Sales Conversion Engine . Agency OS

What it does

Produces the working system that converts inbound leads into signed customers: a pipeline with stages and entry and exit criteria, speed-to-lead instant reply automation, a booking and show-up flow, qualification logic, discovery and closing call scripts with your offer presented live, objection responses for your market, and follow-up plus dead-lead resurrection cadences. Output is GoHighLevel-ready, so stages become GHL stages, follow-up becomes GHL workflows, and qualification becomes a Conversation AI prompt or a form. It is Build 7, sitting between the funnel that generates leads and the delivery system that fulfils them.

Say this to start

This skill has no button. You start it by saying what you want. Any of these will do it:

/build-sales

> build sales

> sales system

> close leads

> objection handling

> speed to lead

When to reach for it

leads are coming in and nobody works them fast enough
no pipeline exists, or every deal sits in one stage
your close rate drops after the discovery call
the same three objections kill deals every week
no-shows and ghosted prospects are never followed up
building a client's closing system inside their workspace

When NOT to use it

If you actually want	Use this instead
Your agency's own cold outbound client acquisition	`agency-sales-system`
The pages, lead capture and nurture sequences that generate the leads before a call exists	`funnel-architect`
Defining or repricing the offer the script presents	`offer-architect`
Writing the GHL Conversation AI bot prompt that handles speed-to-lead replies	`conversation-ai-architect`
Onboarding, delivering to, or keeping a client after they have bought	`client-delivery-engine`

Before you start

What you need	Why
An offer document (`offer-output.md`) for this product	the skill checks for it first and offers to run `/build-offer`, because every script and price conversation anchors to the offer, its guarantee and its price framing	Required
The right workspace active in `brand/_active.md`	the workspace prefix decides the mode: a `client--` workspace builds the client's closing system in the client's name, `_default` builds your agency's inbound closing system	Required
A `verified-claims.md` registry with entries marked `consent: ok-as-stated`	the skill calls this load-bearing here, because an invented 'a client told me' anecdote is most tempting inside closing scripts and objection handling; anything unbacked becomes a `[PROOF GAP]` marker	Required
A name for this sales track	the skill asks for one and files everything under it, since one product can run several tracks such as high-ticket call close, self-serve checkout or retainer renewal	Required
A completed funnel (`funnel-output.md`)	it is read so the pipeline begins exactly where the funnel's nurture ends; without it you risk a gap or double-messaging between nurture and pipeline	Optional
GoHighLevel MCP connected	Build Mode A configures stages, workflows and Conversation AI for GHL and offers to deploy them; without it you get markdown plus GHL setup instructions	Optional
An ideal customer profile (`ica-output.md`)	listed as a context dependency and a schema input; qualification criteria and objection handling are sharper with it	Optional
Web research tools available (Firecrawl, Perplexity, or WebSearch and WebFetch)	market research runs before Phase 1 and is blocking; it pulls competitor closing scripts, close-rate and show-rate benchmarks and speed-to-lead norms	Optional

How it runs

Resolve the workspace and the modeIt reads brand/_active.md and decides whether it is building a client's closing system or your agency's inbound one. If you actually want cold outreach and proposals, it redirects you to /sell-agency and asks whether to continue.
Name the sales track and check prerequisitesIt asks you to name this sales track, then confirms an offer exists and reads the funnel output if present so the pipeline picks up where nurture ends.
Market researchA blocking research pass on competitor discovery and closing scripts, proposal and pricing presentation, objection patterns, follow-up cadence, close-rate and show-rate benchmarks, and the current state of SMS versus email versus voice for follow-up.
Choose a build modeGHL Deployment (the default when GHL is connected) configures stages, workflows and Conversation AI for GHL. Copy Documents gives you markdown to implement anywhere, with GHL setup instructions.
Phases 1 and 2: pipeline and speed-to-leadIt maps stages from lead to closed or lost with one clear entry and exit criterion each, then builds the instant-reply automation, the booking flow, and the show-up sequence of reminders and confirmations.
Phases 3 and 4: qualification and the closeQualification criteria, discovery questions, and either a Conversation AI prompt or a qualification form. Then the discovery and closing call scripts, the offer-presentation script ported from your offer document, and the price-presentation framing.
Phases 5 and 6: objections and follow-upThe top objections for your offer and market with responses, built on the shared closing primitives so they do not drift from /sell-agency. Then the multi-touch follow-up cadence, no-show recovery and the dead-lead resurrection sequence with its cadence maths.
Phase 7 and closeoutGHL workflow configs, a pipeline checklist, KPI targets on the four conversion rates and a launch checklist. It then saves sales-system.md and .docx, writes the sales-output schema, updates the dashboard, offers to deploy to GHL, asks for slides, and asks how the build landed.

What you get

[target]/sales/[track]/sales-system.md plus a .docx, the canonical artifact
A pipeline stage map from lead to closed or lost, with entry and exit criteria per stage and GHL pipeline configuration
Speed-to-lead instant-reply automation and a booking flow with confirmations and reminders
Qualification criteria, discovery questions, and a Conversation AI prompt or a qualification form
Discovery and closing call scripts, an offer-presentation script anchored to offer-output.md, and price-presentation framing
Objection responses for your specific offer and market
A multi-touch follow-up cadence, no-show recovery, and a dead-lead resurrection sequence with cadence maths
GHL workflow configurations, a pipeline checklist and a launch checklist
KPI targets on the four conversion rates: speed-to-lead reply, lead-to-appointment, appointment-to-show, show-to-close
A sales-output schema artifact, an appended assets.md entry, and a Sales section in agency-dashboard.html

Honest limits

Read this before you rely on it

Direct deployment into GoHighLevel is only offered when GHL is connected. Otherwise you get markdown plus setup instructions and you build the pipeline, workflows and bot yourself.
It builds the system, it does not run it. The scripts need a person (or a configured bot) to actually work the leads, and your close rate depends on that person more than on the script.
It deliberately will not rebuild funnel nurture and will not touch onboarding. If those are missing you need Build 4 and Build 8 separately, so this alone does not cover the full journey.
Every anecdote, number or client result inside a closing script must trace to verified-claims.md. Without a proof registry your scripts come back with [PROOF GAP] markers where the persuasion would otherwise sit.
Market research runs before Phase 1 and is blocking, so the build cannot start until live web tools have been called.

Where people go wrong

The mistake	Do this instead
Writing broadcast sequences that talk to the whole list on a schedule	That is Build 4 nurture. This skill works one individual lead through the stages, so stop and let the funnel own the list.
One pipeline stage doing two jobs, such as 'Contacted and Qualified'	Split them. Each stage needs one clear entry criterion and one clear exit criterion.
Treating speed-to-lead as an optional extra	Build Phase 2. A lead worked in 5 minutes converts far better than one worked in an hour.
Generic 'book a call' scripts with no offer in them	Every closing script must present the actual offer from `offer-output.md`, with its guarantee and price framing.
Marking a lead dead after two attempts and moving on	Build the resurrection sequence. Most leads do not close on the first attempt and a large share of revenue is in the ones already written off.
Running this in `_default` expecting cold outreach and proposals	That is `/sell-agency`. In your own workspace this build closes the inbound leads your funnel already produced.

Worth knowing

Run this after the funnel exists. The skill reads funnel-output.md and starts the pipeline exactly where the funnel's nurture ends, which is what stops leads getting two conversations at once or falling into a gap between the two systems.