The question sounds like a retreat icebreaker: who owns revenue, sales or marketing? In 2026 it has become an org-design question with measurable consequences, and the answer arriving across mid-market B2B is structural rather than rhetorical. Thirty-eight percent of mid-market CMOs now report into the CRO or revenue line, up from 12% in 2021, and among the remainder, more than half are evaluating the switch before 2027 (Gartner CSCO 2026). Meanwhile 71% of marketing budgets now carry direct pipeline metrics rather than awareness metrics. The handoff model that defined B2B for two decades — marketing generates, sales converts, and the wall between them is the MQL — is dissolving in real time. This piece walks the evidence for why, maps the five operating models replacing the relay, and gives leaders a selection framework that does not require copying someone else's org chart.
The relay model is dying of a measurement problem before it dies of a political one. Its core artifact, the MQL-centric service-level agreement, assumed a stable definition of what marketing hands over. In practice, organizations redefine MQL an average of 1.8 times per year, and every redefinition re-litigates the handoff dispute from scratch: sales claims the bar dropped, marketing claims sales stopped working the queue. In 2026, 54% of B2B organizations have moved marketing's primary measurement from MQL volume to opportunities and ARR contribution (HubSpot State of Marketing 2026). That migration is not a metric fad; it is a structural admission that counting leads across a wall no longer describes how revenue forms when 94% of buying groups shortlist vendors before ever talking to either function.
The cost of leaving the dispute unresolved is quantified. Organizations without a shared pipeline SLA are 34% less likely to hit quota; organizations that operate one — with weekly joint pipeline reviews rather than monthly lead-metrics reviews — source 46% of pipeline from marketing, versus 21% for unaligned peers (Pavilion 2026 CMO/CRO Alignment Survey). Read those two numbers together and the ownership question reframes itself: it is not which function owns revenue, but whether the two functions share one definition of pipeline at all. The org-chart migrations follow that realization — you move marketing under the revenue line because that is where the shared definition lives, not because marketing lost an argument.
What replaces the relay? Forrester's 2026 research identifies five operating models now running in production. The relay survives in longer-cycle industrial sales where handoffs are genuinely sequential. Centralized RevOps puts one owner — increasingly a VP-level hire — across the full funnel, with marketing, SDR, and AE capacity planned as one budget. Revenue pods group a marketer, an SDR, and an AE per ICP segment or vertical, each pod carrying its own pipeline number; pod penetration grew 27% year over year, the fastest of the five models, because it localizes the alignment problem to a team that shares a standup. Growth-led models let product-qualified leads drive the funnel with marketing and sales in support, viable where self-service revenue exceeds roughly a third of total. Hybrid SLA models run dual tracks — inbound MQL flow for the transactional tier, ABM motions for the enterprise tier — with different definitions legitimately coexisting because they serve different motions (Forrester Revenue Operations 2026).
Selecting among the five is a matching exercise, not a maturity ladder. Four variables do most of the discriminating work. ICP concentration: if your plausible account universe is under roughly 500 accounts, pods or centralized RevOps aligned to ABM fit; at 50,000 accounts, the relay or growth-led models fit because volume motions still work. ACV band: below $10K, growth-led and self-service economics dominate; above $100K, committee-based selling rewards pods and centralized planning. PLG revenue share: past roughly 35%, the growth-led model stops being optional because the product is already the top of funnel. Sales-cycle length: cycles under 60 days tolerate loose SLAs; at 120+ days — which is now the mid-market median at 121 days — the definitional disputes compound every month a deal ages, which is why long-cycle teams migrate first. Score yourself honestly on those four and usually only one or two models remain plausible.
Compensation has to move with the model, and this is where most migrations quietly fail. Under the relay, marketing is bonused on MQL volume and sales on closed revenue, which makes the handoff dispute rational on both sides. Under pods, the pod carries one pipeline number and the incentive splits three ways — a common structure is 40% on pod pipeline creation, 40% on pod bookings, 20% on individual craft objectives — which converts the SDR from a lead-quota role into a pod-member role and converts the marketer from a volume role into a coverage role. Under centralized RevOps, the change is subtler: the VP owns one number end to end, and both functions inherit it, which is why the centralized model concentrates political risk in a single hire and why boards increasingly ask what happens to the model when that person leaves. Decide the incentive design in the same meeting as the org design, because every month the two spend unaligned, the old dispute regrows.
Tooling consolidates along the same seam. The relay era ran a marketing automation platform, a CRM, and an SDR sequencer with three definitions of a lead living in three systems; every model that replaces it requires one shared definition layer, which in practice means a single account spine — the named account universe with committee roles and engagement state attached — that all three tools read from rather than argue with. The 2026 pattern in mid-market is deliberately conservative: keep the CRM as the system of record, retire duplicate lead definitions, and spend the integration budget on the account spine rather than on another dashboard. Teams that skip the spine and buy dashboards instead discover that the dispute did not disappear; it just gained charts.
The migration itself fails predictably when it fails, and the failure mode is cadence, not structure. Teams that change reporting lines but keep the monthly lead-metrics review discover that a new org chart cannot arbitrate an old ritual; the disputes simply relocate. What works is unambiguous in the data: replace lead-metric reviews with weekly joint pipeline reviews walking the same account list, measured on the same definitions, chaired alternately by sales and marketing. The cadence change is also the cheapest experiment available — several mid-market teams run it for six weeks before touching a single reporting line, and the pipeline-review discipline alone typically closes a third of the alignment gap.
The measurement stack that supports the migration is smaller than vendors imply. Five artifacts carry almost all the weight: one account universe file, one pipeline-definition document that both function heads sign quarterly, one weekly review ritual with a standing agenda, one coverage dashboard showing committee-touch state per account, and one incentive sheet that pod members can compute by hand. Everything else — attribution platforms, intent feeds, forecasting AI — is leverage on top of those five, and none of it rescues a team missing any of them. The 27%-a-year growth of the pod model is partly explained by this: pods are small enough that five artifacts are maintainable by hand, which is exactly why they stay aligned after the launch enthusiasm fades.
A worked selection makes the framework concrete. Consider a $20M vertical-software company: 900 plausible target accounts, $85K ACV, zero PLG revenue, 130-day cycles. Concentration and ACV point at pods over relay; cycle length demands the shared-definition discipline that pods institutionalize; growth-led is disqualified by the absence of product-qualified volume. The build sequence that follows: hire or anoint the pod leads, allocate accounts to pods rather than territories, move the SDR team into pods, and centralize only data and ops tooling — the shared account map, the CRM definitions, and the pipeline review ritual. The ninety-day plan writes itself: weeks one to three map the account universe and freeze definitions; weeks four to eight run pods in shadow against the legacy structure with both incentive plans live; weeks nine to twelve cut over fully and retire the MQL report. Ninety days later the org chart catches up to the operating reality, usually with marketing's demand function distributed into pods and brand remaining centralized.
None of the five models functions without one shared asset: a single map of the accounts both functions are working, with committee roles and engagement state attached. That is not a metaphor. Relay, pods, centralized RevOps, growth-led, and hybrid all degrade into the same failure without it — two functions optimizing different lists and discovering the overlap in lost deals. This is the layer Salebrate operationalizes: one account map with buying-committee roles and per-node engagement, so whichever operating model you choose, the denominator both functions share is the actual list of companies trying to decide whether to buy from you.
The 2026 answer to who owns revenue is therefore procedural: revenue is owned by whichever function can see the whole account, and the org design that wins is whichever of the five models your four variables select. Pick the model deliberately, migrate the cadence before the chart, align the incentives in the same meeting, and give both functions one map to argue over — the argument is the alignment.