Sales workflow that actually runs

A sales workflow is the repeatable sequence a deal moves through from first contact to close. The test of a good one is simple: can anyone look at any deal and know what happens next? Most teams have stages. Far fewer have a workflow, because stages describe where a deal is, while a workflow defines what moves it forward. The difference shows up on busy weeks, when deals in stage-only pipelines sit still and deals in real workflows keep moving.

A sales workflow is the repeatable sequence a deal moves through from first contact to close, and Ember is the AI sales assistant that drafts the next email in the seller's voice from the real inbox when a genuine reason exists. Approval is the default; auto-send is optional. Stages describe where a deal sits, while a workflow defines the trigger, the action, and the maximum delay between them — and most elapsed time in a pipeline lives in those gaps, from a reply waiting hours to a quiet deal waiting ten days. Ember reads Gmail, Outlook, Attio, HubSpot, and Zoom history for each contact, notices the trigger, and has the next draft waiting in the approval queue. Mechanical steps — who is due, what was said, the first draft — sit with software, while qualification, negotiation, and the send decision stay with the seller. Ember does not replace the CRM; Ember sits on top of it so the record stays where the team already works.

What are the six stages of a sales workflow?

Workflows vary by market, but relationship-driven sales reduces to six recognizable stages: identify the relationship worth pursuing, make genuine contact, understand the problem, propose a fit, resolve the open questions, and close. You can rename or subdivide them freely. What you cannot skip is the thing most pipelines lack: an exit criterion for each stage, defined by what the buyerdid. "Sent the proposal" is seller activity; "buyer confirmed the proposal matches what their team needs and named a decision date" is an exit criterion. Buyer-defined criteria keep the pipeline honest and make the next action self-evident: it is whatever closes the gap to the criterion.

How do you define the action inside each workflow stage?

For each stage, write down three things: the trigger that starts work (a reply arrives, a meeting ends, a deal goes quiet for too long), the action that follows, and the maximum acceptable delay between the two. That last number is where workflows live or die. A workflow that responds to a buyer reply within hours produces a visibly different pipeline than the same workflow at three days' latency. Same stages, same actions, very different sales cycle. When you map your current workflow honestly, most of the elapsed time sits in these trigger-to-action gaps, which is exactly where optimization effort pays off first.

What does walking a deal through a sales workflow look like?

Here is the shape in practice. A prospect replies to an intro with interest, so the workflow says contact becomes discovery, and a call gets booked inside that same thread, same day. The call ends, and the workflow says discovery produces a recap with next steps, so it goes out before the conversation cools.

The recap names a question for their ops lead, so the workflow says new stakeholder, genuine personal outreach, and that email goes out grounded in the context of the deal so far. Ten days pass with silence, and the workflow says quiet deals get a follow-up with a real reason, so one arrives, carrying something new rather than a bump. Nothing in that sequence is clever. Its entire power is that it happens every time.

Which workflow steps should you automate, and which should stay human?

The reason workflows fail is rarely design. It is that every step depends on a busy human remembering to execute it. The durable fix is moving the mechanical steps to software. The remembering (which deal needs attention now), the logging (what happened, recorded without typing), and the first draft (the recap, the follow-up, the new-stakeholder intro) are all work that Emberdoes natively: it reads each relationship's full history, notices the trigger, and has the next email drafted in your voice when you open the queue. You approve every send. The workflow gains the reliability of automation while every message stays grounded in the real relationship, the standard described in personalization at scale.

How do you instrument a sales workflow, then trim it?

Once the workflow runs, measure it the way you would any system: time-in-stage, conversion between stages, and trigger-to-action latency, the core pipeline metrics. The numbers will tell you which stage to fix next, and they will also tell you what to delete. Every workflow accumulates steps that once mattered: the report nobody reads, the field nobody uses. A workflow is not done when nothing can be added; it is done when nothing can be removed. Review it quarterly with the team that runs it, keep what moves deals, and cut the rest. That discipline is itself one of the most underrated sales best practices.