Your MQL count went up again, and sales still does not believe it. You have probably stopped defending that number and started explaining it. Would your two teams sign the same piece of paper?
It is the third Tuesday of the month, and your B2B demand generation meeting has reached the item about leads. Marketing puts the MQL count on the screen. Sales says the leads were not real. Somebody asks for the MQL-to-SQL rate, nobody has it, and the meeting ends with a promise to look into it. Do you believe the number you send, or do you send it because arguing about it would take longer than the report?
I have watched that meeting run in four companies, and the MQL is rarely what is broken. The break sits one level down. Nobody in the room has written down what a qualified conversation is. So marketing measures its definition, sales measures its own, and both teams read the gap between them as the other team's performance problem. That is what makes the meeting repeat. Fix the definition and it changes shape. Leave it alone and you will be in the same room in December. The number will be bigger. The silence after sales speaks will not have changed.
The size of that gap is now measurable. Research commissioned by Docket and fielded by Heinz Marketing in September 2026 surveyed 192 marketing, demand-gen and RevOps practitioners. Nearly 90% of them send MQL volume to leadership. Only 7% were confident it predicts real pipeline. And 78% said their sales organization rejects more than half of the MQLs it receives. The study is vendor-run and the sample is small, so read it as a signal rather than a law. As a signal it is loud.
You did not come here for another definition of B2B demand generation. If you wanted one, the answer sitting above this page would have ended your search. You also do not need the demand-versus-lead argument again, which we keep in demand generation vs lead generation. What is left is the part those pages skip. Which numbers does your demand gen function read on a Monday? Which two does it read on the first of the month? And which decision is it not allowed to make until the month closes?
Five Definitions Both Teams Sign Before Any B2B Demand Generation Number Moves
The vocabulary is already settled, which is why arguing about it wastes a meeting. HubSpot's glossary defines a sales lead, an MQL, an SQL and a prospect, and it does the job. What none of those entries gives you is the operational layer: who applies the definition, on what evidence, and what happens when sales disagrees. A glossary entry is a meaning. Your team needs a sheet it signs.
Five rows, and each one needs three columns: the definition, the evidence that satisfies it, and an owner your team names out loud.
Lead. A person or account that has expressed interest and has not been assessed against anything. The evidence is an action: a form fill, a reply, a badge scan, a calendar booking. The owner is whoever captured it, usually marketing. A lead is a fact about behavior. The trouble starts the day your company reads the count as a judgment about fit.
MQL. A lead that matches your profile and has crossed a threshold you defined in advance. The evidence has to be nameable: which pages, which actions, which firmographic attributes, what score, over what window. If nobody can list that evidence without opening the automation platform, the threshold is arbitrary. Forrester's Terry Flaherty, in the firm's revenue process series, calls MQL definitions "typically made up". What he describes is a scoring model built around one person while the purchase is made by a group. Forrester has its own opportunity-led process to sell, so read the criticism with that in mind. It is still the clearest statement of the problem.
SQL. A lead that sales accepted after a conversation, on four things: a real need, a person with authority, a timeline, and a budget that exists. The evidence is a note from the call in the salesperson's own words. The owner is sales. If your sales team will not write the note, the stage is decoration.
Opportunity. An account with a deal in it. Account, not person. Forrester puts the share of buying decisions made by groups of more than three people at over 80%. That figure is the structural reason a person-shaped opportunity undercounts every deal you have. From here the shape of the funnel matters more than the count. Our lead generation funnels piece covers the stage between SQL and opportunity that usually gets skipped.
Pipeline. The sum of open opportunities, each with an amount and an expected close date, inside the period you are measuring. The owner is whoever owns the forecast. The definition has to match what finance sees, or the number cannot be used in a budget conversation.
The row that decides whether the sheet survives is a sixth one nobody prints: what happens when sales rejects an MQL. Write it down. The rejection comes back to marketing with a reason chosen from a fixed list. Marketing reviews the reasons weekly, and the criteria change at month end, never mid-quarter. That rule takes the heat out of the room. It turns an argument about judgment into an argument about evidence, and evidence is a much shorter conversation.
The B2B Demand Generation Cadence: What You Read Weekly and Monthly
A cadence is an agreement about which numbers change a decision, and at what interval the decision is allowed to change. Plenty of functions run a reporting schedule instead, which is why your pack grows every quarter while the decisions stay where they were.
Weekly: three numbers, thirty minutes
Pipeline created. This means new opportunities with an amount, sourced to a channel, this week against the same week last quarter. The decision it drives is whether the offer or the audience is working. Pipeline flat for three weeks while spend holds means the offer is the problem. Spend up and pipeline flat means the targeting is.
Stage-conversion deltas. MQL to SQL, SQL to opportunity, as a rolling four-week rate against the previous four weeks. The decision is whether the qualification criteria are being applied or quietly ignored. A rate that moves five points in a month is a signal about the people applying the criteria. Our sales pipeline management piece goes deeper on what a coverage number is really telling you.
Content in market. What shipped, what is live, and what is being read by accounts that later became pipeline. The decision is what to brief next. This is the only leading indicator of the three. It is also the one a weekly meeting tends to skip, because it does not fit on a dashboard. If you are choosing which asset to build next, lead generation content covers which formats convert.
One page, assembled before the meeting, from the systems that own the data. If the pack takes four hours to build, it is dead by the third month. That is why we generate ours inside Allable. A number that costs a morning to produce gets produced less often than one that is already there.
The decision the weekly meeting is not allowed to make
Budget reallocation does not happen weekly. That is a rule, not a preference, and it belongs next to the agenda because the temptation arrives every week.
The mechanism is sample size. A channel that produces eight qualified conversations a month swings by half when two deals slip. In a single week you are looking at two or three. Two or three cannot separate a channel that is failing from a channel that is early. Reallocation on that sample is how a program ends up with twelve half-funded channels and no finished test. The budget moves monthly, on channel contribution computed with the same definitions. The weekly meeting is where you surface that question. The monthly one is where you settle it.
Monthly: two numbers, one decision
The monthly meeting reads cost per qualified conversation and channel contribution, in that order. It exists to answer one question: where does the next dollar go. Once those numbers are on the table, the split between brand and demand spending becomes the real agenda. That argument belongs to demand generation strategy, where we have already made it. Bring the question here. Take the answer there.
One more boundary, because our own site has a page that could be mistaken for this one. Our marketing operations guide covers the campaign lifecycle: the twelve weeks from brief to retro, the intake form, the handoff SLAs, the week-seven midpoint. That clock starts when a campaign starts and stops when the retro is written. What this page covers is the standing rhythm underneath it, the one that runs whether or not a campaign is in flight. The two touch in one place, when a campaign retro feeds the monthly channel read, and nowhere else. Read that page for how a campaign runs. Read this one for what your function measures every week and every month.
The Two Numbers That Should Govern B2B Demand Generation Spend
Every function I have seen with a stable budget runs on two numbers. Everything else on your dashboard is diagnostic.
Cost per qualified conversation
Add up every cost the program incurred in a month: media, people, tools, agency fees, the content spend you would otherwise call fixed. Divide by the number of qualified conversations in the same month. A conversation is defined on the sheet above and evidenced by a note in the CRM.
That is the whole formula, and it has one uncomfortable property. There is no published benchmark for it. I looked. The market has a published cost per meeting, cost per qualified lead, cost per opportunity and cost per dollar of pipeline. None of them prices a conversation. The reason is that a conversation is defined inside a company, so it cannot be compared across companies. The absence is the argument for it. The moment you can benchmark it externally, you have defined it the way everyone else did. That is how the MQL happened.
So the first number you compute becomes your baseline. The only valid comparison after that is your own last quarter on the same definition. When the number moves and the definition held, you have learned something. When the definition moved, you have learned that the definition moved.
Pipeline created per dollar of program spend
Pipeline created in the period, divided by the program spend that produced it. Same window, same definition of pipeline. Some teams write the reciprocal and call it cost per dollar of pipeline; the arithmetic is identical.
There is real data on the shape of this ratio. LeadCoverage's Q4 2025 benchmark put the median at $4.84 of pipeline per dollar of go-to-market spend. The mean was $25.74, the range ran from $0.36 to $204.30, and the middle half of the distribution sat between $8 and $55. That sample is roughly 30 clients of one vendor in one industry, logistics. Its denominator is total go-to-market spend rather than marketing program spend, which makes it broader than your metric and narrower at the same time. Read it for the spread, which is the only thing it can honestly teach you.
Do not adopt $4.84 as a target. Do not adopt $8 to $55 as a range. The finding is the shape. This ratio varies by two orders of magnitude between companies doing roughly the same thing. Any single figure borrowed from it is a number about someone else. Compute yours and compare it with your own last four quarters.
Why MQL volume is the wrong headline
The arithmetic is unforgiving. One published 2026 aggregation uses stage rates it attributes to First Page Sage's paid-search funnel for B2B SaaS: 36% of leads to MQL, 26% of MQLs to SQL, 38% of SQLs to opportunity. Multiply those and 3.6% of leads reach opportunity. At $150 a lead, that is $4,217 per opportunity.
The same page admits that 2026 B2B cost-per-lead figures across datasets run from $66.69 to $524. That is a spread of 7.9×, because every dataset defines a lead differently and counts different costs. A number with a spread like that is not comparable between companies. It is barely comparable between two quarters at the same company.
Now rank your channels by cost per lead. In one 2026 aggregation the spread between cheapest and most expensive is about 9×, from $42 to $389. Rank the same channels by cost per opportunity and the spread is about 28×, from $221 to $6,118. The ranking changes. Channels that look cheap on leads look expensive once a stage conversion sits between the spend and the outcome. Content syndication is the clean case: $164 per lead, 6.8% to opportunity, $2,412 per opportunity. A reasonable buy at the first number, an obvious cut at the second. Treat that aggregation as an illustration of a mechanism, not an industry average.
This is what happens when MQL volume becomes the headline. The weekly meeting optimizes the number in the headline, so the program buys cheap leads. The rejected share climbs, sales trusts the pipeline less, and the meeting repeats with better attendance and worse information. Docket's own summary of its research puts the mechanism plainly. The MQL keeps getting produced because reporting systems and quarterly rhythms were built around it, not because anyone finds it informative. They sell a product that benefits from that conclusion, and the conclusion still holds.
Keep the two of them at the top. On your dashboard, every other metric exists to explain a movement in one of them. Our marketing KPIs piece covers the nine that change a decision. Yours probably carries forty that do not. If the two governing numbers have to survive a board conversation, marketing measurement is where we argue about what makes them reconcilable.
What to Cut First, in Order
Cutting is the part every guide skips, so here is the order I would use. You can stop each item this month without buying anything.
- Anything you cannot see downstream. Lead lists, content syndication with no stage visibility, a partner program that reports clicks. If you cannot compute cost per qualified conversation for a channel, you are not running it, you are paying for it. Cut it or instrument it before the next invoice.
- The weekly budget reallocation. Kill the agenda item, not the meeting. Move the decision to the first working day of the month and give it a real pack.
- MQL volume as a headline. Keep the field and the definition, remove it from the top of the report. Put pipeline created there instead and watch how fast the conversation changes.
- Gated assets on informational queries. If the page answers what something is or how it works, gate nothing. Save the gate for what a buyer needs before a call: a pricing comparison, a migration plan, a template they will use in a meeting.
- Channel reporting on cost per lead. Delete the column. A cost-per-lead table rewards the wrong channels, and it will be rebuilt within a quarter unless you remove it now.
- The dashboard nobody opens. If a report has not changed a decision in two months, it is costing you the credibility of the ones that do.
Six is enough for one quarter, and you should resist adding a seventh. Do the first two and the rest get easier, because the meeting stops rewarding volume and starts asking for conversion.
None of this needs a new platform. It needs one sheet both teams sign, one thirty-minute meeting a week with three numbers on it, one monthly meeting where the money actually moves, and the discipline to leave the budget alone in between. The first month you run your B2B demand generation numbers this way, they will look worse than the ones you reported last quarter. That is because they will be real. That is also the moment a lot of teams go back to the MQL count. A year later, the ones who did not are the ones whose pipeline you are trying to copy.
