You sequence them so each close finishes the previous close's job. The documented example is the Franklin T chain: the T surfaces every remaining concern; each concern gets the framework; the temperature check measures what's left; and the close that fits the remaining situation runs last. A stack is a relay, not a pile — three closes doing one job is volume, not structure.
The short answer
The script itself shows the handoff: after the T's columns are written — reasons for, concerns against — the rep asks the surfacing question (based on everything we've talked about, is there anything else that would keep you from moving forward besides these concerns?), works each listed concern, and checks the temperature (how are we doing so far?) before the decision is asked for. Notice what no step does: repeat another step. The T diagnoses, the framework resolves, the temperature verifies, the close collects. Stacks fail when the rep runs three collection closes in a row — that's not a stack, it's asking three times, and the homeowner can count.
How it runs
Plan the chain from the decision backwards: what must be true before the final close can work (concerns surfaced? temperature known? option chosen?), and which close produces each precondition? That backwards read is the selection skill applied across time. It also caps the stack: when the preconditions are met, stop adding closes. Every unnecessary link spends momentum the final close needed — the stack serves the decision, never the rep's anxiety.
Spar's rounds and the 33-objection library grade your framework under pressure — Round 1 is free; the full camp is $29/month. Spar before you knock.