Leads
AI lead scoring agent that ranks leads 0–100 and builds your shortlist
Lead Qualifier evaluates your pending leads on evidence: it reads their website, looks for buying signals, writes a 0–100 score with reasons, updates each lead's status and sends you a shortlist of up to 10 leads to contact first.
- As a workflow node
agent.leadQualifier- Default model
- Claude Sonnet 5
- Economy model
- Claude Haiku 4.5
- Tools
- 5
- Report
- Lead qualification
What the agent does
Lead scoring works on your leads in the New and Researching states (up to 40 per run), and a saved agent skips leads it has already decided on. For each lead it gathers evidence with at most two tool calls: fit, offering and size cues from the website, plus buying signals such as hiring, funding, new locations, ads, reviews and technology from a web search. Leads that are obviously outside your profile are ruled out without research.
The score has five parts: profile fit (30), budget capacity (20), evidence of need (20), reachability of the decision maker (15) and timing (15). Leads at or above the threshold become Qualified; leads more than 30 points below it, or with a hard blocker (wrong industry, competitor, out of region, closed), become Disqualified; the ones in between stay Researching with a note. Every decision is written to the lead record with the score and a short reason.
Finally, up to 10 qualified leads with the strongest timing signal go on a shortlist with a “why now” note and are sent to you in a single notification. In the ready-made research → qualify → outreach template this step sits between research and first contact, right before an approval step.
Inputs / Outputs
Inputs
- Qualified threshold
- Leads at or above this score become Qualified (default 70); leads more than 30 points below are disqualified and the rest stay in Researching. 70–80 suits most businesses; a low threshold floods your approval and outreach queues.
- Extra qualification rules
- Your own scoring rules, especially who to rule out. Example: “Rule out agencies and freelancers; never score a company without a website above 50; rule out companies outside London.”
Outputs
- qualified
- Leads above the threshold: score, 2–5 evidence points for and against, and a suggested next step.
- disqualified
- Ruled-out leads: score, reasons and the disqualifying factor (outside profile, competitor, out of region, closed and so on).
- shortlist
- Up to 10 leads to contact first: score, reasons and the timing signal that makes each one worth calling now (“why now”).
Saves a report to the Reports page on every run: Lead qualification.
Tools
- List leads
list_leads - Read a web page
fetch_url - Web search
web_search - Update a lead
update_lead - Notify the owner
notify_owner
Integrations it uses
Example tasks
- “Score the leads found this week. Mark 70 and above as qualified, rule out agencies and freelancers, and give a 2-sentence reason for each.”
- “Read each lead's website. Score dental clinics without online booking higher; that's exactly the problem our product solves.”
- “For leads scoring 50–69, note the missing information and leave them in Researching.”
- “Never score a lead without a budget signal above 60; prioritise companies opening new locations or hiring.”
Tips for good results
- Use Extra qualification rules mainly for negative rules; positive criteria already come from your target profile.
- Calibrate scores with examples in the saved agent's Additional instructions: “This lead should be 90 because…; this one 40 because…”.
- If the approval step gets crowded, filter by score in the workflow and send only the top leads to approval instead of raising the threshold.
- Keep the same model and thinking effort so scores stay comparable over time.
- With Serper or Tavily connected, buying signals are searched on the web; otherwise scoring relies on the lead record and the company website.
Model advice
Default model
Claude Sonnet 5
Per 1M tokens: $2 input · $10 output
Economy model
Claude Haiku 4.5
Per 1M tokens: $1 input · $5 output
The default balanced model is recommended for consistent scores and clear reasons your sales team will read. If you are running a rough first pass over many leads, or your profile is very sharp, the economy model is usually enough. Whichever you choose, stick with it so scores stay comparable over time.
Templates that use this agent
Frequently asked questions
How is the lead score calculated?
Out of 100: profile fit 30, budget capacity 20, evidence of need 20, reachability 15 and timing 15. Extra rules let you shift the emphasis, and the reason for every score is written to the lead record.
Can I change a score by hand?
Yes. Edit the score and status on the lead's detail page; you can also move a disqualified lead back to New or Qualified.
Will it score the same lead twice?
A saved agent keeps the leads it has decided on in memory and skips them. When a lead is re-evaluated, its existing record is updated rather than duplicated.
Which accounts do I need?
None are required. With Serper or Tavily connected it searches the web for buying signals; otherwise it scores from the lead record and the company's website.
Get started with OrqLabs
Build your first workflow today
Start on the free plan: connect your accounts, pick a template and stay in control with approval steps. No credit card required.