Similarity Search
Similarity Search helps you find more people like your best people. Pick one or more employees you consider exemplary — your "favorable people" — and the AI surfaces the closest look-alikes across your talent pool, ranked by education, work experience, and skills.
Use it to build shortlists for hard-to-fill roles, discover hidden bench strength, and find candidates who resemble a proven high performer.
How It Works
Similarity Search compares each candidate against a baseline — the set of exemplar employees you choose — and scores how closely their profile matches. Scoring is driven by LinkedIn-enriched data (education, work history, and skills), so employees need enriched profiles to appear as strong matches.
⚠️ Prerequisite: Similarity Search relies on LinkedIn-enriched data. If some employees are missing LinkedIn profiles, a banner appears at the top of the page. Enrich those profiles first (see Data Enrichment) for the best results.
Building Your Baseline
The baseline is the group of "favorable people" you want to find look-alikes for. You can add 1 to 10 exemplars.
- Open Similarity Search from the navigation bar.
- Click Add People (or the + button) to open the picker.
- In the picker, search by name, position, or personnel number — results update as you type.
- Click a person to add them to your baseline. Repeat to add up to 10 exemplars.
- Your baseline is sticky — it stays selected across multiple runs, so you can tweak filters and re-run without rebuilding the list.
- To remove someone, click the × on their chip in the baseline list.
💡 Tip: You don't need to pick 10 people. A single strong exemplar works well; adding more exemplars broadens the "signature" the AI looks for.
Narrowing the Candidate Pool (Optional Filters)
Before running, you can constrain who is eligible to appear as a match using the filters in the left panel:
| Filter | Purpose |
|---|---|
| Job Layer | Limit matches to one or more layers (e.g., Staff, Section Head, Division Head) |
| Age Range | Limit matches to an age bucket (e.g., 30–39, 40–49) |
- Filters apply to the candidate pool, not to your baseline exemplars.
- Leave the filters empty to search across everyone you have access to.
Running a Search
- Once your baseline has at least one person, the Find Similar People button becomes active.
- Click Find Similar People.
- The AI returns the top 10 people most similar to your baseline, ranked by education, work experience, and skills.
Reading the Results
Baseline Signature
At the top of the results, the Baseline Signature summarizes the common profile the AI extracted from your exemplars:
- Exemplar count — how many people are in your baseline
- Job families — the dominant career families across your exemplars
- Layers — the job layers your exemplars sit in
- Common skills — the skills most shared across the baseline
This tells you what the search is matching on, so you can sanity-check the results.
Look-alike Match Cards
Each candidate appears as a card showing:
| Element | What it means |
|---|---|
| Similarity score | 0–100, color-coded (green = strong match, amber = moderate, red = weaker) |
| Name, position & talent class | Who the candidate is and where they sit on the talent matrix |
| Why Similar | A breakdown of the shared attributes — the education, work experience, and skills that drove the match |
Acting on a Match
From a candidate card you can move straight into follow-up actions:
- View Career Path — jump to the candidate's Career Navigator page to see their trajectory, fit scores, and development plans.
- Add to Succession — nominate the candidate as a successor without leaving the results.
💡 Tip: Combine Similarity Search with succession planning — find people who resemble a strong incumbent, then add the best matches directly to that role's succession plan.
Search History
- Past runs are saved in the history panel on the left. Click any entry to reload that run — including its baseline and results.
- You can collapse the history panel to give the results more room, and clear your history when you no longer need it.
Common Mistakes to Avoid
- Expecting matches for un-enriched employees — people without LinkedIn data won't score well because the match relies on education, work experience, and skills.
- Over-filtering — applying a very narrow Job Layer + Age Range combination can leave few or no candidates. Loosen the filters if results look empty.
- Confusing the baseline with the results — filters shrink the candidate pool, not the exemplars you selected.