Sales Plays

A sales play is a document that tells your agents how to handle one type of buyer: who they are, what they need, what to recommend and how to answer their objections. This page shows how to write plays and make them available to your sales agents.

How plays reach your agents

Agents read plays from a knowledge group. Knowledge groups are built from data synced from your connected sources, so keep your plays as files in a connected source, such as a GitHub repository.

  1. Write each play as a Markdown file, for example in a plays/ folder of a repository you have connected to Outrun.
  2. Go to Automation > AI Agents, open the Knowledge tab and select Create Group.
  3. In Sources, choose the repository. In Include Folders (optional), enter plays/. Select .md in File Types.
  4. Name the group, for example Sales plays, and select Create Knowledge Base.
  5. Open each sales agent and select the group in its Knowledge card.

When the agent responds, it searches the group for the parts of your plays that match the conversation.

When you change a play, wait for the source to sync, then open the knowledge group and select Rebuild.

What to put in a play

Write one play per buyer type. A useful play answers five questions:

  1. Who is this buyer? Describe the persona and the signals that identify them in a conversation.
  2. What do they care about? List their specific problems.
  3. What should the agent recommend? Name the plan, product or next step, and why.
  4. What if they push back? Give the response you want for each common objection.
  5. What should the agent avoid? List topics, offers and promises that are off limits.

Keep each play focused on one buyer type. Agents match sections of documents to the conversation, so a play that mixes several buyer types is harder to match.

Play template

Copy this template into a new file for each buyer type.

# Play: <buyer type>

## Who this is for
<Role, company size, industry, budget, and the signals that identify
this buyer in a conversation>

## Their problems
- <Problem 1, described as the buyer would describe it>
- <Problem 2>

## How we help
- <Problem 1> -> <The capability that solves it and why>
- <Problem 2> -> <The capability that solves it and why>

## What to recommend
- Recommend: <plan or product>
- Why: <reasoning the agent can share>
- Suggest an upgrade when: <signal>

## Objections
- "<Objection>"
  -> <The answer you want the agent to give>

## Next steps
- Primary: <demo, trial or proposal>
- If not ready: <resource or follow-up>

## Avoid
- <What not to say or offer to this buyer type>

Example play

This example uses a fictional company, Acme Pty Ltd, that sells project management software.

# Play: Small construction teams

## Who this is for
Site managers and owners at construction firms with 1 to 10 staff.
They mention spreadsheets, paper job sheets or missed handovers.

## Their problems
- Job updates are spread across texts, email and paper
- Handover details get lost between shifts

## How we help
- Scattered updates -> One job record the whole team can update from site
- Lost handovers -> A shift handover checklist on every job

## What to recommend
- Recommend: Acme Team plan
- Why: It covers up to 10 users, which fits their crew size
- Suggest an upgrade when: they mention more than 10 staff or several sites

## Objections
- "We don't have time to learn new software"
  -> Offer the guided setup call and point to the mobile app for site use.

## Next steps
- Primary: Book a guided setup call
- If not ready: Send the getting-started guide

## Avoid
- Do not offer discounts
- Do not mention enterprise features

Improve plays over time

  • Turn on human-in-the-loop review for your sales agents. Edits you make while reviewing become examples for that agent.
  • If you keep making the same edit, update the play instead.
  • Keep pricing in one maintained document, so agents do not quote outdated prices.