AI interview screenings (HireVue, Sapia, Foundire, and similar platforms) evaluate your answers against a scoring rubric with weighted dimensions. The scoring is deterministic: structure, specificity, and relevance account for most of your score. This guide covers exactly what the models look for and how to deliver it.
Covering behavioral, technical, and situational interview types across all major AI screening platforms.
AI interview platforms score your responses across multiple dimensions simultaneously. Understanding the weighting helps you allocate preparation time to what matters most.
| Dimension | Weight | What It Measures |
|---|---|---|
| Relevance | Did you answer the specific question asked? Off-topic answers score poorly regardless of quality. | |
| Structure | Did you use a recognizable framework (STAR, PAR)? Scoring models are trained to detect structured patterns. | |
| Specificity | Did you provide concrete metrics, timelines, and outcomes? Vague adjectives like "significant" carry no scoring signal. | |
| Clarity | Were you concise? Low filler word count, consistent pacing, and logical flow all contribute here. | |
| Ownership | Did you use "I" and describe your specific actions? "We" signals team context but dilutes individual accountability. |
Standard STAR (Situation, Task, Action, Result) works. STAR+ adds a Signal step at the start that helps the scoring model classify your response immediately, and tightens each section to fit within the 60-90 second window that AI platforms reward.
Total target: 80-90 seconds. Not 45 seconds (scores as incomplete). Not 3 minutes (scores as unfocused and raises your filler word count).
Name the competency upfront: "This demonstrates my approach to stakeholder alignment."
Helps the scoring model classify your response immediately.
Set context in 1-2 sentences. Business, team size, challenge. No backstory.
Establishes the frame without burning time.
Your specific responsibility. One sentence. Distinguish your role from the team.
Signals individual ownership to the scorer.
What you specifically did. Verb-rich, first-person. This is the highest-weighted section.
This carries the most scoring weight across all platforms.
Quantified outcome. Revenue, time saved, percentage improvement. Then one sentence on what you learned.
Closes the loop. Numbers here are what separates pass from fail.
Not all AI interviews work the same way. The format determines what gets weighted. Ask your recruiter which platform the company uses, then prepare accordingly.
Four patterns account for the majority of AI interview failures. Each is fixable with targeted practice.
Fix: Repeat the core noun of the question in your first sentence. "A time I led a project" → start with "The project I led was..."
Fix: Prepare proxy metrics even for soft skills. "Improved morale" → "Reduced voluntary turnover from 22% to 9% over two quarters."
Fix: Practice with a visible timer. Your Action section should be 30-45s maximum. If you hit 90s total, wrap to your Result.
Fix: Replace every "we" with what you specifically did. "We redesigned the flow" → "I mapped the existing flow, identified 3 drop-off points, and proposed the new architecture."
Under the EU AI Act (2026), recruitment AI is classified as high-risk. This means employers must disclose AI use in hiring, provide human oversight of decisions, and give candidates the right to request an explanation of how AI influenced their outcome. In the US, Illinois (AIPA), Maryland, and New York City have enacted separate AI hiring disclosure requirements.
If you are not informed that AI is evaluating your interview, that may be a compliance violation. You can ask your recruiter directly: "Is AI being used to evaluate my responses in this process?"
Under 2 hours of practice. AI scoring models reward the same patterns every time — structure, specificity, timing. Build those patterns before your real screening.
Start Practice Session