Quota Setting: Top-Down vs Bottoms-Up Reconciliation in 2027

The scenario that breaks most planning cycles

It is the second week of November. The board deck is due in eighteen days, and the CEO has already told the lead investor the company will commit to $40M in net new ARR next year. The CRO nods in the room, then walks back to their desk and pings RevOps: "Build me the model that gets us there."

That sequence — number first, model second — is how the majority of annual planning cycles actually run, and it is the origin of nearly every quota failure that shows up nine months later. The model does not produce the number; the model is reverse-engineered to justify it. RevOps opens a spreadsheet, divides $40M by a headcount plan that assumes everyone is hired on January 2nd and fully productive by February, and lands on a per-rep quota that looks defensible in a cell but is fiction in a territory.

Here is what actually happens next in that company. The plan distributes quota to 34 AEs. Six of those seats are open on January 1 and will not be filled until March. Four of the reps carrying full quota were hired in October and have closed nothing yet. Two territories were re-cut in December, so the reps in them lost their pipeline and are effectively new hires without the ramp allowance. The productivity assumption baked into the per-rep number is the *top quartile* figure from the prior year — because that was the number sitting in last year's plan doc, and nobody re-derived it from actuals.

By the end of Q1, bookings are 22% behind plan. The CRO's explanation is "slow start, we'll catch up in Q2." By the end of Q2, the gap is 31% and the board has stopped accepting the catch-up narrative. The comp plan pays out to almost nobody, so the three strongest reps — the ones who *were* hitting numbers — start taking recruiter calls, because a rep at 78% attainment with no accelerator earned looks around and realizes their W-2 will be 60% of OTE. The company loses its best sellers in the worst quarter, which mechanically worsens Q3 and Q4.

The failure was not execution. The failure was that no one ever built the second model. Nobody sat down and asked: given the reps we will actually have, in the seats they will actually occupy, on the ramp curves our last twelve hires actually followed, what can this organization book? That number — the bottoms-up number — might have been $31M. Had it existed on paper in November, the conversation would have been an *eight-million-dollar gap discussion with four named levers*, held in a conference room with the CFO, rather than a two-quarter erosion of credibility followed by a leadership change.

The broader point extends past sales quota. The same top-down/bottoms-up tension runs through every capacity-constrained plan in the revenue org: the marketing team handed a pipeline target with no math on how many campaigns, events, or partner motions it takes to generate it; the CS org given a gross-retention goal without a model of how many accounts each CSM can meaningfully touch; the SDR team assigned meeting quotas derived from what sales needs rather than from dial-to-connect-to-meeting conversion rates on a real list. In each case, the discipline is identical — build the capacity-out model, put it next to the target-in model, and force the gap into the open.

How the reconciliation mechanism actually works

Reconciliation is not a negotiation between two opinions. It is the deliberate construction of two independent models that are *never allowed to see each other* while being built, followed by a structured comparison. The independence matters more than the arithmetic. If the person building the bottoms-up model knows the top-down answer, they will unconsciously tune the productivity assumption until the two match, and you have produced one model wearing two hats.

The top-down side works backward from the operating plan. The starting input is not the revenue commit — it is the *composition* of that commit. Take a company with $125M in base ARR committing to $40M net new. If net revenue retention is assumed at 108%, expansion contributes roughly $8M against a gross churn assumption of 12% that costs roughly $15M. The new-logo ARR the sales team must actually source is therefore commit plus churn minus expansion: $40M + $15M − $8M = $47M. Most plans skip this decomposition entirely and hand the AE team the $40M gross figure, which is how a $7M hole appears in Q4 that nobody modeled.

Then you apply the coverage multiplier. You cannot distribute exactly the number you need, because not every rep hits their number. If historical attainment says roughly half the team lands at or above 100%, you must distribute quota summing to more than the target. Coverage in the 1.15x–1.30x range is the common band: 1.15x expresses confidence in the plan and a stable team; 1.25x is typical for a growth-stage org with a mixed-tenure roster; 1.30x is what you use when attainment history is weak or the ICP is unproven — and you should expect the sales team to push back hard, because a high multiplier is mathematically a statement that you expect most of them to miss. On the $47M example, a 1.25x multiplier means distributing about $58.75M of quota across the seat plan.

The bottoms-up side starts from the individual and never looks at the commit. Three inputs drive it: the productivity assumption per fully-ramped rep by segment, the ramp curve applied to every partially-ramped seat, and the attainment factor drawn from trailing performance. Multiply, sum, and you have expected bookings — not distributed quota, but the dollars this specific roster is likely to produce.

The comparison is where the work happens. Distributed quota and expected bookings are different units, so you normalize: compare the top-down *target* to the bottoms-up *expected bookings*, both as booked-dollar figures. A healthy plan has top-down running modestly above bottoms-up — enough to be a stretch, not so much that the comp plan is designed to fail. When the gap is under about 10%, you likely sandbagged and left growth on the table. When it exceeds roughly 15%, you have built a plan whose predictable outcome is missed accelerators, mid-year re-forecasting, and voluntary attrition among precisely the people you cannot afford to lose.

The loop in that diagram is the important part. Reconciliation is iterative — you apply a lever, rebuild the affected side of the model, and re-check the gap. What you must never do is apply the lever *inside* the spreadsheet by quietly raising the productivity assumption until the gap closes. That is not reconciliation; that is a wish with a formula attached.

Real numbers, ranges, and the benchmarks worth arguing about

The single most-disputed input in the entire cycle is the productivity assumption: the annual new ARR a fully-ramped rep produces. Everything downstream inherits its error, so it deserves the most rigor.

The correct source for the productivity assumption is the *median fully-ramped rep's trailing-four-quarter actual*, segmented. Not the average — the average is dragged upward by one or two outliers and downward by anyone mid-PIP. Not the top performer — building a plan on your best rep's output is how you get a roster of people who all miss. Median, fully-ramped only, trailing four quarters.

Productivity scales with average contract value, but not linearly, and attainment generally falls as deal size rises. As a rough shape: an SMB AE working small-ACV transactional deals ramps in roughly a quarter and carries a lower per-rep number at higher velocity; a mid-market AE typically needs four to six months to reach steady state; an enterprise AE working six-figure deals often needs six to nine months; and a strategic or named-account seller working the largest deals can need nine to twelve months before their first closed-won is even plausible. The per-rep number rises across that progression while attainment rates fall, because larger deals mean fewer shots on goal and a single slipped deal can move a rep from 110% to 70%.

The discipline on year-over-year change: cap the productivity uplift at something modest — roughly 10% — unless you can name the *specific* structural change driving more. "The team will be more experienced" is not a reason. "We are raising list price 9% on January 1 and 70% of the pipeline will transact at the new price" is a reason. "We are cutting territory count from 4.2 accounts per rep to 2.8" is a reason. "We are adding a second product that attaches to 30% of new logos at a 22% ACV lift" is a reason, and it should carry its own line in the model rather than being smeared into the base assumption.

Ramp is where bottoms-up models lie most often. A graduated quota curve by month in seat is the standard structure: minimal or zero expected production in the first quarter while the rep learns the product and builds pipeline; something in the 40–60% range in months four through six as first independent deals close; 70–85% in months seven through nine; full quota from month ten. The critical modeling step is that a rep's *capacity contribution* is the area under that curve across the planning period, not their annual quota. A rep starting January 1 on a six-month ramp contributes roughly three-quarters of a fully-ramped rep's annual capacity. A rep starting July 1 contributes closer to a quarter. Planning headcount by hire date without integrating the ramp curve over-credits the model by 15–25% — and that error alone accounts for a meaningful share of Q1 misses across the industry.

Worked example, mid-market, 30 AEs, using a $1.2M productivity assumption:

• 15 fully-ramped reps × $1.2M = $18.0M of raw capacity
• 8 reps mid-ramp, averaging 70% of full-year contribution × $1.2M = $6.72M
• 7 planned new hires, averaging 35% of first-year capacity × $1.2M = $2.94M
• Raw capacity total: $27.66M
• Apply a 52% historical attainment factor → ≈$14.4M expected bookings

That last multiplication is where people flinch, and it is worth being precise about what it means. Raw capacity is the sum of quota you could distribute. Expected bookings is what the roster is likely to actually produce. If your top-down distributed-quota figure is $58.75M against $27.66M of raw capacity, you are not 15% apart — you are proposing quotas roughly double what capacity supports, and no comp design rescues that. The gap is the finding. Surfacing it in November is the entire value of the exercise.

Benchmarks the board will already know, and that you should be ready to speak to without notes: magic number (net new ARR divided by prior-period S&M spend, annualized), payback period on fully-loaded customer acquisition cost, quota-to-OTE ratio (commonly in the 4x–6x range for mid-market and higher for enterprise — below roughly 4x you are overpaying relative to production, and well above the band attainment tends to collapse), and percentage of reps at or above 100%. On that last metric: when fewer than about 40% of reps hit plan, the honest diagnosis is that the comp plan and quota are broken, not the reps. Firing your way to attainment when the quota is mathematically unreachable replaces experienced sellers with new hires who then need six months of ramp, which makes next year worse.

One adjacent number worth modeling alongside quota: pipeline coverage. A quota is a claim about bookings; bookings require pipeline; pipeline requires marketing and SDR capacity. If the plan needs 3.5x coverage entering each quarter and marketing's budget supports 2.6x, the quota model is already dead and the reconciliation meeting should catch it. This is why the head of marketing belongs in the room.

Trade-offs: one number, two numbers, and what each choice costs

The most consequential structural decision in quota setting is not the size of the number — it is how many numbers exist.

The single-number approach. One figure, committed to the board, cascaded to the field, anchored in the comp plan. Its virtue is clarity: everyone knows the target, forecast conversations are unambiguous, and there is no confusion about what "hitting plan" means. Its defect is that a single point estimate always becomes the floor in the next re-forecast. If you commit $40M and deliver $40M, the next plan starts at $40M and grows from there, regardless of whether the $40M was achieved by pulling deals forward or discounting to close. Single-number plans also create a perverse political incentive: the CRO is rewarded for negotiating the number *down* in November, because the only visible measure is attainment against it.

The two-number system. Internally you carry a commit and a stretch; externally the board sees one. The commit sits near the bottoms-up figure and carries high confidence — it is what capacity planning, hiring, and cash modeling anchor to. The stretch sits near the top-down figure and carries genuine uncertainty — it fuels the investor narrative and the upside case. Comp design bridges them: accelerators begin above 100% of commit and scale toward a meaningful multiplier as a rep approaches the stretch figure. This structure gives the CFO something defensible to model cash against while preserving the upside, and it removes the sandbagging incentive because both numbers are visible to the same audience.

The cost of the two-number system is real: it requires discipline nobody enjoys. If stretch is just "commit plus 20%" with no itemized math, the board will quietly treat stretch as the real commit by Q2 and you have made things worse. Stretch has to be built as commit *plus a named, dollar-quantified stack of conditional assumptions*: the pricing increase lands in Q2 (worth $2.1M), the new module reaches GA in Q3 (worth $1.4M against a 30% attach assumption), the two strategic renewals convert to multi-year with expansion ($3.0M). Each line has an owner and a trigger date. If a line does not land by its date, it comes out of the stretch number publicly at the next QBR.

A third option worth naming: banded quotas with quarterly true-ups. Rather than one annual number set in November and frozen, you set the annual plan with explicit quarterly checkpoints where quota is adjusted for actual hiring, actual ramp, and actual territory changes. The advantage is accuracy — the plan tracks reality instead of diverging from it. The cost is that reps hate mid-year quota changes, and any adjustment mechanism that only ever moves quota *up* destroys trust immediately. If you band, the band must be symmetric and the rules must be written down before the year starts.

Whichever structure you pick, pick it *before* the reconciliation meeting rather than during it. Choosing the structure in the same session where you are arguing about the gap guarantees that the structure gets chosen for tactical reasons — whichever framing makes the current gap look smallest — rather than on its merits.

Pitfalls, and the operating rhythm that prevents them

Building the models in sequence rather than in parallel. The single most common process failure. If bottoms-up is built after top-down is known, it is contaminated. Assign the two models to different owners, have them built against a shared data set but without visibility into each other's output, and reveal both in the same meeting. This feels bureaucratic and it is worth every minute.

Using a productivity assumption that was never re-derived. Plan documents are copied year to year. The $1.4M in this year's model is frequently the $1.4M from three years ago, when the ICP, price book, and sales motion were all different. Every planning cycle, re-pull eight quarters of per-rep new ARR by segment and tenure and recompute the median from scratch. If the new median is materially below the old assumption, that is not a data problem to be smoothed over — that is the finding.

Ignoring tenure mix. A team that will be 40% new hires next year cannot produce last year's per-rep number, no matter how good the hires are. Ramp is arithmetic, not attitude.

Pulling forward uncontracted expansion to shrink the new-logo number. It makes the top-down math look easier in November and it is caught in the first quarterly review. Finance sees it immediately, and it costs more credibility than the gap it hid.

Modeling attrition at zero. Sales orgs lose people. If trailing voluntary and involuntary attrition runs at, say, 20% annually, the plan must assume seats go empty and take time to backfill, with the replacement re-entering the ramp curve at month one. A model with 30 seats occupied for 12 months each is not a model of a sales team.

Running the reconciliation meeting with the wrong people. The room needs the CRO defending top-down, the VP of Sales defending bottoms-up as the capacity owner, RevOps owning both models and the reconciliation itself, finance controlling the headcount unlock, and marketing owning the pipeline coverage assumption. Without finance, no lever involving headcount can actually be pulled in the room. Without marketing, you will commit to a bookings number with no pipeline math behind it and discover the shortfall in Q1.

Leaving the reconciliation undocumented. Produce a one-page defense memo, signed by all four functions: the top-down commit with its NRR and churn decomposition; the bottoms-up capacity with productivity, ramp, and attainment assumptions stated explicitly; the gap and the specific lever used to close it; three named risks that would invalidate the model, each with a trigger condition; and the quarterly true-up checkpoints. The purpose of this document is not compliance. It is that when Q2 comes in light, the conversation is "risk two triggered, here is the pre-agreed response" rather than "why did you miss." One converts a miss into a planned variance; the other converts it into a leadership change.

The operating rhythm. A workable cadence runs roughly ninety days. The first thirty are audit: pull eight quarters of per-rep production, measure the *actual* ramp curves your last dozen hires followed rather than the assumed ones, and document the delta between your assumed attainment and your realized attainment. The next thirty are construction: build both models in parallel, then sensitivity-test each — flex productivity ±10%, ramp ±30 days, attainment ±5 points — and show finance the resulting spread rather than a single figure. A model that produces a range is more credible than one that produces a point. The final thirty are reconciliation and defense: hold the meeting with all five roles present, apply exactly one lever, sign the memo, build the board narrative in three parts (how we derived the number, what has to be true, what we do when it slips), and pre-wire the board chair and lead investor with the math before the formal meeting so the first time they see the gap is not in front of the full board.

The question every sophisticated board member eventually asks is simply: "What's the bottoms-up number, and how big is the gap?" A leader who answers it cleanly — naming both figures, the lever, and the true-up date — buys two quarters of trust. A leader who answers "we're aligned around the number" without showing the derivation gets a board observer assigned before the next meeting.

Related questions

How is quota setting different for a first-year sales team with no history?

With no trailing data, borrow segment benchmarks for the productivity assumption, discount them 20–30% for the absence of a proven playbook, and set an explicitly provisional quota with a written 90-day true-up. Model the ramp longer than you think. Commit to the board in ranges, not points.

Should SDR and CS quotas go through the same reconciliation?

Yes, with different inputs. SDR capacity builds from activity-to-meeting-to-opportunity conversion rates on a real list; CS capacity builds from accounts per CSM and touch frequency. Both should be modeled bottoms-up and compared against the pipeline and retention targets handed down from the plan.

What happens to quota when territories are re-cut mid-year?

A re-cut territory resets a rep's pipeline, so treat the affected reps as partially re-ramping and adjust their capacity contribution accordingly. If quota does not move, you have silently raised it. Document the adjustment rule before the re-cut, not after.

How does pipeline coverage connect to the quota model?

Quota is a bookings claim; bookings need pipeline. Multiply the quota by the required coverage ratio to get the pipeline generation target, then check it against marketing and SDR capacity. If pipeline capacity cannot support the coverage, the quota is unreachable regardless of rep quality.

Is a high coverage multiplier a sign of a bad plan?

Not automatically, but it is a statement. A 1.30x multiplier says you expect most of the team to miss. That may be honest given weak attainment history, but it also means the comp plan must pay meaningfully below 100% or you will lose mid-attainers who are performing acceptably.

FAQ

What is the difference between top-down and bottoms-up quota setting?

Top-down starts with the revenue target the board or operating plan requires and divides it across seats, usually applying a coverage multiplier so distributed quota exceeds the target. Bottoms-up starts with the individual rep — productivity assumption, ramp stage, expected attainment — and rolls upward into expected bookings. Top-down describes what the business needs; bottoms-up describes what the roster can produce. Reconciliation is the structured comparison between them.

What is a healthy gap between the two models?

Roughly 10–15% is the commonly cited band, with top-down running above bottoms-up. Under 10% suggests you sandbagged the plan and left achievable growth unclaimed. Over 15% means the comp plan is mathematically designed for widespread misses, which produces low payouts, disengagement, and attrition among strong performers — a cost that frequently exceeds the value of the ARR you were reaching for.

How do I calculate rep capacity for the bottoms-up model?

Take the median fully-ramped rep's trailing-four-quarter new ARR for each segment. Apply a graduated ramp curve to every seat based on month-in-seat during the planning period, taking the area under that curve rather than the annual quota. Sum across the roster to get raw capacity, then multiply by your historical attainment factor to get expected bookings. Model open seats and expected attrition explicitly.

What if I cannot close the gap between the two models?

Four levers exist: add capacity, raise productivity with a named structural cause, increase pipeline with committed budget, or lower the commit. If none closes the gap credibly, present the bottoms-up figure as the commit and the top-down figure as an explicitly conditional aspiration with itemized assumptions. Boards generally prefer a credible lower number to a fantasy higher one that resets in Q3.

How should the reconciliation be presented to the board?

Three parts. First, the derivation: top-down decomposition including churn and expansion, bottoms-up build including productivity, ramp and attainment, and the resulting gap. Second, what has to be true — five named assumptions with honest confidence levels. Third, the slip plan: specific early-warning triggers and the pre-agreed response to each. Transparency about the gap earns more latitude than confidence without math.

What are the most common mistakes in reconciliation?

Building the two models sequentially so the second is contaminated by the first; reusing a productivity assumption that was never re-derived from actuals; ignoring ramp for new hires and tenure mix across the team; assuming zero attrition; pulling forward uncontracted expansion to shrink the new-logo requirement; and closing the gap by quietly editing an assumption inside the spreadsheet rather than pulling a named, owned lever.

Sources

• https://hbr.org/2017/07/how-to-set-sales-quotas-that-motivate-your-team
• https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights
• https://kellblog.com/2017/01/16/the-quota-capacity-model/
• https://www.saastr.com/category/sales/
• https://openviewpartners.com/blog/
• https://www.bridgegroupinc.com/research
• https://www.salesforce.com/resources/articles/sales-quota/
• https://www.gartner.com/en/sales/topics/sales-performance-management
• https://www.bain.com/insights/topics/sales-and-marketing/
• https://a16z.com/tag/enterprise/

Related on PULSE

• [How to build a sales capacity model that finance will actually sign off on](/knowledge/ra0640)
• [Ramp curves: how long should a new AE really take to reach full quota](/knowledge/ra0641)
• [Designing accelerators when attainment is below 50%](/knowledge/ra0642)
• [Pipeline coverage ratios: what 3x actually has to mean](/knowledge/ra0643)
• [Territory design and its hidden effect on quota attainment](/knowledge/ra0644)