Formulate Success Criteria

Formulate Success Criteria

Turn customer meeting notes and/or call transcripts into a filled copy of the success criteria workbook. The output is a .xlsx file based on the template in references/success-criteria-template.xlsx with one row per success criterion, populating four columns: Success Criteria, Current Challenge, Future State, and Product(s) Tested. The Validated Y / N and In-app Recording columns are left blank for the SE to fill in later.

The #1 rule: ground challenges, formulate outcomes

• Current Challenge must come from the notes/transcript. Do not invent a pain the customer never expressed. If it is not in the input, it is not a challenge — ask about it (see Step 3) or leave it out.
• Success Criteria and Future State may be formulated. Once a challenge is grounded in the input, you may write a polished, outcome-oriented success criterion and a Datadog-oriented future state, even if the customer did not phrase it that way. This is the value of the skill.
• Never invent products the customer's context does not support. Only map to Datadog products that plausibly address a stated challenge.
• Follow the column style guide below, not fabricated content. Mirror the voice of each column; never invent details the customer's input does not support.

Inputs

The user provides one or more of:

• Meeting notes (pasted text, a file, bullet points from a discovery call)
• Call transcript (raw or cleaned)
• Both

There may also be an account/customer name. If none can be identified, use unnamed and proceed (do not block on it).

Workflow

Step 1: Ingest the input

Acknowledge what you received (notes, transcript, or both) and read all of it. Extract the customer/account name for the output filename if present.

Step 2: Extract current challenges (grounded)

Read through the input and pull out the customer's actual pains, gaps, and frustrations. Typical signals: tool sprawl / consolidation, reactive detection ("customers report issues first"), slow/manual triage, alert fatigue, no centralized logging, no end-to-end visibility, high MTTD/MTTR, cost/FinOps pressure, security/compliance gaps, SLO/reporting needs.

Group related pains into distinct challenges — each challenge becomes one row.

Step 3: Clarify when unclear (ask, do not guess)

Before formulating, decide whether the input is clear enough. Ask the user targeted clarifying questions (use the AskQuestion tool when available) if any of these are true:

• The challenges are vague or high-level with no concrete detail.
• It is unclear which Datadog products/solution areas are in scope.
• A stated pain could map to several very different success criteria.
• The input is too thin to formulate meaningful, testable criteria.

Keep it to a few focused questions at a time. Wait for answers before continuing. If the input is already clear and specific, skip straight to Step 4.

Step 4: Formulate each row

For each grounded challenge, formulate:

1. Success Criteria — outcome-oriented and testable, from the customer's desired-outcome perspective. Use a short headline (e.g. Decrease MTTD/MTTR, Tool Consolidation) or a numbered/bulleted list of concrete capabilities to validate, matching the depth of the input.
2. Current Challenge — the grounded pain from Step 2, in the customer's context. Preserve their specifics (tool names, environments, numbers).
3. Future State — how Datadog resolves it. Prefix with [DD] and name the specific Datadog products/capabilities that deliver the outcome. Keep it concise and benefit-led (single pane of glass, reduced MTTR, lower TCO, proactive detection).
4. Product(s) Tested — the Datadog products/pillars in scope for that row (e.g. Infrastructure Monitoring, APM, Log Management, RUM, DBM, NPM, CCM, CSM, Monitors/SLO/Event Management).

Use the column style guide below for phrasing patterns and apply the voice only where the customer's input supports it.

Step 5: Preview for approval

Present the proposed rows as a Markdown table (Success Criteria | Current Challenge | Future State | Product(s) Tested) so the user can review and edit. Ask whether to adjust anything before generating the workbook. Incorporate edits and re-preview if requested.

Step 6: Generate the filled workbook

On approval, write the rows to a JSON file and run the fill script:

python3 scripts/fill_success_criteria.py \
  --rows /tmp/success-criteria-rows.json \
  --output success-criteria/<customer>/success-criteria-<customer>-<ISO-date>.xlsx

Where the rows JSON is a list of objects:

[
  {
    "success_criteria": "Decrease MTTD/MTTR",
    "current_challenge": "Issues are reported by customers first; triage is slow and manual.",
    "future_state": "[DD] Datadog correlates APM traces, logs, infra metrics, and RUM in one place to detect issues proactively and cut MTTD/MTTR.",
    "products": "Infrastructure Monitoring, APM, Log Management, DBM, NPM, Incident Management"
  }
]

The script is stdlib-only (no installs), copies the branded template, appends one row per entry with the template's styling preserved, and leaves Validated Y/N and In-app Recording blank. Report the output path and row count to the user.

Column style guide (from the validated examples)

Files

• references/success-criteria-template.xlsx — blank branded template (base for output)
• scripts/fill_success_criteria.py — fills the template from a rows JSON file