Find the constraint, then fix only that
The whole point of seeing the funnel as a chain is that it tells you where to spend your attention, and for a solo or lean operator, attention is the scarcest resource you have. You cannot work on everything, so the question is which single thing, and the chain answers it.
List the rates, find the lowest
Write your five stage-to-stage conversion rates in order, down the page. Your end-to-end conversion is their product, so multiply them and you have your funnel's true throughput. Then circle the lowest rate. That is your constraint, the weakest link, and it is the only stage worth working this quarter. Everything downstream of it is being starved, because nothing past a bottleneck can flow faster than the bottleneck allows.
This is why pouring more leads into a funnel with a downstream constraint is not just wasted effort, it actively raises your CAC. If your MQL-to-SQL joint converts at 12 percent, every extra lead you buy mostly dies at that joint, so you pay full acquisition cost for leads that never reach the close. You have spent more to acquire the same number of customers, which is the definition of rising CAC. The fix is to repair the weak joint first, then open the tap, in that order, never the reverse.
The constraint is a diagnosis, not a volume problem
Take that 12 percent MQL-to-SQL rate as the worked example, because it is the most common trap I see. It looks like a lead-volume problem, the kind a founder solves by spending more on ads. It is almost never that. A rate that low is a scoring or definition problem, marketing and sales disagreeing about what qualified means, or a lead score that is wrong, or follow-up that is too slow to catch leads while they are warm. None of those is fixed by more leads, and all of them are fixed by attention to the joint itself. And the payoff is concrete: closing a five-point gap there lifts revenue by roughly 18 percent, from a fix that costs you a definition workshop and a scoring tweak rather than a bigger ad budget.
Work one constraint, prove it, then move
The operating cadence is to pick the one constraint metric, work it for two to four months, and prove it actually moved on the real surface before you declare victory. One focus metric tied to the current bottleneck, not a scattered assault on the whole funnel. This is the operational expression of measure the constraint, not the wall, and it keeps a small team from spreading itself across problems it cannot all solve at once.
The subtle part is what happens after you fix it. The constraint migrates. Repair the weakest link and some other link is now the weakest, because the chain has a new lowest rate. So you re-run the exercise, list the rates again, circle the new lowest, and work that one next. The funnel's constraint is a moving target, and the method is not a one-time audit but a loop you run every quarter, always pointing your scarce attention at the single joint that is throttling everything else right now.