Transition from manual filtering to strategic discovery with our reasoning-capable AI engine.
The latest update to TestGorilla Sourcing introduces Smart Search and Ranking Criteria. This "Smarter Teammate" approach transforms recruitment from a reactive waiting game into a proactive, evidence-first discovery engine. By using natural language, you can bypass the "AI application slop" of optimized resumes and identify candidates with scientifically proven capabilities.
Approx. reading time: 3 minutes
In this article
- What is Smart Search?
- How to use Smart Search
- Understanding Ranking Criteria
- The "Skills to Tests" logic
- FAQs
What is Smart Search?
Smart Search is a natural-language search experience that allows you to describe your ideal candidate in plain English. Instead of manually toggling dozens of filters, our LLM (Large Language Model) "reasons" through your request to identify structured data and nuanced role intangibles like culture fit or specific industry experience.
How to use Smart Search
To begin, simply enter a description of the talent you need in the freeform search box on the Sourcing homepage.
Enter your prompt: For example, "JavaScript Developer with FinTech experience in London with 5 years of experience".
AI Extraction: The engine automatically extracts up to six outputs: Job Title, Years of Experience, Location, Salary Range, Skills, and Csutom Criteria.
Review and Edit: Transparency is key. You will see exactly what the AI extracted ("We interpreted this as..."). You retain full control to manually edit or delete any of these attributes in the filter panel.
Note on Custom Criteria: These are freeform phrases (e.g., "startup background") that the system uses for semantic search to find candidates who match the intent of your search, even if those exact words aren't on their resume.
Understanding Ranking Criteria
Once your results are displayed, we provide a visual indicator of how closely a candidate matches your specific search. This is known as the Ranking Criteria.
The Match Score
Each candidate receives a numerical score from 0% to 100%.
Skills Match: Based strictly on verified test scores from our library.
Custom Criteria: Our AI evaluates how well the candidate's profile matches your Custom Criteria, labeling them as Perfect Match (100%), Partial Match (50%), or No Match (0%).
Visual "Pills"
On the candidate card, you will see color-coded labels for each requirement:
Match (Green): The candidate meets the criteria perfectly.
Partial (Yellow/Hyphen): The candidate has some relevant experience or scores.
No Match (Red): The candidate does not meet this specific requirement.
The "Skills to Tests" Logic
We have bridged the gap between how you think (Skills) and how our library works (Tests).
Automated Mapping: When you enter a skill (e.g., "Python"), our Test Recommender automatically suggests relevant TestGorilla tests to prioritize candidates who have proven those skills.
Prioritization: Candidates with high scores across multiple recommended tests are ranked at the top of your list.
Full Transparency: You can hover over any "Verified Skill" tag in the filter panel to see exactly which tests are being used for prioritization. You can toggle these tests on or off to suit your organization's needs.
FAQs
Does the AI scan resumes to calculate the match score?
For Phase 1, the Skills Match is based strictly on verified test scores to ensure objective, merit-based discovery. However, the AI does use profile information to evaluate Custom Criteria, like industry experience.
Can I still select specific tests manually?
Direct test selection is retired in this version. You now work with Skills, and our engine recommends the most relevant tests. This ensures you find the best talent without needing to be an expert on our 400+ test catalog.
What happens if the AI fails to understand my prompt?
If extraction fails or takes longer than 3 seconds, the system will gracefully fall back to a manual filter view, ensuring your search is never blocked.
Is this available in all languages?
At this time, Smart Search and Ranking Criteria are available in English only.