AI Scout — Recruiter Guide
For: Recruiters and hiring managers using Lope’s AI Scout
Last updated: March 2026
Table of Contents
1. How AI Scout Works
1.1 What is AI Scout?
AI Scout is Lope’s intelligent candidate search engine. When you describe the kind of candidate you are looking for — by job title, skills, location, experience, or industry — AI Scout searches your entire candidate pool and returns the most relevant matches, ranked from best fit to least. Unlike a traditional search that only finds exact keyword matches (e.g. typing “Software Engineer” would miss candidates titled “Full-Stack Developer”), AI Scout understands the meaning behind your words. It knows that “Full-Stack Developer” and “Software Engineer” are closely related roles. It knows that “Python” and “Machine Learning” are related skills. It knows that “London” and “United Kingdom” refer to the same geography. This means you spend less time tweaking search terms and more time talking to great candidates.1.2 The Five Dimensions of Search
When you run a search, AI Scout evaluates every candidate in your pool across five independent dimensions. Think of these as five separate expert opinions about how well each candidate fits your criteria:
Each dimension is evaluated independently. No single dimension can override the others. This prevents scenarios where a candidate with a perfect title match but zero skill overlap ends up at the top of your list.
1.3 How Ranking Works
After AI Scout evaluates every candidate across all five dimensions, it needs to combine those separate scores into one final ranking. Here is how: Step 1 — Score within each dimension For each dimension you searched (e.g. you specified a job title and skills), every candidate gets a relevance score. Candidates with the same relevance in a dimension share the same rank — the system is fair and does not arbitrarily break ties. Step 2 — Convert scores to points Within each dimension, candidates are awarded points based on their rank. The top-ranked candidate in that dimension gets the most points, the second gets slightly fewer, and so on. Candidates with identical scores get identical points. Step 3 — Add up the points across all dimensions Each candidate’s points from every dimension are summed. A candidate who ranks highly across multiple dimensions accumulates more total points than one who is excellent in only one area. Step 4 — Final sort Candidates are ordered by:- Total points (highest first) — the primary ranking factor
- Average match quality (highest first) — used to break ties
- Number of dimensions matched (most first) — a final tiebreaker
1.4 Why AI Scout is Different from Keyword Search
2. Writing Effective Search Prompts
2.1 The Five Search Fields
Every search in AI Scout can use up to five fields. You do not have to fill in all five. Use only the ones that matter for your role. The more fields you provide, the more refined your results will be.
At least one field must be filled in to run a search.
2.2 Job Title — Best Practices
The job title field is one of the most powerful search dimensions. AI Scout doesn’t just look for the exact words you type — it understands what the role means. Do:- Use the most common version of the title:
Software Engineer,Product Manager,Recruiter - Use natural language:
Senior Data Scientist,Head of Marketing - Be specific about seniority when it matters:
Junior UX DesignervsLead UX Designer
- List multiple different roles in one search (run separate searches instead)
- Use internal or proprietary title conventions that wouldn’t appear on LinkedIn profiles
- Add unnecessary modifiers like company names
Tip: If you are looking for a very niche role, try the most commonly used industry title. AI Scout works best with titles that candidates actually put on their profiles.
2.3 Skills — Best Practices
The skills search checks for both exact matches (the candidate has that exact skill listed) and related skills (semantically similar skills). Exact matches are given significantly more weight. Do:- List specific, concrete skills:
Python,Figma,SQL,Project Management - Separate multiple skills with commas:
Python, Docker, Kubernetes - Use the names that appear on professional profiles (e.g.
ReactnotReactJS library) - Include 3–7 skills for the best results — enough to differentiate, not so many that no one matches all of them
- Use full sentences:
experience with Python programming languagePython - Mix skills with other criteria:
Python, 5 years, London - List too many skills (15+) — this dilutes the signal
Python, Docker, AWS:
Tip: The first candidate in the example above would rank highest because exact skill matches carry the most weight. Candidates with related (but not identical) skills still appear, ranked below exact matches.
2.4 Location — Best Practices
AI Scout understands geography at multiple levels — from specific cities to entire continents. Do:- Type naturally:
London,Berlin, Germany,United States,Europe - Use common location names that appear on profiles
- Search for regions when you’re flexible:
EU,Europe,DACH
- Use postal codes or zip codes
- Combine locations with other search criteria in this field
Tip: If you’re open to multiple locations, run the search with the broadest acceptable location. For example, if you’d hire anyone in the DACH region, search
Germany or DACH rather than running three separate searches for Germany, Austria, and Switzerland.
2.5 Years of Experience — Best Practices
The experience field lets you filter by total years of professional experience. Unlike a simple cutoff, AI Scout scores candidates on a curve — candidates closest to your target get the highest scores, while those slightly outside your range still appear but rank lower. Supported formats:
How experience scoring works in practice:
If you search for
> 5 (at least 5 years):
Key insight: The system doesn’t use a hard cutoff. Candidates at 4.5 years still appear when you search
> 5, because in practice, someone with 4.5 years of experience might be just as qualified. They simply rank below candidates with 5+ years.
In addition to total years, the system also considers:
- Average tenure — how long the candidate typically stays at each company
- Current tenure — how long they’ve been in their current role
> 3, < 8) when you have a specific seniority band in mind. This avoids burying your ideal mid-career candidates under very senior profiles with 20+ years.
2.6 Industry — Best Practices
The industry field searches the companies the candidate has worked at, not the candidate’s job title. This is a powerful way to find candidates with relevant domain experience. Do:- Use broad industry terms:
FinTech,Healthcare,E-commerce,SaaS - Separate multiple industries with commas:
casino, lottery, gambling - Think about the kinds of companies your ideal candidate would have worked at
- Confuse industry with job title (use the job title field for that)
- Use overly niche terms that wouldn’t appear in a company description
- What industry the company operates in
- What the company’s description says
- What the company’s specialties and tags are
- An AI-generated summary of what the company does
Tip: The industry search is especially useful when you’re hiring for a role that requires domain expertise — for example, a Product Manager who has worked in FinTech before, not just any Product Manager.
2.7 Combining Fields for Powerful Searches
The real power of AI Scout comes from combining multiple fields. Here are some strategies: Strategy 1: Start Broad, Then Narrow- Start with just a job title:
Data Scientist - Review the results — if too many, add skills:
Data Scientist+ skills:Python, TensorFlow - Still too many? Add location:
Data Scientist+Python, TensorFlow+Berlin
Software Engineer + React, TypeScript + London Potentially too narrow: Software Engineer + React, TypeScript, Node.js, GraphQL, Docker, AWS + London + > 5, < 8 + FinTech
Strategy 3: The “Domain Expert” Search
When you need someone with specific industry experience, lead with industry:
- Job title:
Product Manager+ Industry:FinTech - Job title:
Sales Director+ Industry:SaaS, Enterprise Software - Job title:
Nurse Practitioner+ Industry:Healthcare
- Job title:
Senior Software Engineer+ Experience:> 5, < 12 - Job title:
Junior Designer+ Experience:< 3 - Job title:
VP of Sales+ Experience:> 10
2.8 Quick Reference: Search Examples
Here are ready-to-use search examples for common hiring scenarios:3. Understanding Match Scores and Explainability
One of the most important features of AI Scout is transparency. Every result comes with a detailed breakdown that explains exactly why a candidate was ranked where they are. This section teaches you how to read and interpret those scores.3.1 How Each Candidate Gets a Score
Every candidate in your results has:
Below the summary, you’ll see a per-dimension breakdown — one section for each search field you used.
3.2 What the Score Breakdown Tells You
Each dimension in the breakdown shows:Job Title Breakdown
Example: You searched for “Software Engineer”. A candidate titled “Full-Stack Developer” might score 85%, while “Project Manager” might score 35%.
Skills Breakdown
Example: You searched for
Python, Docker, AWS. The breakdown shows:
- Matched: Python, Docker (2 out of 3 exact)
- Missing: AWS
- The candidate also has “Azure” (related to AWS), which slightly boosts their similarity score
Location Breakdown
Example: You searched for “Berlin”. A candidate in “Berlin, Germany” scores 100% (exact). A candidate in “Munich, Germany” scores ~70% (same country, different city). A candidate in “London, United Kingdom” scores ~30% (different country entirely).
Experience Breakdown
Example: You searched for
> 5. A candidate with 7.2 years shows “Within Target: Yes” and gets high points. A candidate with 4.5 years shows “Within Target: No” but still appears with moderate points because they’re close.
Industry Breakdown
Example: You searched for “FinTech”. A candidate who worked at Stripe scores 95%. A candidate who worked at a traditional bank scores ~60% (related to finance, but not specifically FinTech). A candidate who worked at a restaurant chain scores ~10%.
3.3 Reading the Results: A Practical Walkthrough
Let’s walk through a real example. Imagine you searched for:- Job Title:
Senior Software Engineer - Skills:
Python, React, Docker - Location:
Berlin - Experience:
> 3
Rank #1 — Total Points: 350 | Average Score: 88% | Matched: 4/4 dimensions
Why they’re #1: Perfect skill match (3/3), exact location match, closely related title, and experience within range. Strong across all four dimensions.
Rank #2 — Total Points: 310 | Average Score: 82% | Matched: 4/4 dimensions
Why they’re #2: The title is actually a slightly closer match (92% vs 87%), but the missing React skill (2/3 vs 3/3) and lower experience score bring their total points below Rank #1.
Rank #3 — Total Points: 280 | Average Score: 76% | Matched: 4/4 dimensions
Why they’re #3: They match on skills and experience, and the title is related. But they’re in Munich (not Berlin), which lowers the location score, and their title is less directly relevant.
3.4 Why Candidate A Ranks Higher Than Candidate B
The ranking system is designed to be intuitive and fair. Here are the key principles: Well-rounded beats one-dimensional. A candidate who scores 80% across all four dimensions will rank higher than a candidate who scores 100% on job title but only 50% on skills, 40% on location, and 0% on experience — even though the second candidate has a “perfect” title match. More dimensions matched = higher ranking. If two candidates have the same total points, the one who appeared in more of your search dimensions ranks higher. A candidate found in all 5 dimensions is generally a better fit than one found in only 2, even if the 2-dimension candidate has slightly higher individual scores. Exact matches matter most for skills. For skills specifically, having the exact skill listed on a profile counts much more than having a related skill. “Python” matching “Python” is worth significantly more than “Java” matching “Python” (even though they are somewhat related). Ties are broken fairly. When multiple candidates have the same total points, the system breaks ties by their average match quality (higher is better), then by how many dimensions they matched in (more is better). Candidates with identical scores in a dimension always receive the same number of points — no candidate is arbitrarily ranked above another equally qualified person.3.5 Common Score Patterns and What They Mean
Here are patterns you’ll frequently see in your results, along with what they indicate:3.6 Frequently Asked Questions About Scoring
Q: Why does a candidate appear in my results even though they don’t meet all my criteria? A: AI Scout is designed to surface the best available candidates, not to hard-filter your pool. A candidate who matches 4 out of 5 dimensions strongly is almost certainly worth reviewing, even if they’re slightly short on one criterion. Hard filters (like requiring exactly 5+ years) would hide candidates at 4.9 years who might be perfect in every other way. Q: Why does a candidate with a slightly lower title match sometimes rank higher than one with a “perfect” title? A: Because ranking is based on the combined score across all dimensions, not just one. A candidate with an 85% title match, 95% skill match, and 100% location match will rank higher than a candidate with a 95% title match but 50% skill match and 40% location match. The system rewards overall fit. Q: What does it mean when a candidate has 0% in one dimension? A: It means the candidate had no data or no relevance in that dimension. For example, if a candidate hasn’t listed any skills on their profile, they’d score 0% on skills — but they can still rank well if they’re strong on job title, location, and experience. A 0% doesn’t mean they’re unqualified; it means there wasn’t enough profile data to evaluate that dimension. Q: How often are scores and rankings updated? A: Scores are calculated fresh every time you run a search. There’s no stale data — AI Scout always uses the latest candidate profiles in your pool. Q: Can I trust a 90%+ match? A: A 90%+ match across multiple dimensions is a very strong signal. However, AI Scout evaluates profile data, not the person themselves. A 90% match means the candidate’s profile is highly aligned with what you’re looking for. You should still review their full profile and assess fit through a conversation. Q: Why do some candidates appear in my results but with very low scores? A: These are candidates who have some relevance to your search but are weak matches overall. They appear at the bottom of your ranked list. Focus your attention on the top-ranked results — the system puts the most relevant candidates first. Q: I searched for a very niche role and got few results. What should I do? A: Try broadening your search:- Use a more common job title (e.g., “Software Engineer” instead of “Platform Reliability Engineer”)
- Reduce the number of required skills
- Broaden the location (e.g., “Europe” instead of “Berlin”)
- Widen the experience range
- Remove the industry filter
Skills: Python, Docker will rank candidates purely by their skill relevance. This is perfectly valid, though combining multiple fields typically produces more meaningful rankings since the system can evaluate candidates from multiple angles.
This guide is maintained alongside the AI Scout product. If you have questions not covered here, please reach out to the Lope team.