Discovery Brief
Captures what is actually known about one opportunity before anyone writes a proposal. The brief is the input every later sales skill reads.
Most proposal time is lost to hunting for context, not to writing. A brief that records the buyer's own words, the decision process, and — critically — what is still unknown, is what makes a later proposal fast and specific.
Read First
workspace/firm/profile.md— what the firm sells.workspace/firm/services/{slug}.md— the candidate service, its fit and anti-fit criteria.workspace/marketing/icp/{slug}.mdand its personas — how this buyer type usually buys.- Existing folders in
workspace/sales/opportunities/— do not open a duplicate.
When to Use
- A prospect replied and there is something real to record.
- A fit or discovery call happened and notes exist.
- An RFP or inbound brief arrived.
- Someone wants a proposal and no brief exists yet.
Core Rules
Brand scope. In a multi-brand workspace (2+ brand homes in workspace/firm/brands/), every path this skill reads or writes is brand-scoped: definitional entities (services, ICP, personas, proof, channels) live in firm/brands/{brand}/…, working pipelines carry a {brand}/ segment under their entity root, and the working brand comes from the user's choice or B2BFORCE_BRAND — never guessed. Rules: Brand Scope Gate in AGENTS.md; paths: docs/WORKSPACE.md.
- One opportunity = one folder. Do not create a folder per conversation.
- Record only what was said. Everything else goes in
## Open questions. A brief that reads as complete when it is not is how a firm ends up proposing into a decision process it never understood. - Never invent budget, timeline, decision makers, or competitors. Absent is
unknown, not a guess. - Distinguish the buyer's words from your interpretation. Quote the buyer for the
problem statement; put your reading under
## Firm assessment. - Do not write a proposal in this skill and do not decide bid/no-bid here. Those
are
sales-proposalandsales-bid-qualification.
Output
Folder: workspace/sales/opportunities/{account}--{service}--{YYYY-MM}/
File: !_discovery.md
The !_ prefix marks it as needing attention until qualification runs, per the
repo's naming convention.
Frontmatter:
---
account: acme-industrial
account_public: false
service: platform-migration
icp: mid-market-manufacturing
persona: cto-mid-market
source: inbound | referral | outbound | rfp | existing_client
opened: 2026-07-25
updated: 2026-07-25
budget_signal: unknown # stated | range_hinted | unknown | no_budget
timeline_signal: unknown # dated | quarter_hinted | unknown | no_timeline
decision_process: unknown # mapped | partial | unknown
competitors_known: []
bid_decision: pending # set by sales-bid-qualification
brief_quality: thin # thin | workable | strong
---
Body sections:
## Situation
## Problem in the buyer's words
## Impact
## Desired outcome
## Decision process
## Constraints
## Firm assessment
## Open questions
## Next action
Field-by-field rules and a worked example: references/brief-fields.md.
Brief Quality
Set brief_quality honestly, because the Proposal Gate reads it.
| Value | Means |
|---|---|
thin |
Problem is vague, or no impact, decision process, and timeline |
workable |
Problem and desired outcome are concrete; one or two gaps remain |
strong |
Problem, impact, outcome, decision process, and constraints all recorded |
A thin brief is a normal first state after a short inbound email. Say what is
missing and what question would close the gap.
Refuse To Proceed
If the user asks for a proposal from a thin brief, do not write one. Return the
three questions that would move the brief to workable and offer to draft the
follow-up message that asks them.
This refusal is the point of the skill. A proposal built on a thin brief is where the average firm spends eight hours and loses.
Workflow
1. Identify the opportunity
Slug the account in kebab-case. Pick the service the request most likely maps to;
if two services could fit, record both under ## Firm assessment and let
qualification decide.
Check for an existing folder for the same account and service. If found, update it instead of creating a second one.
2. Extract from the source
Work through the source material once, and write down only what is present:
- the problem, quoted where possible;
- what it costs them today — money, time, risk, or blocked work;
- what they say they want instead;
- who decides, who influences, and what has to happen for a yes;
- constraints: timing, budget, incumbent vendor, internal capacity, compliance.
For an RFP, also record the submission deadline, required format, and evaluation criteria if stated.
3. Mark the gaps
Everything not stated goes under ## Open questions as a question you would
actually ask, not a field name. "Who else has to approve this besides you?" beats
"decision process: unknown".
4. Assess fit
Compare against the service's fit and anti-fit criteria and the ICP. Write your
reading under ## Firm assessment — including reasons this might not be a good
fit. Do not soften anti-fit signals; they are what makes the next step cheap.
5. Set the next action
One concrete next step with an owner. Then hand off to
sales-bid-qualification.
Testing Requirements
- Build a brief from a three-line inbound email; confirm
brief_quality: thinand that## Open questionsis the longest section. - Ask for a proposal from that brief; confirm the skill refuses and returns questions instead.
- Build a brief from full call notes; confirm
workableorstrong. - Re-run for the same account and service; confirm it updates rather than creating a second folder.
Reference Files
| File | Content |
|---|---|
references/brief-fields.md |
Section rules, signal vocabularies, worked example |
Related Skills
| Skill | When |
|---|---|
sales-bid-qualification |
Next — decides bid or no-bid on this brief |
sales-proposal |
After a bid or conditional decision |
marketing-icp |
The ICP or persona for this buyer does not exist yet |
firm-context |
Firm profile or service definition missing |