What Is a Sales Funnel? Stages, Conversion Logic, and Model Limits

A sales funnel is a staged model of how potential buyers move from first awareness of a product or problem toward purchase. It usually narrows because some prospects drop out, fail to meet later-stage criteria, or remain unresolved. The useful part is not the funnel picture. It is the conversion logic underneath it: comparable stage counts, clear entry evidence, and an honest view of where a linear model stops matching how buyers actually buy.

Salesforce’s current sales funnel guide defines the funnel as the journey from awareness to purchase and makes an important qualification: prospects can bounce among stages or jump ahead. Salesforce Canada’s introductory comparison explains why the visual narrows, with more prospects near the top than customers at the bottom.

There is a formula, but it is a measurement formula rather than a definition of the funnel itself:

Stage conversion rate (%) = records from the same cohort that reach the next stage / records that entered the current stage x 100

Overall funnel conversion rate (%) = final-stage conversions / first-stage entrants x 100

Stage drop-off rate for a completed cohort = 100% - stage conversion rate

Grid’s stage conversion definition uses the same idea: the numerator is the count that advances beyond a stage, and the denominator is the count that entered that stage. Salesforce’s lead conversion guide gives the broader lead-conversion formula and adds the crucial operating point: the team has to agree on what conversion means and track it consistently.

Here is an illustrative example, not real company data. A cohort starts with 500 qualified leads. Of those, 200 become sales accepted, 80 become opportunities, and 20 become customers.

Because each later stage in this simple example is an ordered subset of the same cohort, the stage rates reconcile to the total: 0.40 x 0.40 x 0.25 = 0.04, or 4%. That reconciliation stops being reliable when records can enter mid-funnel, skip steps, move backward, repeat stages, merge, or remain open after the reporting window.

Four neighboring terms are worth separating before a team trusts the dashboard:

The terms overlap because the same commercial motion can be viewed from several angles. Salesforce distinguishes the pipeline as the seller’s view of prospects and stages, while the funnel is broader and buyer-oriented. Shopify’s sales funnel guide draws one practical boundary between marketing and sales: marketing brings the right people in, while sales removes friction for people with purchase intent. That boundary is useful, but it is an operating choice, not a law.

Common sales funnel stages are vocabulary, not a standard#

Sales funnel stage names often come from one of three vocabularies. One familiar shorthand is AIDA: attention, interest, desire, and action. A broader operating view uses top, middle, and bottom of funnel: TOFU, MOFU, and BOFU. Salesforce’s current six-stage example uses awareness and discovery, interest, consideration, intent, evaluation, and purchase.

None of these lists is universal. IBM’s sales funnel overview also notes that the number of stages depends on the business, industry, and purpose. A high-velocity ecommerce funnel can define product-page view, cart, checkout, and purchase. A B2B SaaS team may need lead, accepted lead, opportunity, evaluation, proposal, procurement, and closed outcome. A founder-led enterprise sale may need fewer dashboard stages and more buyer-verifier evidence.

The practical test is whether a stage changes a decision. If a buyer is merely reading more content, the next action may be nurture. If the buyer has confirmed a problem, a buying group, and an evaluation path, the next action may be sales ownership. If the buyer is in procurement, the next action may be commercial risk management. A stage earns its place when it changes what the team does.

Avoid defining stages as moods. “Interested,” “warm,” and “ready” become measurement traps when no one can inspect the evidence. Translate them into observable states:

Conversion logic needs a cohort, a clock, and one unit#

A common funnel error is dividing today’s stage counts by each other. If 90 records are currently in evaluation and 30 are currently in proposal, the ratio 30 / 90 may look like a conversion rate. It is not. Those 30 proposal records may not have come from the 90 evaluation records, and both groups may contain different ages, segments, sources, owners, and stage definitions.

A defensible funnel report needs one clear sentence before it needs a chart. State the unit, the entry stage and evidence, the cohort window, the next-stage evidence, and the observation window in that sentence.

Each bracket matters.

The unit might be a person, account, buying group, lead, opportunity, subscription, checkout session, or user. Do not count contacts at one stage and accounts at the next. The cohort window defines who is eligible. The observation window gives records enough time to mature. The stage evidence keeps the numerator from becoming a collection of opinions. The terminal policy says how to count open, lost, disqualified, recycled, duplicate, skipped, and merged records.

Google Analytics’ funnel documentation is useful here even for teams measuring sales in a CRM. Its custom funnel reports separate retained users from abandonments, and its funnel exploration documentation shows that open and closed funnels count different entry populations. The same principle applies to sales operations: the dashboard’s answer changes when the counting rule changes.

Benchmarks can orient you, but they cannot define your funnel#

There is no universal good sales-funnel conversion rate. A “good” rate depends on the unit, channel, stage definition, offer, price, sales motion, buyer segment, observation window, and data hygiene. A product-led free-trial funnel, a sales-led enterprise funnel, and an ecommerce checkout funnel can all be healthy with very different shapes.

Use external benchmarks as context, not as a target to force the system toward. First Page Sage’s B2B SaaS funnel benchmark report is a bounded example: it discloses a data set from more than 50 B2B SaaS clients over ten years, mostly small-to-midsize businesses, and reports stage rates under its own definitions. That can help a SaaS team ask better questions about channel mix and handoffs. It is not an independent universal standard, and it will not tell a company whether its own MQL, SQL, opportunity, or closed-won labels match the report’s.

An internal benchmark is often more useful: compare a new cohort with an older cohort under the same definitions. If one stage gets worse, then inspect the records. Did the lead source mix change? Did the qualification rule change? Did the CRM timestamp change? Did the offer change? Did a new routing rule delay follow-up? A rate locates a transition worth investigating. It does not name the cause.

The funnel model breaks when it pretends to be the buyer’s mind#

A sales funnel is a compression tool. It reduces many buyer histories into a few states so a team can count movement. That is why it is useful. It is also why it can mislead.

Gartner’s B2B buying journey guidance describes B2B buying as nonlinear, with buyers revisiting buying jobs and multiple stakeholders working through tasks. McKinsey’s consumer decision journey made a related critique of the classic funnel: modern buyers may add and remove options during evaluation, and post-purchase experience can feed future decisions.

A funnel can answer:

• Which comparable records reached each defined state?
• Where does movement slow or stop?
• Which handoff needs inspection?
• Did a later cohort behave differently after a documented change?

A funnel cannot answer by itself:

• Why buyers changed their minds
• Whether marketing caused a stage movement
• Whether a single buyer followed the modeled path
• Whether one stakeholder’s activity represents the buying group
• Whether the current pipeline will close on time
• Whether a benchmark from another company should become your target

For those questions, pair the funnel with other evidence. Use pipeline review for active deals, customer interviews for buyer reasoning, cohort analysis for timing, path and attribution analysis for touchpoint-allocation questions, experiments or credible causal designs for causal claims, and post-sale data for retention and expansion.

How to use a sales funnel without over-trusting it#

Start with the decision the funnel needs to support. If the decision is budget allocation, keep channel as a breakdown and compare mature cohorts under stable stage definitions. If the decision is sales follow-up, define handoff states and owner clocks. If the decision is product onboarding, use product activation events rather than CRM labels. One funnel rarely serves every decision well.

Then write a small measurement contract:

This contract is the thing worth remembering. The funnel drawing is only the visible output. The contract determines whether the numbers are comparable.

Once the contract exists, use the funnel in a simple diagnostic rhythm:

1. Name the weak transition.
2. Reconcile the numerator and denominator to record history.
3. Separate volume, conversion rate, and time in stage.
4. Read representative records and buyer evidence.
5. Change one bounded rule, offer, message, handoff, or follow-up process.
6. Compare the next mature cohort under the same definitions.

The best funnel report is not the one with the most stages. It is the one another operator can audit without asking what the labels really mean.

Use a sales funnel when you need a shared, measurable view of buyer progression. Stop using it as the whole explanation when buyers loop, multiple people influence the decision, or the next question is causal. The funnel can show where to look.