1. What Calibration Is (and Why It Matters)
Calibration is the vital intake process where a recruiter defines the skills, experience, and expectations for a role, enabling Eightfold to generate a ranked pipeline from your Talent Network.
Strong calibration directly improves hiring accuracy, speed, recruiter efficiency, DEI outcomes, and candidate experience. It establishes a shared language between recruiters and hiring managers regarding must-haves, trade-offs, and role signals.
2. The Step-by-Step Calibration Workflow
Treat calibration as a live, structured intake conversation with your hiring manager, not just a system step. Aim for at least 80% calibration completeness as a practical baseline.
Step 1: Start Broad. Don't over-filter early. Cast a wide net by defining your foundational criteria: 3 ideal candidates, 5 preferred skills, 3 alternate job titles, a years-of-experience range, and the job location.
Step 2: Add Context and Trade-offs. Use the Calibration Notes section to document hiring context, such as dealbreakers, internal leveling guidance, and the difference between must-haves and nice-to-haves.
Step 3: Validate Using "Extremes". Check the top and bottom leads generated by your initial criteria. This confirms whether the calibration signal is behaving as intended. Adjust only if you strongly disagree with the results.
Step 4: Keep Key Sections Enabled. As a recommended configuration, always leave the Experience and Education sections turned on.
Step 5: Treat as Iterative. Recalibrate as requirements or candidate pools evolve. Do not aim for a "perfect" score; match scores of 3+ stars can be excellent depending on market availability.
3. Breaking Down Calibration Components
Use the table below to optimize each specific section of your calibration criteria.
| Component | Best Practices | Important Nuances |
| Ideal Candidates | Select 3 to 5 individuals currently succeeding in the exact or similar role. Use current employees, applicants, or external profiles (e.g., LinkedIn). | Ideal candidates have the highest impact on match scores. Do not use individuals who are no longer performing the role, as outdated profiles skew results. |
| Skills / Keywords | Focus on hard skills. Select a maximum of 5 skills total. Limit "Required" skills to 1, as required skills will automatically knock out missing leads. | Review the "related skills" Eightfold suggests to ensure they align with the role. Use quotes (e.g., "Computer Science") for exact word matches. |
| Job Titles | Provide at least 3 alternate, commonly used industry titles. Ideal candidate profiles are a great place to source these. | Avoid generic titles, internal company jargon, and acronyms. Do not mark titles as "Required," as titling conventions vary wildly by company. |
| Experience | Prioritize "relevant" years over "total" years for technical, specialized, or management roles. Set a realistic range based on the job description. | Candidates having more experience than preferred impacts the match score far less than candidates lacking enough experience. |
| Education | Use for Interns and Early Career roles (e.g., select Bachelor's and specify the anticipated graduation year). | Higher-level achievements (like a Master's degree) will not negatively impact the match score. |
| Seniority | Reserve this filter primarily for executive, top-level positions (CXO, VP, Director) or clearly defined Interns. | Avoid using this for mid-level initial calibrations unless your resulting leads are drastically misaligned in their career stage. |
Pro-Tip on Grouping Skills: Grouping skills allows multiple skills to be evaluated as one, reducing skill dilution in the calculation. The criteria between individual skills in a group acts as an "OR" function.
4. Post-Calibration: Filter Best Practices
Calibration sets the direction; filters narrow the destination. Once you are in the "right neighborhood" with your calibrated pool, use filters to drill down without breaking the core relevance of the search.
Core Principles for Filtering:
Be Selective, Not Restrictive: Focus on essential criteria. Overusing "Exclude" or "Required" will damage your pipeline.
Match Score is a Guide, Not a Gate: Use the score to focus your pool, but do not treat it as a guarantee of performance or an automatic disqualifier. Lower-scoring applicants remain visible and reviewable.
Save Your Views: If you create a highly effective filtered view, use the "Save view" option to reuse it across current or future positions.
Commonly Standardized Filters:
Highlights: Strong Career Growth, Top School, Similar Industry, Likely to Respond.
Logistics: Location, Visa Status, Languages.
History: Last Application Date, Last Rejection Date, Last Contacted Date.
Attributes: Skills/Keywords (can specify recent experience), Companies (current only), Education, Seniority Level.
5. Operational System Quirks to Remember
Copying Calibrations: You can copy a successful calibration to a new position. However, Title, Location, and Years of Experience will not copy over and must be set manually. Always validate copied Ideal Candidates to ensure they fit the new role.
No Error Messages: Eightfold does not generate standard error messages. If your leads suddenly drop from 12,000 to 600 after entering a skill, check your text formatting and grouping logic.
Pipeline Indexing: If a pipeline is extremely large (1,000+ leads), pages do not load all at once. You may see leads shift on the screen as the system indexes the full pool.
6. Quick Reference: Do & Don't Checklist
| DO | DON'T |
| Do calibrate live with the hiring manager as your primary intake motion. | Don't overuse "Required" and "Exclude" modifiers, which shatter pipelines. |
| Do start broad: 3 ideal candidates + preferred skills + alternate titles + experience range. | Don't assume calibration is "done forever"—revisit it as candidate pools shift. |
| Do use the Calibration Notes to document nuances and trade-offs. | Don't treat the Match Score as a guarantee of quality or an automatic disqualifier. |
| Do validate your setup by reviewing the highest and lowest matched leads. | Don't use every available filter at once; you will isolate candidates who match most, but not all, criteria. |