Lead Generation Funnels: Stages, Metrics and B2B Examples

To see how many leads in your funnel fit your ideal customer profile (ICP), the ICP fit score calculator explains how a 0 to 100 fit score works and scores your own leads in Cleanlist AI.

The stages of a B2B lead generation funnel

Use this table as your funnel template. Copy the rules into your CRM, then change the examples to fit your market.

If your team is small, merge SAL and SQL into one stage. If you have a free plan or a trial, add a product qualified lead (PQL): a user who reached a usage mark that predicts buying. The labels MQL and SAL come up in Forrester's writing on the SiriusDecisions Demand Waterfall, a B2B funnel model first launched in 2006 (read November 4, 2026).

Lead funnel vs sales funnel

The two meet at the handoff from MQL to SAL. Write that step down most carefully: how fast sales replies, what makes a rep accept a lead, and the reason a rep gives when rejecting one. Pipeline generation covers the sales side: coverage, velocity and how many meetings a bookings target needs.

How to build a lead generation funnel in 7 steps

1. Write your ICP and your definition of a lead

A lead is a person you can reach at a company that could buy. Write the company rules (industry, size, region) and the roles that buy, and share them with sales before you build anything. If you have no ICP yet, ICP in sales builds one from your closed-won deals.

2. Put the stages in your CRM

Give every stage a written entry rule and a date stamp, so you can count each stage and the days between them. HubSpot's default lifecycle stages, read on its knowledge base on November 4, 2026, are Subscriber, Lead, Marketing Qualified Lead, Sales Qualified Lead, Opportunity, Customer and Evangelist, plus Other for records that fit none. HubSpot's automatic updates only move a record forward, and you can customize the stages. There is no SAL stage by default, so add a custom stage or track acceptance in a lead status field. In Salesforce, write down which lead status matches each stage and when a rep converts the lead.

3. Give each stage an offer and a next step

Visitors get content and free tools. Leads get the next offer that fits what they did. MQLs get a rep or a high-intent offer, such as a demo. Every stage needs one action that moves a person to the next one, or the funnel stalls there. The offers and forms themselves are covered in lead generation on your website.

4. Capture clean data

Ask for a work email, look up the rest, and confirm the mailbox exists before the lead enters nurture. A typo or a fake address at capture becomes a bounce in every email after it. The free email verifier checks 2 addresses per IP every 30 days.

5. Score fit and interest, then set the MQL line

Score fit and interest separately. Fit asks whether the company and the title match your ICP, from 0 to 100. Interest counts what the person did: pricing visits, demo requests, replies. A simple MQL rule is fit of 70 or more plus at least one buying action. Leads with high interest and low fit go to nurture. HubSpot lead scoring shows the setup in HubSpot, and account scoring shows how to weight the traits.

6. Write the handoff rules

Agree three things with sales and put them in writing: how fast a rep replies to an MQL, what makes a rep accept or reject one, and the reason codes for a rejection. Give rejected and "not now" leads a path back, with a nurture track and a date to score them again. Review the rejections together every month. Lead routing software covers how each lead reaches the right rep.

7. Measure every week and fix one stage at a time

Count each stage, the conversion rate between stages and the time in each stage. Pick the step with the biggest drop that you can change, fix it, and measure again before you touch the next one. The metrics and the math come after the examples.

5 lead generation funnel examples for B2B

Demo request. The shortest funnel, and the one with the most at stake. A visitor on the pricing page asks for a demo, books a time on the thank-you page and can count as an SAL at once. It leaks when nobody books a time and the reply comes the next day.

Free trial or free plan. The product does much of the qualifying. Pick a usage mark that predicts buying, such as inviting a teammate or connecting a CRM, and call a user who reaches it a PQL. Sales then reaches out to the accounts that fit and hit the mark.

Free tool. A calculator or a checker that does part of the job brings people who have the problem today. Give the first result free, ask for a work email for the full one, and score every user for fit, because a useful tool also draws people outside your market.

Content or webinar. The longest funnel. Treat a download as a lead, and move it to MQL only when fit is high and the person takes a buying action, such as a pricing page visit.

Outbound. This funnel runs in reverse: you choose the accounts first, so fit is set before the first email, and the stages count replies and meetings. What is outbound sales covers the motion. Watch two leaks: addresses that bounce and people who do not own the problem. Verify every address before it enters a sequence.

Lead funnel metrics, with formulas

Measure by cohort: take the leads created in one month and follow them through the stages. Counting this month's SQLs against this month's leads mixes people who entered at different times, and a long sales cycle makes every recent month look worse than it is.

Use your own last two quarters as the baseline. Published conversion benchmarks mix B2B and B2C, different stage rules and different deal sizes, so they tell you little about your funnel. Your own rates, tracked the same way every week, show whether a change worked.

A worked example: more leads, or a better handoff?

A made-up company sells IT asset management software to US companies with 200 to 2,000 employees. It spends $32,000 a month on marketing programs. All the numbers are invented to show the arithmetic.

Deals = leads × MQL rate × SAL rate × SQL rate × win rate, so 400 × 0.25 × 0.64 × 0.50 × 0.25 = 8 deals. Cost per lead is $80, cost per SQL $1,000 and cost per deal $4,000.

Option A: buy 25% more leads. At $80 a lead, 100 more leads cost $8,000 a month. At the same rates that gives 500 leads, 125 MQLs, 80 SALs, 40 SQLs and 10 deals. That is two more deals, and cost per deal stays at $4,000.

Option B: fix the handoff. Sales rejects 36 of every 100 MQLs. Say the reason codes show most of those leads were reached too late, or had a bad email or no phone number. Reply within minutes, verify every email at capture and add phone numbers to MQLs, and suppose acceptance rises from 64% to 80%. That gives 100 MQLs, 80 SALs, 40 SQLs and 10 deals: the same two extra deals with no extra program spend, so cost per deal falls to $3,200.

What it means. After the fix, 2.5% of leads become deals, up from 2%, so every lead you buy afterwards is worth a quarter more. Fix the handoff first, then buy the extra leads.

The data cost in Cleanlist AI. Checking the 400 new leads' mailboxes costs 200 credits, fit scores cost 2,000, and phone numbers for the 100 MQLs cost at most 1,000, since a phone number costs 10 credits only when one is found. That is at most 3,200 credits a month, which fits Pro's 5,000-credit level at $229 a month for one seat. Unused credits roll over.

How to qualify the leads in your funnel with Cleanlist AI, step by step

Cleanlist AI is our product, so here is the qualification job from this page done in it, with what each step costs.

1. Save your ICP once. Put your ICP, personas and exclusions in your Cleanlist AI company profile. Clu, Cleanlist AI's GTM agent, reads them for every list and agent.
2. Bring in your leads. Upload a CSV of recent leads, or connect HubSpot or Salesforce on Pro.
3. Check every mailbox. A mailbox check costs half a credit a lead and catches typos and fake addresses before nurture sends a single email.
4. Score the fit. Clu scores each lead from 0 to 100 against your ICP at 5 credits a lead and writes down the reasons. The ICP fit score calculator explains how the score is built. Set your MQL line on that score.
5. Fill the gaps on MQLs. A verified work email costs 1 credit and a phone number 10, and a miss costs nothing. Cleanlist AI checks 25+ data providers in turn and stops at the first deliverable result.
6. Keep it running on Pro. The Qualify inbound leads agent scores each new lead, within seconds when your form sends a webhook, and posts who to call first in Slack. The Route inbound to owners agent gives each new lead to the owner of its account, at no credits.
7. Fill the top of the funnel with outbound. Search 1B+ profiles by job title, company and location at no credit cost, then pay 1 credit for each verified work email you keep. People Search shows the filters.

What it costs: every new workspace starts with a 14-day Pro trial with up to 3 seats and 250 credits, and a card is required. Pro is $89 a seat a month with 1,500 credits a seat and 5 active agents. Starter is $49 a seat a month with 750 credits a seat. ICP scoring shows the scoring in the product.

Frequently asked questions

What is the 5 minute rule for leads?

It is the rule of thumb that a new inbound lead should hear from a rep within five minutes, while the person is still at their desk and thinking about the problem. Harvard Business Review covered lead response in March 2011 in "The Short Life of Online Sales Leads" by James Oldroyd, Kristina McElheran and David Elkington, and the article's summary says most companies answer online inquiries too slowly (read November 4, 2026). Treat five minutes as the target for demo requests in business hours, and track your median reply time every week.

What is the biggest mistake in lead generation?

Judging the funnel by lead volume. A month with more leads from outside your ICP looks good in a lead report and adds nothing to pipeline. Report the share of leads inside your ICP and the cost per SQL next to the totals. The other common mistakes, such as slow replies and an MQL rule sales never agreed to, show up in those two numbers.

Can ChatGPT generate leads?

It can help plan a funnel, write offers, forms and nurture emails, and research accounts. Asked for contact details from memory, a chatbot can invent them, so connect it to a data source or verify everything it suggests before it enters your funnel. ChatGPT for lead generation has the prompts and the setup, and Cleanlist AI's MCP server, included from the Starter plan, lets an AI assistant look up verified contact data directly.

What should happen to leads that are not ready to buy?

Put them in a nurture track: a short series of useful emails, each pointing to the next offer that fits what they did, such as a calculator or a webinar. Score them again every month and whenever they take a buying action, and move them to MQL the day they cross the line. Keep the track short and stop it when someone unsubscribes or stops engaging.

What is a lead magnet?

A lead magnet is something useful a visitor gets in exchange for a work email: a template, a calculator result, a benchmark report or a webinar seat. In B2B, the ones that bring buyers solve a job the reader has this week. Measure each one by the ICP leads and MQLs it brings.

Do you need funnel software to build a lead funnel?

For a B2B funnel, the CRM is the funnel: it holds the stages, the dates and the owners. Add a form tool, an email tool for nurture, and enrichment and scoring so each lead arrives with what sales needs. Funnel builders help with landing pages and quiz forms at the top, and the stages after capture still live in the CRM.