Planning season runs on a familiar script. The CEO wants 30% more pipeline next year. You take the number back to your team, divide it by this year's conversion rates, and the model says you need 40% more leads. The budget request follows the model. Nobody in the room asks what sales did with the leads you sent this year.
Pull one report before that request goes to finance. Count the leads marketing passed to sales last quarter that nobody touched within five business days. Those leads cost money to make, and they are aging in a queue in front of the slowest stage in your funnel.
My default is simple. If more than 20% of accepted leads sit untouched for five business days, stop buying volume. Your constraint is downstream, and every new dollar spent above it buys more of the same queue. Drop the threshold to 10% when your average deal is above $100K, because each abandoned lead costs more.
What Goldratt saw on the plant floor
Eliyahu Goldratt made this argument about factories in 1984. The Goal is a novel about Alex Rogo, a plant manager given three months to make a failing plant profitable (Wikipedia)). The theory underneath it holds that any system has at least one constraint, and the system's output is capped there (Wikipedia). An hour lost at the bottleneck costs total throughput equal to the bottleneck's hourly capacity, and the fastest way to find it is to look for the biggest pile of work stacked in front of a step (Shortform).
Goldratt turned the idea into five focusing steps. Identify the constraint. Exploit it, which means getting the most out of it with what you already have. Subordinate everything else to it, including how much work you release into the system. Elevate it with new capacity, and only after the first three steps are done. Then go back to step one, because the constraint moves (TOC Institute). Release work faster than the constraint can absorb and, in the TOC Institute's words, the result is "bloated WIP inventory, elongated lead times, and frequent expediting/firefighting" (TOC Institute). In a funnel, that is a lead queue and a lead-quality fight. Goldratt took the method to marketing himself in his 1994 sequel, It's Not Luck (Wikipedia).
I call the marketing version Spend at the Constraint. Find the stage that caps pipeline, put the next dollar there, and hold every other stage to the pace the constraint can absorb.
Where the pile usually sits
In a plant you walk the floor. In a funnel you pull the CRM. When a regional sales lead told me marketing had sent nothing in five weeks, I pulled it and found about 150 leads untouched in a queue that routing never put in front of the team. The market data says the pile usually sits there, downstream of marketing.
Response is the first place to look. RevenueHero submitted demo requests to 1,000 B2B SaaS websites and got 365 replies back, automated ones included. The average reply took 1 day, 5 hours and 17 minutes (RevenueHero, March 2024). Oldroyd, McElheran and Elkington reached the same finding in 2011, that most companies are "not responding nearly fast enough" to online inquiries (Harvard Business Review, 2011). A demo request is the highest-intent lead you can buy, and the replies that did come took more than a day on average.
Sales capacity is the second. Salesforce surveyed 5,500 sales professionals and found that non-selling tasks consume 70% of reps' time, and that 84% of reps missed quota the year before (Salesforce, July 2024). The Bridge Group's 2025 study of 351 B2B companies put 60% of SDRs at quota, the lowest in the study's history (The Bridge Group, February 2025). If reps spend 30% of the week selling, 40% more leads land in the same 30%.
Then check the other direction, because sometimes the pile is thin and sales is waiting. 6sense's 2025 survey of nearly 4,000 buyers found they choose one of the four vendors on their Day One shortlist 95% of the time, and they initiated 79% of contact with sellers (6sense, 2025). A company missing from that list has an upstream constraint, and faster follow-up will not reach a buyer who never fills in the form.
In Partnership with Attio
Frontier models perform better in Attio’s agent harness.
The Constraint Map
This is the page I bring into planning. Each row is a stage that can cap pipeline, the signal that says it is the constraint today, and where the next dollar goes. The defaults are my positions, set from the accounts I have run. Reset them to your own sales cycle.
Stage | It is the constraint when | Next dollar goes to | Default |
Day One shortlist | Reps have open calendars and inbound demo volume is flat | Reach into the buying group, category presence | Fund reach before more capture |
First response | Median reply to demo requests exceeds one business hour | Routing, instant scheduling, SDR coverage hours | One-business-hour SLA on hand-raisers |
SDR capacity | Over 20% of accepted leads untouched after five days | SDR hours or self-serve paths | Lead target equals worked capacity plus 10% |
AE capacity | First meetings slip past ten business days | Rep-free buying paths, sales engineering support | Move volume to self-serve tiers |
Win rate | Pipeline holds while close rate falls two quarters | Competitive proof, pricing clarity, sales content | Freeze new top-of-funnel spend |
Only one row is the constraint at a time. If three rows look guilty, take the one furthest downstream, because fixing anything above it feeds the same pile.
Exploit before you ask for money
Step two costs the least, so it goes first. A response constraint is usually a routing problem. Demo requests land in a shared queue, round-robin skips reps who are out, and nobody owns the gap between 5pm and the next morning. Instant calendar booking on the form, a named owner for after-hours requests, and a one-business-hour SLA on hand-raisers fix most of it without a dollar of media. Content downloads are the exception. Those buyers did not ask to talk, so a nurture sequence is the right response.
Exploiting a capacity constraint means protecting its hours. If AEs are the constraint, every unqualified meeting marketing books spends the scarcest hour in the system. Raise the bar on what counts as a sales-ready meeting, even though your meeting count drops. Who sets that bar is its own fight, which I covered in The RevOps Land Grab.
Set the lead target from the bottom up
The planning model in the opening runs top down. Revenue target, divided by conversion rates, equals leads. Goldratt's third step runs the other way. You start from what the constraint can absorb.
Here is the math with round numbers. Eight AEs can each run 25 first meetings a month alongside their open deals, so sales absorbs 200 first meetings. If 40% of accepted leads become a first meeting, the funnel can work 500 accepted leads a month. A plan that sends 900 buys 400 leads a month of inventory at full media cost.
My default caps the lead target at worked capacity plus 10%. Goldratt's own scheduling method keeps a buffer in front of the constraint so it never sits idle, and the 10% does the same job (Wikipedia). Go above the cap only when you are building a nurture pool on purpose, ahead of a launch or a new sales hire class, and write that into the plan so nobody mistakes it for pipeline.
In the example, the cap is 550 accepted leads a month. The media budget for the other 350 is what you bring to the reallocation conversation.
Elevate with marketing's budget
When sales capacity is the constraint, the reflex is to ask for more reps. That is the CRO's budget, and new SDRs take three months on average to ramp (The Bridge Group, February 2025). Marketing can raise the constraint faster by moving work off the reps.
Buyers will take that path. Gartner surveyed 632 B2B buyers and found 61% prefer an overall rep-free buying experience. The same buyers still want seller input on whether a product fits their company's needs (Gartner, June 2025). Public pricing, recorded demos, a self-serve trial, and an ROI calculator a buyer can run alone all take questions off the rep's calendar. The rep's hours then go to the fit conversation, which is the one buyers want a person for.
My default, when AEs are the constraint, moves a fifth of next year's lead-capture budget into that rep-free path. Deviate when your product cannot be bought without a scoping call, as in most deals with custom implementation. In that case, fund sales engineering support from the same reallocation and agree the swap with the CRO in writing.
When the constraint is the shortlist, elevating it means reach. The 95% figure from 6sense says most deals go to a vendor the buyer already had in mind on day one. Capture spend at that stage harvests the same small pool of buyers who already know you, and the next dollar belongs in getting known by the buying groups who do not.
The constraint moves
Goldratt's fifth step sends you back to the start. Fix response time and SDR capacity often shows up as the new limit. Fix that and the win rate gets exposed. Re-run the Constraint Map every quarter and expect the guilty row to change, which means the budget split should change with it.
Before your 2027 plan goes to finance, put one number at the top of the demand section, the leads sales can work per month. Then sit down with the CRO, agree which row of the map is the constraint today, and fund that row first.
James