Discovery Template & Question Bank

How to use this

Discovery is structured listening. Each conversation has a target persona, a set of questions calibrated for them, and a defined output you'll add to the synthesis doc. Don't ask all the questions in one meeting — pick the 4–5 that matter most for that conversation, and dig.

Personas in order of typical sequencing:

1. Executive sponsor — the person who signed the contract. Sets vision and definition of success.
2. Customer's data lead — your primary peer. You'll work together for months.
3. Procurement / legal analysts — daily users; their friction is your highest-leverage signal.
4. IT & security — gatekeepers; file tickets early.
5. Output consumers — sometimes a different group from analysts; often higher-leverage stakeholders.

Executive sponsor (30–45 min)

Usually a VP or C-level — CPO, CFO, GC, or a divisional EVP. Don't waste their time on logistics; aim at vision and decision context.

Questions

1. What business outcome are you trying to drive with this deployment?
2. If we delivered exactly the right thing in 90 days, what would change in your operation?
3. What's the metric you'd point a board member at to show this worked?
4. What's the thing you're worried about that I should know early?
5. Who are the three people whose buy-in is most important for this to succeed?
6. Are there freezes or executive priorities I should plan around in the next 90 days?
7. If something has to give — scope, timeline, or quality — which one?

What to capture

• One sentence on the business outcome they want.
• The metric they'd report on.
• The political map (who matters, who's a watch-out).
• Their stated tradeoff preference (the last question is gold; ask it).

Customer's data lead (60–90 min)

Your primary peer. Usually a director or staff-level engineer in their data org. Goal: understand their stack, their constraints, and what they think will be hard.

Questions

1. Walk me through your data stack today — ingestion, warehouse, transformation, BI.
2. Where does the contract / invoice / PO data live today? What format?
3. What's the cleanest path for us to pull data in? What's actually realistic given your team's bandwidth?
4. What's your change-management process for deploying into production?
5. Who else on your team will I work with day-to-day?
6. What's the most painful integration you've done recently, and what made it painful?
7. What's a system in your stack where you'd want to push back if I asked to touch it?
8. Do you have a sandbox / dev environment I can use?
9. What tooling does your team use for orchestration and observability?
10. If I needed to add a new column to a customer-facing dashboard, what's the path?

What to capture

• Stack diagram with system names + which team owns each.
• Their preferred ingestion mechanism (drop bucket, API pull, etc.).
• Change-management process and freeze windows.
• Names of every collaborator you'll touch.
• Their pain points — listen for "the most painful integration" answer; it tells you where to be careful.

Procurement / legal analysts (45–60 min, ideally with 2–3 of them)

The daily users — paralegals, contract analysts, procurement analysts. Their workflow is what the platform will change. Their endorsement is the leading indicator of expansion at renewal.

Questions

1. Walk me through how you'd answer this question today: "what's expiring in the next 90 days that we need to act on?" Every system you touch, every spreadsheet you open.
2. What contracts give you the most trouble? Why?
3. What do you wish you could ask the data but can't today?
4. What part of your workflow is repetitive and you'd hand off to a robot?
5. If you had to pick one number you wish was always accurate and up-to-date, what is it?
6. Where do you keep your own spreadsheets / notes that aren't in any official system?
7. What's a recent contract issue that took longer to resolve than you wanted?

What to capture

• The actual workflow they use today (specific systems, specific friction points).
• The "shadow IT" — spreadsheets and notes that won't show up on the data-stack diagram.
• The repetitive workflow you can replace with the highest leverage.
• The one number they care about most.

IT & security (30–45 min)

Usually one combined call with the customer's IT director + security lead. Lower-frequency contact than the data team, but high-leverage on access.

Questions

1. What's the auth pattern you require for the platform's UI access — SAML SSO, what IdP?
2. Do you require SCIM provisioning for user lifecycle?
3. For service-to-service traffic (our pipeline → your warehouse), what auth do you require?
4. Do you require PrivateLink / VPC peering, or is public-internet OK with TLS?
5. What's your data-residency requirement? Where can our data live?
6. What's the security review process before we can ship to production?
7. Do you have an existing security questionnaire for us to fill out?
8. Who at your end approves new integrations?
9. What's the timeline you typically see for these approvals?

What to capture

• The auth requirements (concrete: which IdP, which protocols).
• The security review process and timeline — the long pole of every deployment.
• Who approves; how to escalate when they're slow.
• Any data-residency or regulatory constraints.

Output consumers (30 min each)

Often different from the analysts. Sometimes it's a VP who saw a demo and signed the contract but doesn't use the system directly; sometimes it's a CFO's chief of staff producing board materials. They define what success looks like in artifact form.

Questions

1. How would you use this output in your day-to-day or monthly work?
2. What format do you need the output in — dashboard, weekly email, embedded in another tool?
3. What's a decision you've recently made where this data would have changed your call?
4. If we showed you a dashboard at the end of month one, what charts would you most want to see?
5. Who shares this output with you today, and how?

What to capture

• The artifact format they need (dashboard, report, etc.).
• Specific charts / numbers they'd want to see.
• Existing reporting they expect to replace or augment.

Reading the corpus (10–15 hours over the first two weeks)

Read at least 20 representative documents from the customer's corpus. Not for completeness — for pattern. What you're looking for:

• Layout variance. Are all the MSAs visually similar, or does each year / template version look different? Are they templated or fully bespoke?
• Completeness. Are amendments present? Are SOWs separately filed? Do you see signature pages, or do they trail off?
• Age range. Pre-2010 contracts may use different vocabulary and weaker scans. The full corpus often spans 15+ years.
• Language and jurisdiction. Multi-region customers have English / French / German / Mandarin contracts. Plan for translation if needed.
• Quality of scans. Native PDF, image-based, mixed, handwritten amendments?
• Edge cases. What weird things show up — embedded contracts inside emails? Multi-tenant contracts where one signed doc covers multiple buying entities? Bundled SOWs?

Take notes per document class. By end of week two, you should have a clear sense of where extraction will work well and where it will struggle.

Synthesis template

Use this Markdown template to consolidate discovery into a draft before writing the working agreement. Share with the customer's data lead for accuracy check before going to the sponsor.

# [Customer] — Discovery Synthesis
Last updated: YYYY-MM-DD

## Business outcome
- Stated outcome:
- Metric the sponsor would point to:
- Confidence we can deliver this in 90 days:

## Stakeholders
- Executive sponsor: [name, role]
- Data lead (primary partner): [name, role]
- Procurement / legal analysts: [names]
- IT / security gatekeeper: [name]
- Output consumers: [names + what they want]

## Stack
- Source documents: [where they live, what format]
- CLM: [if any]
- ERP / system of record: [vendor + version]
- Procurement suite: [if any]
- Warehouse: [vendor]
- BI: [vendor]
- Identity: [vendor]
- Orchestration / observability: [vendors]

## Corpus profile (from reading 20+ docs)
- Document classes in scope, ranked by importance:
  1.
  2.
  3.
- Layout variance per class: [low / medium / high, notes]
- Completeness: [amendments present? signatures? SOWs linked?]
- Age range:
- Languages / jurisdictions:
- Scan quality: [native / image / mixed; specific concerns]
- Edge cases observed:

## Workflow being replaced
- Current manual process:
- Hours per week today:
- Pain points called out:

## Constraints
- Change-management process:
- Freeze windows:
- Security review timeline:
- Data residency requirement:
- Other regulatory:

## Risks (ranked)
1. [Risk]: [mitigation]
2.
3.

## Proposed first slice
- Document class:
- Source:
- Output target:
- Success criterion:

## Open questions
-
-
-

This synthesis doc becomes the input for the working agreement (next chapter).