B2B Marketing Funnel: Follow Buying Groups
Suppose a marketing team reports a busy month: more visitors reached its website, more people downloaded a guide, and more demo forms arrived. Sales still has few deals to work. The easy response is to ask for more traffic. The useful question is narrower: which organizations have a purchase to make, who is involved, and what must happen before those people can choose a supplier?
A B2B marketing funnel is a way to answer that question with defined steps and counts. It can show where interest becomes a sales conversation, where a conversation becomes an opportunity, and where an opportunity advances or stops. It cannot, by itself, describe every move a buyer makes. Gartner describes organizational buyers as revisiting buying tasks, including defining the problem, building requirements, selecting a supplier, and creating consensus, rather than completing them once in a fixed order. Gartner’s account of the B2B buying journey explains why a neat sequence on a dashboard should not be mistaken for the buyer’s actual path.
The practical answer is to keep the funnel, but give it a more demanding job. Count progress toward a purchase at the level of a defined opportunity, connect the people who influence that purchase, and make each stage mean something observable. Keep person-level responses for marketing operations, but do not let a form submission stand in for a buying decision. That choice takes more care with records and sales handoffs. In return, a falling conversion rate can point to a real question: Is the team attracting the wrong organizations, failing to help a buying group agree, or losing viable deals after they reach sales?
A funnel measures your process while buyers revisit theirs
The familiar top, middle, and bottom of funnel picture is useful as a first sketch. At the top, a potential customer recognizes a problem or encounters a supplier. In the middle, people look for possible approaches and decide what a solution must do. Near a purchase, they compare suppliers, settle objections, and work through approval. The sketch helps a team decide what information to offer. It becomes misleading when it assumes that one person moves from an introductory article to a guide to a demo to a contract without going back.
Consider an illustrative organization examining payroll software. Someone in payroll might begin by looking for a way to reduce manual corrections. An operations colleague could ask how the software fits existing work. A person with purchasing authority might enter only when terms are discussed. New requirements could send everyone back to compare options after a demonstration. One individual’s sequence of clicks would show only part of that organization’s decision. The example is a way to plan the funnel, not a report of an actual buyer or a typical purchase.
The roles are more than a storytelling device. Organizational purchases can involve users, gatekeepers, influencers, decision makers, and buyers, with different people affecting the same decision. OpenStax’s description of the B2B buying center supports that distinction. A payroll user may need proof that a task will become easier; a decision maker may need a reason to commit resources. Treating both people as interchangeable “leads” hides the reason a deal is progressing or stalled.
This creates two related views. The buyer view asks what the group still needs to decide: Is the problem agreed? Are requirements clear? Is a supplier acceptable to the people who can approve the purchase? The company view asks what the marketing and sales teams have observed and accepted: Was an inquiry received? Was it qualified? Did sales open an opportunity? A stage in the company view should reflect a recorded event or decision. It should not pretend that the buyer has finished every earlier task.
The distinction also explains why a funnel and a sales pipeline are related but different. A marketing funnel includes the path from early contact or interest into a qualified sales opportunity. A pipeline follows opportunities through a seller’s defined sales steps and eventual outcome. Both may appear in one report, but the unit and stage rules should be explicit at the join. If one chart counts people at the beginning and deals at the end, its narrowing shape alone does not tell you how many potential purchases were lost.
Decide what one row represents before choosing stages
The hardest funnel decision is often the denominator. A person, an account, a buying group, and an opportunity answer different questions. A person can download a resource or request a call. An account identifies an organization. A buying group connects people involved in a particular purchase. An opportunity identifies a possible sale for a particular need. The same account may consider more than one purchase; the same purchase may involve more than one person. Those relationships must be represented before any conversion percentage is useful.
Forrester recommends identifying buying group members and associating their activity with potential opportunities, rather than treating every inquiry as an isolated lead. Its example allows distinct opportunities and buying groups within one target account. Forrester’s buying group guidance provides a useful basis for choosing the opportunity as the primary unit once a specific purchase has emerged. This does not make contact records disposable. It gives each contact a place in the purchase it may influence.
For an early stage, the organization may not yet know whether a purchase exists. It can count known inquiries as people while marking the account and, when possible, the need that prompted contact. Once sales and marketing can describe a potential purchase, report its progress as one opportunity linked to those people. Do not silently add three colleagues’ form submissions to the numerator of three separate deals. Conversely, do not merge two unrelated projects at the same company into one deal just because the email domain matches.
The illustrative payroll case shows the cost of the choice. If three employees ask for different materials about the same proposed replacement, a person-level report has three inquiries. An opportunity-level report may have one potential purchase with three participants. Both counts are true under their definitions. Only the second can be compared directly with the number of opportunities later won or lost. The first is still useful for understanding reach and participation, provided it retains its person-level label.
Start the funnel definition with the commercial question. If the question is “Are our campaigns creating more conversations?” count inquiries and meetings, then disclose how people were deduplicated. If the question is “Are we creating more deals?” count distinct, defined opportunities and their outcomes. If the question is “Are we reaching the people needed to make a decision?” examine roles attached to each opportunity. No single number answers all three.
Give each stage an entry event and an owner
Stage names are choices, not universal facts. “Awareness,” “consideration,” and “decision” describe broad buyer conditions but leave too much room for disagreement in a record. A workable operating funnel might use inquiry, marketing qualified, sales accepted, opportunity opened, active evaluation, and closed outcome. The exact labels can change. What matters is that a person can tell when an item enters a stage, who makes that call, and what later action counts as progress.
Inquiry could mean that an identifiable person asked for information or contact. Record the request, the organization if known, and the stated need. An anonymous page view is useful for website reporting, but without a reliable connection it should not become a named organization’s sales opportunity. At this point, marketing can help someone understand the problem and decide whether the supplier’s category of solution is relevant. The stage should not imply that a budget, project, or buying group has been established.
Marketing qualified can mean that the inquiry meets agreed conditions for follow-up. Those conditions might include an appropriate organization, a plausible need, and a request that warrants contact. The record should show which condition was met, rather than carrying only a status label. A score can help prioritize this work, but the score is a rule for changing a record, not a finding that a purchase will happen. Microsoft’s lead qualification documentation describes a threshold that can mark a lead marketing qualified or sales ready; it also notes that lead stages can later change through automation or manual updates. Microsoft’s qualification guidance is a concrete reminder to keep the threshold distinct from subsequent sales judgment.
Sales accepted should record that someone responsible for follow-up has reviewed the handoff and taken it on. Acceptance is different from a meeting being booked and different again from a buying group agreeing to evaluate the product. A handoff that sales declines should have a reason: the organization is outside the intended market, the request does not concern this product, the contact cannot be reached, or the timing is unclear. Those are operational categories to define locally, not universal diagnoses. They give marketing a way to improve targeting and context without reclassifying every rejected inquiry as a lost deal.
Opportunity opened should require a named organizational need and a sales decision to pursue a possible purchase. Link the known participants and the contact history that explains why the opportunity exists. If a second person at the same organization appears later, attach that person to the relevant opportunity when the connection is clear. Forrester’s buying group approach specifically calls for associating multiple inquiries from one organization with the opportunity they concern. Its guidance on qualifying buying groups is useful here because the second or third inquiry can be additional context, rather than another independent deal.
Active evaluation should mean that there is an observable exchange about requirements, a possible solution, or a comparison. One firm may use a meeting; another may require a documented review of needs. Select one rule that sales can apply consistently. Repeated visits to a pricing page might prompt a question, but they do not by themselves prove that a group has agreed on requirements. A late new participant may restart part of the evaluation. The record can remain in the same commercial stage while notes show that the buying group has returned to an earlier task.
Closed won and closed lost describe an outcome under the company’s chosen sales rules. Keep “no decision” distinguishable from choosing another supplier if those outcomes require different responses. A deal that has simply gone quiet also needs a declared treatment: still open, paused, or closed under a stated condition. Otherwise, the rate of progress changes according to how long each owner leaves silent opportunities in the system. These are design choices that require agreement between the people using the report.
This stage model deliberately costs some simplicity. A business with a short, straightforward purchase may need fewer steps. A team with longer evaluations may need a distinct approval stage. Add a stage only when it changes what someone does or reveals a meaningful handoff; every extra label creates another opportunity for inconsistent entry. Whatever the model, write down entry criteria, owner, allowed exits, and how to handle a skipped or reopened stage. Then a change in the chart is at least a change against known rules.
Match information to the buying task that is still open
Content is often assigned to “top,” “middle,” or “bottom” as if every member of a buying group shares one location. A better starting point is the unresolved decision. Gartner describes buying tasks that can recur, including problem identification, solution exploration, requirements building, supplier selection, validation, and consensus creation. Gartner’s buying journey guidance supports planning for repeated questions, not a one-way content sequence. The aim is to make the next useful piece of information available when a group needs it.
When a team is still defining the problem, help it name the work that is failing and the scope of a possible change. An explanatory article can be useful here if it tells readers what to examine in their own setting. A product pitch at this point may leave the core question unanswered. In the illustrative payroll purchase, someone might need to separate a recurring correction problem from a one-time data issue before a software comparison means much. That is a proposed content choice for the illustration, not a claim that every payroll buyer behaves this way.
When requirements are being built, give readers something they can use with colleagues: a clear account of which tasks a solution covers, what it needs from existing processes, and which questions remain for the buyer to answer. A demonstration can help if it follows the buyer’s stated requirements. A generic demo may generate a meeting without resolving the reason the organization hesitates. The practical test is whether a participant could take the material to another person involved in the decision and explain what the supplier can and cannot do.
At supplier selection, information becomes more specific. A buyer may need product details, implementation conditions, terms, and a way to compare alternatives against the requirements already agreed. The user of a product and the person approving its purchase may ask different questions. Record which question belongs to which participant instead of sending the same sequence of messages to everyone who shares an account. This is where the buying group view becomes a working tool: the opportunity record can show what is known, what remains unresolved, and whom a seller needs to involve.
Consensus deserves attention even when a deal appears close. Gartner includes consensus creation among the tasks buyers revisit, while the buying-center account shows that different roles can influence the same organizational choice. Gartner and OpenStax together support a simple inference: a supplier can have an enthusiastic contact without a group that is ready to decide. When that happens, more reminder emails to the contact may be less useful than material that helps the group resolve a stated objection. A seller should first find out which objection actually exists.
Marketing’s role continues after an opportunity is opened. It can make explanations and comparison material available to newly involved participants, while sales handles the specific conversation. Forrester describes marketing and sales working together to engage buying group members throughout an opportunity. Forrester’s account of buying group progression supports keeping that cooperation in view. The division of work should follow the question to be answered, not an arbitrary rule that marketing stops at a qualified lead.
Treat the handoff as a decision with a reason
The handoff between marketing and sales is often where an apparently healthy funnel becomes hard to interpret. Marketing may mean “qualified” as a score above a threshold. Sales may mean an organization with a current project and a reachable participant. Both can use the same word and count different things. Put the definition in the shared record: what qualified the inquiry, which organization and need it concerns, which participants are known, and what response the person asked for. The owner receiving it should accept, return, or defer it with a reason.
This does not require marketing to know the full buying group before contacting sales. At an early stage, those people may be unknown. It does require honesty about what is known. “Downloaded a guide” is an observation. “Has an approved purchasing project” is a different claim that needs its own basis. The handoff should preserve that difference, because the appropriate follow-up depends on it. A person who requested an explanation of a concept should not receive the same opening as someone who asked for a proposal.
A rejected handoff can be useful information if its reason is specific. If sales repeatedly returns inquiries from organizations outside the intended market, revisit the campaign’s audience and message. If sales accepts relevant inquiries but cannot open opportunities, examine whether the buyer’s need was recorded clearly enough and whether other participants are missing. If opportunities open and then wait through evaluation, investigate the unresolved buying task rather than changing the lead score. These are decision paths, not automatic conclusions from one percentage; each requires looking at the underlying records.
The feedback should also change what marketing asks people to do. A form that offers a guide and a form that offers a sales conversation create different expectations. Count both, but do not treat them as equivalent intent merely because both created contact records. If a campaign attracts many early readers, its contribution may be reach into relevant organizations or useful education. The opportunity report can then ask separately whether any of those organizations developed an active purchase. That separation protects both the reader’s experience and the meaning of the funnel.
Read every conversion rate with its denominator attached
A stage conversion rate is the number that advanced beyond a defined stage divided by the number that entered it. Grid’s stage conversion definition and formula makes the denominator explicit. The calculation is simple; the stage and population rules carry the meaning. “Conversion improved” is incomplete unless the report says what converted, from which stage, over what period, and under which entry rule.
Take an illustrative cohort of 120 distinct opportunities that entered an evaluation stage during one period. If 72 later advanced to a defined approval stage, the evaluation-to-approval conversion is 72 divided by 120, or 60%. Suppose 18 of those same 120 opportunities eventually close won. The cohort’s evaluation-to-win rate is 18 divided by 120, or 15%. The two percentages answer different questions. Neither says that a specific campaign caused a win, and neither describes the experiences of the people who participated in those opportunities.
Time matters as much as arithmetic. A report of opportunities opened this week will include cases that have not had a chance to reach an outcome. A report of opportunities closed this week includes cases that may have entered months earlier. Either can be useful, but mixing the two as though they are one cohort makes a stage-to-stage rate hard to interpret. Choose an entry period and a reasonable observation window for a progression question, and mark which opportunities are still open. A current pipeline snapshot answers a different question: how much work sits in each stage now?
Also declare whether an opportunity can enter a stage without appearing in every earlier one. Google Analytics illustrates the general reporting issue with open and closed funnels: an open funnel includes users who enter at any step, while a closed funnel includes only users who enter at the first step. Google’s custom funnel report documentation describes those different eligibility rules. An opportunity report needs the same clarity. If a buyer comes directly through a proposal request, excluding that opportunity from an earlier-stage cohort is reasonable; counting it as if it had completed an earlier marketing step is not.
Use drop-off to ask what happened, not to declare that all exits are failures. A qualification stage should filter out some inquiries that do not fit the intended purchase. A high exit rate there may mean the filter is doing its job, or that marketing is attracting poor fits; the disposition reasons help tell those stories apart. A fall between sales acceptance and opportunity opening may instead signal a mismatch in what the two teams call qualified. An evaluation stall may reflect unresolved requirements or another participant entering the decision. The percentage locates a question; it does not answer it alone.
Track counts alongside rates. A jump from one of two opportunities advancing to two of two is a large percentage change on a small denominator. A campaign can raise the number of inquiries while lowering the share that become opportunities, yet still produce more opportunities in absolute terms. To decide whether to change spend or content, inspect both the population size and the later outcome for the same defined group. If different products, purchase types, or organization sizes are combined, consider separate views when their routes through the funnel differ. Keep the stage rules visible so the comparison remains meaningful.
Give attribution a smaller, clearer job
The funnel tells you where recorded opportunities moved. Attribution asks how credit for a recorded conversion is distributed among recorded touchpoints. Google describes an attribution model as rules or an algorithm that assigns credit across touchpoints on a user’s path to an important action. Google’s attribution explanation is a useful boundary for interpreting the result. A channel credited for a website action is credited under that model and its available records; the credit is not a complete account of everyone who shaped an organizational purchase.
Consider the illustrative payroll opportunity again. One employee might read an article, a second might attend a product conversation, and a third might approve the purchase. A model built from the first person’s recorded web actions cannot reveal all three people’s influence. Even if records are linked at the opportunity level, the answer still depends on what interactions were captured and how credit was assigned. Use attribution to compare the touchpoints it actually observes, with the model named. Use the buying group and opportunity record to understand the purchase in a wider context.
This matters when deciding what to cut. A late-stage contact may be easy to connect to a won deal, while earlier material helped another participant frame the problem. That is a possible sequence, not a measured claim about any channel. Conversely, an early touch can receive credit without evidence that it helped the group decide. Before moving budget on attribution alone, check which conversion is being credited, which participants are represented, and whether opportunity outcomes tell the same story. The goal is a decision the recorded data can support.
Use the funnel to choose the next improvement
A useful review starts with one break in the path and the records behind it. If relevant organizations arrive but few people ask for the next step, examine whether the page answers their current question and offers a fitting action. If inquiries meet marketing’s criteria but sales rejects them, compare the criteria with sales’ recorded reasons. If opportunities form but linger, ask which buying task remains open and whose participation is needed. If deals reach a final decision and are lost, inspect the recorded outcome before changing the content that brought them in.
Make one change that addresses the identified problem, then compare the same defined stages and population afterward. An improved rate with fewer suitable opportunities may be a poor trade. A slower path with more well-understood buying groups may be acceptable if the purchase requires more participation. The right call depends on the organization’s commercial goal and on what the opportunity records actually show. Funnel charts are strongest when they make that trade visible.
The durable B2B marketing funnel is therefore a compact operating agreement: define the possible purchase, connect its participants, record the events that move it forward, and calculate rates over clearly named populations. It gives marketing a way to create and support opportunities, sales a way to explain what happens next, and both teams a reason to look past a pile of leads toward the buying group trying to make a decision.