AI lead qualification by phone: how it works

Start with your definition of a qualified lead

Qualification only works if it reflects your business. Generic frameworks miss what makes your buyers different. Peregrine lets you define a qualified lead two ways today: describe your best customers in plain words, or upload a spreadsheet of past leads marked won or lost. If your deals live in HubSpot or Salesforce, connect it and Peregrine learns from your closed-won and closed-lost deals or opportunities directly.

From any of those it drafts a profile: weighted criteria, must-haves and deal-breakers, each with the numbers behind it when it learned from data. Nothing is called until you approve it.

Look it up before you ask

Nobody enjoys answering questions a quick search could have answered. Before dialing, Peregrine enriches the lead from the company behind the email address: size, industry and location. Criteria it can settle from that data are scored immediately, and they never turn into questions.

The result is a shorter call. If you qualify on company size, industry and timeline, the call may only need to ask about the timeline.

The call itself

Peregrine introduces itself as an AI assistant in its first sentence, confirms it is speaking to the right person, and asks the remaining questions one at a time in natural language, adapting to what the person has already said. It keeps each turn short, never invents prices or discounts, and leaves the selling to your rep.

If the caller asks for a different time, it books a callback. If they ask not to be called again, it marks the lead do-not-call and ends politely.

Scoring that you can explain

Every criterion carries a weight. The fit score is the weighted share of criteria that pass, and a lead qualifies when no deal-breaker applies, every must-have passes, and the score reaches the threshold you set. Each verdict records where it came from: the enrichment data, the form, or the call.

Because the decision is explainable, you can tune it. If good leads keep scoring just under the line, lower the threshold or adjust a weight, and the next call reflects it.

From qualified to a conversation with your rep

When a lead qualifies, Peregrine tells them it is connecting them and starts a live transfer to the next available rep, with a one-line summary whispered to the rep first. See our guide to live transfers for how the round-robin and fallbacks work.

Writing good qualification criteria

Good criteria are observable and specific. "Serious buyer" is too vague to score. "Plans to purchase within 90 days" works. Keep must-haves to the few that truly rule a lead out, and use weights for the rest, so a lead that misses one nice-to-have can still reach a rep. Name deal-breakers explicitly, such as students, job seekers, or locations you do not serve, so the call can end politely instead of wasting a rep's time.

Each criterion needs a question for the call and a plain-English description of what passes. Peregrine uses the description to judge answers, so "at least $100 a month" works better than "high bill."

An example profile

Take a company that sells scheduling software to field-service businesses. Its profile might read like this. Company size between 10 and 500 employees, weighted heavily and usually answered by enrichment before the call. Industry in HVAC, plumbing or electrical, also settled by enrichment. Timeline to go live within the next quarter, a must-have asked on the call. The person on the call makes or shares the buying decision, weighted moderately. Deal-breakers: students, job seekers and companies outside the US and Canada. Threshold: transfer at a fit score of 60.

For a lead from a 48-person HVAC company in Denver, enrichment settles size, industry and location. The call asks two questions: when they want to go live, and who decides. That takes about two minutes, and a qualified caller is on the phone with a rep right after.

Learning from past deals

If you upload a spreadsheet of past leads with an outcome column, Peregrine compares the leads you won with the ones you lost, column by column, and finds simple rules that separate them: a company-size range, an industry, a budget threshold. Each suggested criterion shows the win rate when the rule passes and when it fails, so you can judge whether it is real. Nothing is trained in a black box, and every suggestion is editable.

When a lead does not qualify

A lead that hits a deal-breaker hears a polite close, such as a thank-you and a note that your team may not be the right fit. Nobody is transferred. A lead that asks for another time gets a callback at that time. A lead that asks not to be called is marked do-not-call and never dialed again. In every case the answers, the transcript and the reason for the verdict stay on the lead, so you can review any call that ended without a transfer.

Common objections

"People will hang up on a bot." Some will. The lead asked to be contacted minutes earlier, the assistant says it is an AI in its first sentence, and the call is short. People who want to buy tend to stay on a two-minute call that gets them to a person.

"The AI will misjudge answers." Every verdict records where it came from and the transcript shows what was said, so you can check any decision. Read a sample of transcripts each week in the first month and adjust the descriptions where judgments were off.

"Our form already asks these questions." Good. Peregrine scores criteria the form already answers, so the call only asks what is still unknown.

What to measure

Track the share of reached leads that qualify, the share of qualified leads that reach a rep live, and what your reps say about the leads they receive. If reps report unqualified transfers, tighten a must-have. If good leads are being held back, lower the threshold. A week of transcripts usually tells you which.