Around this time every year, a finance partner sends marketing a number. It is usually expressed as pipeline, sometimes as sourced opportunities, occasionally still as leads, and it is almost always derived the same way. Take next year’s bookings target, divide by the win rate, adjust for the share of pipeline marketing is expected to generate, and there is marketing’s commitment. In many companies that number will prove wrong by the end of the second quarter.
The problem is rarely the arithmetic. It is the inputs. Most reverse-funnel models still run on conversion rates and cycle lengths that were measured two or three years ago and carried forward because nobody had a reason to revisit them. Since then buyers have changed how they research and sales teams have changed how they qualify. A model built on those rates produces a confident number with little connection to how the business now acquires customers.
I have sat on both sides of this negotiation, as the marketer handed the number and as the person signing off the plan. The commitments that hold up share a few characteristics, and none of them is a cleverer spreadsheet.
Start With What Happened in 2026, Stage by Stage
Before anyone agrees a target, have marketing or revenue operations pull the actual 2026 funnel for the trailing four quarters, and do it at the level of individual stage transitions rather than overall conversion. The aggregate figure, how many early-stage records eventually became closed revenue, hides almost everything useful.
What you are looking for is which transitions moved, in which direction, and why. In the analyses I have seen this year, a few patterns recur often enough to check for specifically.
Volume at the top has fallen while late-stage conversion has risen. Fewer people fill in forms, fewer inbound requests arrive from organic search, and fewer cold outbound sequences produce meetings. But the buyers who do engage tend to arrive later in their process and better informed, so the rate at which they progress to a qualified opportunity is often higher than it was. A model that applies the old top-of-funnel conversion rate to the new, smaller volume will understate pipeline. A model that applies the old volume assumption to the new, higher rate will overstate it badly. Both errors are common.
Sales qualification has tightened. Many sales organizations raised the bar for accepting an opportunity during 2025 and 2026, partly in response to forecast accuracy problems and partly because leadership grew tired of inflated pipeline. That is a healthy change, but it means an opportunity in 2026 is not the same unit as an opportunity in 2024. If your historical pipeline-to-close rate was calculated on the looser definition, it will understate the real win rate on today’s opportunities, and your model will ask for more pipeline than the business needs.
Cycle lengths have lengthened for larger deals. Buying committees have grown, security and procurement reviews take longer, and AI-related purchases in particular attract additional scrutiny from legal and risk teams. If the model assumes that pipeline created in the first half closes within the year at the historical rate, and enterprise cycles have stretched by a quarter or more, a meaningful share of that pipeline will land in 2028.
Source attribution has drifted. As buyers increasingly encounter vendors through AI assistants, peer communities, and marketplace listings, more opportunities arrive with no clear source or with a source field filled in by whichever rep created the record. The share of pipeline credited to marketing may have changed for reasons that have nothing to do with marketing’s performance.
Rebuild the Model Around Motions, Not Channels
The second change I would make is structural. Most reverse-funnel models treat pipeline as a single flow with a single set of conversion rates. In practice, a typical B2B company now acquires pipeline through several motions that behave very differently, and blending them produces averages that describe none of them.
A workable 2027 model separates at least these:
Inbound from new logos. Buyers who find you and raise their hand, through the website, a demo request, a free trial, or an inbound call. Volume has generally fallen, conversion has generally risen, and the cost per opportunity is usually the most sensitive to changes in search and AI visibility.
Target account programs. Opportunities created from a defined list of accounts through a coordinated combination of advertising, outbound, events, and executive engagement. Conversion rates here should be measured at the account level, not the contact level, and cycles tend to be longer.
Partner and marketplace sourced. Opportunities that arrive through resellers, systems integrators, or cloud marketplace co-sell. These often carry different win rates and deal sizes and should not be blended with direct inbound.
Expansion and cross-sell in the installed base. For many companies this is now the largest single source of new bookings, and it is frequently left out of marketing’s pipeline model entirely, or included with a flat assumption borrowed from the customer success plan. If marketing runs customer programs, product adoption campaigns, or user communities, this motion deserves its own conversion rates and its own target.
Each motion gets its own volume assumption, its own stage conversion rates, its own average deal size, and its own cycle length, all taken from 2026 actuals where the data allows. The output is still a single pipeline number, but it is now the sum of several defensible ones. When the year starts to diverge from plan, and it will, you can see which motion is responsible rather than arguing about a blended figure.
This also changes the budget conversation. When finance asks what a fifteen percent program cut would do, you can answer by motion. Cutting paid search hits inbound volume within a quarter. Cutting the target account program shows up two or three quarters later, mostly in large deals. Cutting customer marketing may never appear in marketing’s sourced number but will appear in net revenue retention.
Agree on the Unit Before You Agree on the Number
A large share of the end-of-year disputes I have seen between marketing and sales were not really about performance. They were about definitions that were never written down. Marketing believed it had delivered its pipeline commitment, sales believed it had not, and both were reading accurately from different reports.
Before the target is signed, agree in writing on four things.
What counts as pipeline. Which opportunity stage, which record types, and whether expansion opportunities count. If sales accepts an opportunity at one stage and finance recognizes it as pipeline at another, pick one.
How sourcing is determined. Whether the rule is first touch, the presence of a marketing interaction within a set window before opportunity creation, or a manual source field. If it is a manual field, who sets it and who can change it. The rule matters less than its being fixed in advance and applied by the system rather than by negotiation.
How influence is reported. Most marketing teams now report some measure of influenced pipeline alongside sourced pipeline, and they should, because so much of what marketing does affects deals it did not originate. But influenced pipeline should not be the committed number. It is too easy to inflate, and every CFO I have worked with knows it. Report it as context and commit to the sourced figure.
When the measurement happens. Pipeline created in the period, pipeline open at period end, or pipeline that eventually closes. These produce very different numbers, and a commitment made on one basis and reported on another will cause trouble.
Build in the Cycle Lag
The single most common way a sound marketing plan fails is timing. Marketing generates the pipeline it committed to, but too much of it arrives in the second half, and too little of that closes within the fiscal year. Sales misses its number, and the post-mortem concludes that marketing pipeline was poor quality, when in fact it was simply late.
The fix is to plan pipeline creation by quarter against the bookings it needs to support, using the cycle length for each motion. If enterprise deals now take nine months from creation to close, pipeline for third-quarter enterprise bookings needed to exist by the end of the previous year. That means part of marketing’s 2027 commitment is really about the first quarter, and some of it is already in the current pipeline or not at all.
The exercise usually reveals two uncomfortable facts. The first quarter of 2027 was largely determined by what marketing did in the second half of 2026. And the programs that produce enterprise pipeline need to be running in January, not approved in March. If your budget process releases program spend late, say so now and put the consequence in the plan.
Set a Range, and Name the Assumptions
A single-point pipeline target implies a precision nobody has. I would recommend committing to a base figure while documenting the range around it and the three or four assumptions most likely to move it. For example: inbound volume holds at second-half 2026 levels; the target account list stays at its current size and coverage; partner-sourced opportunities grow in line with the partner team’s own plan; sales acceptance criteria remain unchanged.
Writing the assumptions down does two things. It gives you a defensible basis for revisiting the number mid-year if one of them breaks, which is far better than explaining a miss after the fact. And it forces the other parties in the room to own their share. If the commitment depends on the partner team doubling marketplace-sourced deals, the partner team should be in the conversation. Agree a formal review at the end of the first quarter, when the model is re-run against actual conversion rates and the target confirmed or adjusted with finance. That beats both rigid adherence to a number everyone knows is wrong and a quiet rewrite that damages credibility.
Where AI Changes the Model and Where It Does Not
Most marketing teams will go into 2027 with more AI in their operations than they had a year ago, and there will be pressure to reflect that in the pipeline commitment. Be careful here.
AI tools have reduced the cost and time of producing content, campaigns, and outbound sequences. They have not, on the evidence I have seen, increased the number of buyers in market for what you sell or shortened their decision processes. In several categories outbound got worse, because every competitor automated at once and buyers stopped responding. A plan that assumes more pipeline because production capacity rose is assuming that the constraint was capacity. For most companies it was not.
Where AI does change the model is in the cost line and in the speed of adjustment. If the same pipeline can be produced with less agency spend or fewer production hours, that is a legitimate efficiency and belongs in the budget. And if better signal detection and faster campaign execution let you respond to changes in a motion within weeks rather than quarters, that is an argument for planning in shorter cycles with more frequent reviews. Neither is an argument for a larger number.
What to Take Into the Room
When the planning meeting happens, the marketing leader should be able to show four things on a page or two: the 2026 funnel by motion with the stage conversion rates that changed and the reasons; the 2027 commitment by motion and by quarter, with the cycle lag built in; the written definitions of pipeline, sourcing, and timing; and the named assumptions with the agreed review date.
That will not guarantee the number is hit. Nothing does. But it changes the nature of the conversation in September from a dispute over whose report is correct to a discussion of which assumption broke and what the business decided to do about it. Over a long career, I have found that difference matters more to how marketing is regarded by the executive team than whether any single year’s target was met.