Module 4 · The Operations Disciplines · Chapter 13 of 32
Pipeline Management and Deal Velocity
A stage is a set of facts that are true or not, never a feeling. Run the pipeline on observable exit criteria, read the two conversions that decide the quarter, and clear zombie deals before they book you a miss you never saw coming.
10 min read
The deal had been in Negotiation for two quarters when I finally asked the rep what Negotiation meant. She said the buyer was “really close.” I asked what had to be true for any deal to sit in that stage, and got five different answers from the five people in the room, none of which matched the field help text nobody had read. That opportunity had no logged activity in six weeks, a close date that had slid three times, and a champion who had left the company in March. It was still counted at full amount in the CRO’s commit. On that team, stage was a feeling, and the forecast was a feeling with a dollar sign in front of it.
The Go-To-Market Model handed operations a shortlist: the two or three motions the math allows, and the cost each one is permitted to carry. This chapter is where the High and Medium Touch motions on that shortlist live or die, on the board, day to day. Choosing a motion is a quarterly decision. Running its pipeline is a Tuesday. The question stops being which motion and becomes which deals, moving how fast, through which gates.
Two conversions decide the quarter
The metrics spine numbered the Bowtie one through nine and gave every stage a volume, a conversion, and a time. Two of those conversions sit at the knot where deals are won or lost, stages 3 and 4, and they are the two you manage a pipeline on. CR3 is prioritization quality: the share of discovery meetings that clear the bar to become a qualified opportunity. CR4 is the win rate: the share of qualified opportunities that reach mutual commit. The spine already showed they compose, OTC = CR3 × CR4, opportunity to close. Managing the pipeline is reading them apart.
Read either one alone and it will lie to you. A high win rate looks like health until you notice the gate upstream is rubber-stamping every discovery call, so the win rate is high on a tiny, over-served set while faster deals wait. A tight gate looks like discipline until you notice the deals that survive it still lose, which means the problem was never the funnel. The two numbers only tell the truth as a pair, which is why Revenue Architecture arranges them into a diagnostic grid rather than two separate KPIs.
| Win rate (CR4) \ Gate (CR3) | Loose gate: nearly every discovery meeting is marked qualified | Tuned gate: 60 to 80% qualify | Tight gate: under 60% qualify |
|---|---|---|---|
| Strong, above 22% | Over-qualifying. You win most of what you accept, but you accept everything, so sellers over-serve marginal deals. Raise the bar and reclaim the time. | The operating band. Hold it and self-benchmark quarter over quarter. | Healthy but starved. You win a high share of a thin set. The constraint is volume, so feed more discovery in. |
| Fair, 18 to 22% | Volume masking quality. A full funnel converting at an average clip. Tighten CR3 and watch the win rate climb on its own as junk stops entering. | Normal. Coach the middle of the team toward the top row. | Fine gate, thin top. Push discovery volume before touching anything else. |
| Low, under 18% | No gate at all. Reps advance everything to look busy, and the pipeline is mostly junk. Fix CR3 first; this is an operations and qualification problem. | A selling problem emerging. The gate is fine, the close is not. Route it to enablement, not to the funnel. | Pure selling problem. A strict gate and still losing means the deals that survive should convert and do not. Coach the deal, not the pipeline. |
Same soft quarter, five different owners depending on which cell you land in. This grid is the logic behind a pipeline-conversion dashboard I keep for a sales leader: stage-to-stage conversion by segment, with CR3 and CR4 broken out per rep, so a bad quarter routes to the desk that owns the weak cell instead of landing on “the funnel” as a group failing. When win rate falls and the gate is loose, the fix is a qualification standard, which is ops. When win rate falls behind a tight gate, the fix is coaching, which is enablement. The matrix keeps those two from being confused, because they get confused constantly.
Deal velocity is the whole engine in one number
CR3 and CR4 tell you where the fix goes. Deal velocity tells you what the engine is producing while you fix it. It folds the two conversions together with deal size and time into a single number, dollars of pipeline that convert to revenue per day.
of pipeline throughput per day
The one metric that captures the whole engine. Move any lever and watch the daily number. Shortening the cycle and lifting win rate compound; adding raw opps is the slowest lever of the four.
of pipeline throughput per day: $29k
Four levers, and they are not equals in practice even though the algebra makes them look identical. Three of them (opportunity count, win rate, deal size) sit in the numerator and multiply the result the same way. The fourth, cycle length, sits in the denominator, so cutting it lifts velocity more than the same percentage move anywhere else, and it buys a second thing the formula cannot show: a rep who runs a 60-day cycle instead of 75 gets more shots on goal a year from the same seat. Cycle length is also the only lever operations owns outright, because it moves with configuration (exit criteria, dead-deal removal, routing latency, approval turnaround) rather than with rep talent. I keep the derivation out of the course arc on purpose; the full decomposition, and why “just add more opps” is the lever that lies, lives in Pipeline velocity has four levers.
A stage is a set of facts, not a feeling
The deal that opened this chapter was not a talent problem. It was a definition problem. Nobody had written down what Negotiation meant, so it meant whatever the rep needed it to mean on a Friday. The repair is stage-gates: each stage advances only when a set of observable facts is true, and each fact is a field a machine can check, not a judgment a human volunteers.
I built this as a stage-gate engine in Salesforce. The exit criteria for each stage live in custom metadata rather than hard-coded in a flow, an Apex service evaluates them when a rep tries to advance a deal, and a Lightning component shows the rep exactly which criteria are still unmet before the stage will save. A deal cannot reach Proposal until the fields that define Proposal are populated: an economic buyer named, a critical event dated, a mutual plan attached. The rep is not blocked by a person, they are blocked by facts they can see and go fill in. That is the difference between a gate and a nag.
- 1
Write each stage as observable facts
For every stage, list what must be true to be in it, phrased so a machine could check each item. "Champion engaged" is a feeling; "an internal Champion is named and has confirmed a critical event date" is a fact. If a criterion cannot be checked without asking the rep, it is not an exit criterion yet.
- 2
Map every fact to a real field
Each criterion points at a populated field, not a checkbox nobody validates. Keep the metadata separate from the logic so you can tune a gate without a code change, which you will do often in the first two quarters.
- 3
Enforce at the point of save, and show the gap
The gate fires when the rep advances the deal, and it names the specific criteria still missing. Silent enforcement breeds workarounds; a gate that explains itself gets respected.
- 4
Instrument conversion and time in stage
Track stage-to-stage conversion and how long deals sit in each stage. A gate that is too tight shows up as a conversion cliff; a gate that is too loose shows up as deals piling in a late stage and never leaving. The instrumentation is how you know which way to tune.
Zombie deals: clear them before the forecast counts them
Gates stop the pile from growing. They do not clear the deals already rotting in it. Roughly 30% of open pipeline has had no activity in 21 or more days, no email, no call, no meeting, and it sits green in the forecast anyway (Ebsta, Clari). A team walking in at 4x coverage with 30% zombies is really at 2.8x live, and the difference is a miss it has not booked yet. The reps know which deals are dead; they leave them open because an open late-stage deal flatters the pipeline while a closed-lost is a visible loss a manager reads. Until you change what is safer, the graveyard refills no matter how many times you sweep it.
One driver of the cycle-length lever deserves its own flag, because it is the slowest gate on most High Touch deals and it is not at the top of the funnel. It is the pricing approval at the very end, the deal desk queue where a discount waits on someone’s inbox. That turnaround is a real chunk of Δt4, and it is configuration you own, which means it is velocity you can recover. Deal desk gets its own discipline later in this module; the deal desk reference topic carries the standing playbook if you need it before then.
Handing off
You now have a pipeline that means what it says: stages defined as facts and enforced at the gate, the two conversions that decide the quarter read as a pair so the fix routes to the right desk, a velocity number you can defend against trailing bookings, and the dead deals cleared before they inflate coverage. That was the whole point of the discipline, because the next one cannot run without it. A forecast is only as honest as the stages underneath it, and every technique in the next chapter assumes the pipeline feeding it has already been cleaned. In Forecasting we turn this instrumented pipeline into a number leadership trusts, starting from coverage as a capacity question and ending at a calibrated call built from more than one input.