What is a hot lead? Definition, hot vs warm vs cold, and a test for replies

What is a hot lead? The short definition

A hot lead is a prospect who has a need right now and has shown they want to talk: they asked for a demo, proposed a time, or replied to your email with interest. That is the working hot lead definition in this guide.

A warm lead fits your target customer and has engaged a little, but has not said yes to a conversation. A cold lead has shown no interest, usually because you contacted them first.

Sources define these words differently, so your own rule matters more than the label. Salesforce's warm-lead guidance (September 2017) says sales and marketing should agree on what counts as warm before any program starts, because misunderstandings are common. The useful question is "what is our rule, and can we count it without fooling ourselves?"

FoxKnock sells outbound software, so treat the rule in this guide as one team's opinion.

Hot vs warm vs cold leads

Temperature describes how close someone is to buying, judged from what they have done. In outbound email the three tiers read like this:

Published definitions agree on direction and differ on detail:

• A Dialics blog post (April 2024) describes hot leads as prospects one step from buying who know the product and need it, and warm leads as people who know the company but are not sure they need it.
• Salesforce's warm-lead article (2017) uses "warm" for prospects nurtured through marketing who have said they are open to being contacted.
• HubSpot's SQL guide lists buying signals such as asking about pricing, asking about features and requesting a demo.

In cold outbound the person did not ask to hear from you, so the first real signal is how they reply.

Sales qualified lead vs hot lead (MQL vs SQL)

MQL and SQL are CRM lifecycle stages. They mark who owns a lead and what has been checked, not how warm the person is.

Salesforce (article dated March 18, 2024) defines a marketing qualified lead as a prospect who took a marketing action, such as signing up for an email list or downloading an e-book, and describes that person as still browsing. It defines a sales qualified lead as one a sales team member has flagged as likely to convert, typically through lead scoring. The lead fits the buyer persona, has a pain point you can solve, and has no purchase barriers in the BANT framework (budget, authority, need, timing). Salesforce adds that an SQL is usually qualified through personal contact, such as a call or video meeting.

HubSpot (updated July 31, 2026) defines an SQL as a prospect who has been vetted and deemed ready for a direct conversation with sales. It advises teams to define clear qualification criteria and agreed scoring thresholds.

Pure cold outbound usually skips the MQL stage, because marketing did not generate the lead. On sales qualified lead vs hot lead: a hot lead is a signal, and an SQL is a decision. A positive reply starts qualification; a first call finishes it. A workable order:

1. Contacted (cold)
2. Positive reply (hot lead)
3. Meeting booked
4. SQL, once a first call confirms fit, need, authority and timing

Is every reply to a cold email a hot lead?

No. Reply rate counts whatever lands in the inbox, and much of that is not a person with interest:

• Auto-replies and out-of-office messages. Software wrote them. RFC 3834, a Standards Track RFC from the IETF, says automatic responders should include an Auto-Submitted header. That is a recommendation, so a message without the header may still be automatic.
• Not interested, remove me and wrong person. Replies, not leads.
• One-word replies. "Yes" to "worth a quick call?" is positive. "Thanks" is not. Someone has to read them.

Metabase's guide to reply rate says to exclude auto-replies and out-of-office messages from genuine replies, because they inflate the rate.

How large is the auto-reply problem?

One vendor publishes a number. Sales.co, a cold email service, classified 61,770 replies from campaigns run between 2024 and early 2026 (more than 2 million emails). It reports 45.1% auto-replies, 29.9% negative and 14.1% positive. Its positive group includes questions and "contact later" replies, so it is wider than the hot lead rule in this guide.

Companies that sell outbound publish many figures like this, and this one comes from a single dataset. Read it as "auto-replies can be a large share", not as a number to plan around. One blog that repeats the figure credits it to Instantly, but we found no auto-reply share in Instantly's own 2026 benchmark report.

Vendor rules move too. Explee's public API documentation (read 2026-10-17) defines a hot lead as a person who replied to outreach with real interest. It also says its project reply total now excludes out-of-office and auto-acknowledgement answers, so campaigns with heavy auto-responder traffic show a lower number than before. A headline count is only as good as the rules behind it, and you can check them only when the vendor publishes them.

A worked example

A hypothetical team emails 400 people and 24 reply:

Count any reply and the team has 24 "hot leads" and a 6.0% reply rate. Count real replies (excluding the 9 automatic ones) and it has 15, about 3.8%. Count positive replies and it has 6, a 1.5% rate.

Cost per hot lead differs fourfold depending on which of these counts you divide by (see cost per lead for cold email).

A four-part test for hot leads in cold email

A reply is a hot lead when all four are true:

1. A real person wrote it. Not an auto-reply, an out-of-office message, a bounce or an anti-spam challenge.
2. It shows interest or asks to talk. The person says they are interested, asks for a call or demo, or proposes a time. "Send me something" with no sign of interest does not count, nor does "not now".
3. It comes from the right person. The sender owns the problem or can buy. If they point you to a colleague, log a referral and treat the colleague as a new lead.
4. The need is current. If the timing is next year, you have a warm lead with a date to revisit.

Reasonable teams disagree on edge cases. Fuzzy AI's guide to positive reply rate (October 1, 2026) counts a call request, a question about fit or price, and a referral to a named colleague as positive.

This guide keeps referrals out of the hot count because the person who replied is not the buyer. A price question is a judgment call, since it can be curiosity or a buying question. Pick a rule and apply it to every reply. Our AI SDR buyer's checklist turns this into a question to ask vendors.

How to track hot leads

Count people, not messages. Metabase calls people who replied at least once divided by people contacted the most common and most stable choice, and warns that switching denominators is what breaks reporting.

Give every inbound message one label from a short list (see the reply classes) and read the ambiguous ones by hand each week, including every one-word reply.

Then follow each hot lead to an outcome. If hot leads rarely turn into meetings, your definition is too loose or your follow-up is too slow. A few replies swing every rate, so wait for a larger sample before you draw conclusions.

How FoxKnock handles hot lead replies

FoxKnock's AI inbox labels every inbound message with one of 13 classes: interested, meeting_request, question, objection, not_now, not_interested, out_of_office, referral, wrong_person, unsubscribe, bounce, auto_reply and other. Each label has a confidence score, anything below 0.6 is marked Needs review, and you can correct a label. Claude, by Anthropic, reads the reply and the email it answers, so that text is sent to Anthropic. A keyword fallback labels replies when the AI is unavailable.

Analytics counts a reply as positive when the label is interested or meeting_request. Interested is positive engagement without a proposed time. A meeting request proposes or accepts a time, shares availability or a calendar link, or asks for a call or demo. A price question is labeled question and a referral has its own class, so neither counts as positive.

Alerts are looser than the count. The in-app Hot reply notification and the reply.interested webhook also fire for a confident question, because someone should answer it quickly.

In Analytics, bounces, auto-replies and out-of-office messages never count as replies, and reply, positive and meeting rates are per person contacted. A meeting counts only when a person clicks "Mark meeting booked". Whether the sender is the right person with a current need is still your call, and a classifier can be wrong, which is why low-confidence labels are flagged. Reply labels cost no credits at default credit prices (they can change; see pricing).

If you want to see how your own replies would be labeled, you can sign up for FoxKnock. The Free plan needs no card and includes 500 one-time credits, granted once you verify your email.

Replies sync from a mailbox you connect with an app password over SMTP and IMAP: Google Workspace or any SMTP and IMAP mail server. Microsoft 365 is limited. Microsoft no longer accepts passwords (Basic authentication) for IMAP in Exchange Online and is retiring it for SMTP AUTH, so a Microsoft 365 mailbox may not connect or may not sync replies, and FoxKnock does not offer Microsoft sign-in.