Skip to content
TheThinkLoop
Try the playground

Template - Revenue

Predicate

Lead qualification

Whether an inbound lead is worth a salesperson's time.

A predicate decision for inbound inquiries, with evidence checks for authority, fit, and timing.

Try it in the playgroundCopy the template

The decision

PredicateA probability from 0 to 1 that the condition holds.

Is this inbound lead worth a sales call?

Evidence checks (asked in the same request)

  • Decision-maker: The sender holds a senior title or budget authority.
  • Team size fits: The company or team size is 20 people or more.
  • Buying timeline: The sender states a timeline or an active evaluation.

Expected output: A probability from 0 to 1 that the lead is worth a call, plus one probability per evidence check.

Input

The Decisions API reads one input. Combine these fields into it as labeled text so the model sees the facts the decision depends on.

Input fields
FieldTypeRequiredDescription
messagestringYesThe customer's message, verbatim. Include the subject line if there is one.
enrichmentstringNoFirmographic data you already hold: company size, industry, existing customer or not.

Worked examples

  1. Senior buyer with budget and a timeline

    Hi, I'm Head of Support at a 240-person B2B SaaS company. We have 45 agents on Zendesk and budget approved to automate triage this quarter. Can we get a demo and pricing?

    A careful reviewer would say: Yes. Authority, a sizeable team, an approved budget, and a stated timeline.

    Run this example

  2. Student project

    I'm a student writing a thesis on AI classification. Is there a free forever plan I could use for a personal project?

    A careful reviewer would say: No. No commercial intent. Better served by documentation and a free tier.

    Run this example

Expected answers are human judgments written for this template, not model output.

How it works

"Worth a sales call" means something different at every company. Replace the evidence checks with your own qualification criteria. The three here are a common starting set, not a recommendation.

The checks matter more than the headline number. A lead that passes on fit and authority but has no timeline may belong in nurture rather than in a rep's calendar.

Human review and thresholds

Recommended review policy

Automate the confident ends: book a call for strong leads and send self-serve resources to clear non-fits. Have a person review the middle. Check a sample of automated rejections every week, because lost revenue is invisible unless you look for it.

Choosing thresholds

A wrongly rejected lead costs a deal you never hear about. A wrongly accepted one costs a rep thirty minutes. When pipeline is thin, lower the bar for a call. When reps are overloaded, raise it.

Starting automation threshold
85%
Starting don’t-decide threshold
55%

Starting points for the playground, not recommendations. The right values depend on your costs and on how the model’s confidence tracks accuracy on your data. Calibrate them.

Limitations

  • Short form submissions contain little evidence. Enrich the input with firmographic data before deciding.
  • Signals like job title are easy to misstate, and good leads sometimes write one line.
  • The template scores the inquiry, not the opportunity. It knows nothing about your pipeline or territory rules.

Use it in your code

The template definition is plain JSON: the question, the evidence checks, the input fields, and the starting policy. The SDK call below sends the same question and checks to the Decisions API from your server.

lead-qualification.jsonjson
{
  "$format": "thethinkloop.template/v1",
  "slug": "lead-qualification",
  "title": "Lead qualification",
  "decides": "Whether an inbound lead is worth a salesperson's time.",
  "question": {
    "name": "decision",
    "type": "predicate",
    "instructions": "Is this inbound lead worth a sales call?"
  },
  "evidenceChecks": [
    {
      "name": "factor_1",
      "label": "Decision-maker",
      "type": "predicate",
      "instructions": "The sender holds a senior title or budget authority."
    },
    {
      "name": "factor_2",
      "label": "Team size fits",
      "type": "predicate",
      "instructions": "The company or team size is 20 people or more."
    },
    {
      "name": "factor_3",
      "label": "Buying timeline",
      "type": "predicate",
      "instructions": "The sender states a timeline or an active evaluation."
    }
  ],
  "inputs": [
    {
      "name": "message",
      "type": "string",
      "required": true,
      "description": "The customer's message, verbatim. Include the subject line if there is one."
    },
    {
      "name": "enrichment",
      "type": "string",
      "required": false,
      "description": "Firmographic data you already hold: company size, industry, existing customer or not."
    }
  ],
  "expectedOutput": "A probability from 0 to 1 that the lead is worth a call, plus one probability per evidence check.",
  "examples": [
    {
      "title": "Senior buyer with budget and a timeline",
      "input": "Hi, I'm Head of Support at a 240-person B2B SaaS company. We have 45 agents on Zendesk and budget approved to automate triage this quarter. Can we get a demo and pricing?",
      "expected": "Yes"
    },
    {
      "title": "Student project",
      "input": "I'm a student writing a thesis on AI classification. Is there a free forever plan I could use for a personal project?",
      "expected": "No"
    }
  ],
  "startingPolicy": {
    "automateAt": 0.85,
    "abstainBelow": 0.55,
    "note": "A starting point for exploration, not a recommendation. Calibrate against your own labeled examples."
  }
}
Equivalent SDK calljavascript
import OpenAI from "openai"; // openai >= 7.30.0

const client = new OpenAI(); // reads OPENAI_API_KEY on the server

const decision = await client.decisions.create({
  model: "gpt-6-luna",
  input: "Hi, I'm Head of Support at a 240-person B2B SaaS company. We have 45 agents on Zendesk and budget approved to automate triage this quarter. Can we get a demo and pricing?",
  questions: [
    {
      type: "predicate",
      name: "decision",
      instructions: "Is this inbound lead worth a sales call?"
    },
    {
      type: "predicate",
      name: "factor_1",
      instructions: "The sender holds a senior title or budget authority."
    },
    {
      type: "predicate",
      name: "factor_2",
      instructions: "The company or team size is 20 people or more."
    },
    {
      type: "predicate",
      name: "factor_3",
      instructions: "The sender states a timeline or an active evaluation."
    }
  ],
});

// Answers come back in question order. A question can also be refused.
const [main, ...checks] = decision.answers;
if (main.type === "refusal") throw new Error("The model declined to answer.");
if (main.type === "predicate") {
  console.log(main.probability); // 0–1 estimate that the condition is true
}

Download it as lead-qualification.json, or browse all developer resources.

Need this decision tuned to your policy?

We are running a small number of pilots with B2B teams: your policy, your examples, your thresholds. Request early access.

Related