit. *Why:* The proxy feels easier to measure. *What goes wrong:* The proxy creates bias or legal risk. It's often not actually predictive. *How to avoid:* Collect what you actually need to assess, not proxies. Want to assess communication? Use an interview. Want to assess culture fit? Define what that actually means and assess it directly.

PRACTICE PROMPTS

1. Data Inventory Exercise: List everything you currently collect from candidates. For each, write one sentence explaining why you collect it. If you can't write that sentence, consider eliminating it.
2. Predictiveness Audit: For three data categories you collect, write down: What job performance outcome does this predict? If you can't confidently answer that, the data probably isn't necessary.
3. Assessment Evaluation: If you use pre-screening assessments, evaluate each one. Is it predictive of job performance? Are you using the results, or just going through the motion?
4. Candidate Burden Calculation: Calculate how long it takes a candidate to go through your recruiting process--resume, cover letter, assessments, interviews. If it's more than 4 hours for a non-executive role, consider what you can eliminate.
5. Retention Policy Review: For each category of data you keep, write down: How long do we keep this? Why? What's the legal requirement? Eliminate retention periods longer than necessary.

KEY TAKEAWAYS

1. Data minimization protects candidates and simplifies your process. Less data to manage, less privacy risk, faster decisions.
2. Necessity, not convenience, should drive collection. Ask "do I need this?" not "might this be useful?"
3. Predictiveness matters. Collect data that actually predicts job performance. Discard data that doesn't.
4. Assessment load should be proportionate. A short coding assessment for technical roles makes sense. A 3-hour assessment battery for any role is excessive.
5. Retention should match legal requirements, not habits. Review how long you're keeping data and why. Delete what's no longer necessary.
6. Candidates notice and appreciate minimization. When you ask for less data and explain why, candidates feel respected.

GLOSSARY

• *Data Inventory:** A catalog of all data categories you collect from candidates.
  - *Necessity Test:** Asking whether data is actually required for hiring decisions (not just nice-to-have).
  - *Predictiveness:** Whether data actually predicts job performance.
  - *Proxy Data:** Data collected as a proxy for what you actually want to assess. Often introduces bias.
  - *Disparate Impact:** A seemingly neutral data requirement that systematically disadvantages certain groups.
  - *Retention Period:** How long data is kept before deletion.
  - *Assessment Battery:** Multiple assessments given to candidates. Should be evaluated for cumulative value.

[SYNTHESIS AND APPLICATION]

Data minimization isn't just about privacy compliance. It's about decision quality. The best hiring decisions come from relevant data, not volume.

When you collect only what you need, you're forced to think carefully about your evaluation criteria. You reduce noise. You focus on signal. And candidates appreciate that you respect their time.

Minimize ruthlessly. You'll make better decisions faster.

[REFLECTION EXERCISE]

1. What's one data category you collect that you've never actually used to make a hiring decision?
2. If you had to defend your current data collection practices to a privacy auditor, where would you feel most exposed?
3. How long is your average candidate spending on assessments and data entry during your recruiting process?
4. What's preventing you from eliminating the data categories that aren't predictive?
5. If you cut your current data collection in half, would your hiring decisions actually get worse?

[CLOSING REMARKS]

Data minimization forces clarity. When you ask only for what you need, you're forced to be clear about what you actually value.

AI for Recruiters Certification Program

Level 3: Independent Practice | Privacy Discipline And Data Handling | Lecture 15.2

A SkillsClinic initiative.

Duration: ~75 minutes | Word Count: ~2200