Orlob Discovery Framework

1. Business Problem — what's the actual problem behind the ask? (not the feature)
2. Cause Analysis — what's driving that problem? (root cause, see reference/epiphan-root-causes.md)
3. Negative Impact — what does it cost them? (quantified: time, money, students, cases, congregants)
4. Future State — what does "fixed" look like, in their words?
5. Close — lock a specific next step with a date and an owner.

Tim's proven opener for Step 1/2 (works across verticals):

"Where do you tend to lose the most time — capture, switching sources, or getting recordings out the door?"

Never feature-dump. Ask, listen, confirm the pain back, then connect a product.

Scores 3/3 on the discovery dimension of sdr-call-coaching when cause analysis surfaced a root cause the buyer owns (see Scoring Rubric below).

The 5-Step Framework

Discovery is a funnel, not a checklist. Each step earns the right to the next. You cannot quantify impact (Step 3) on a problem you haven't diagnosed (Step 2), and you can't diagnose a problem the buyer hasn't admitted is a problem (Step 1). Move down only when the buyer has agreed with you at the current step.

Step 1  Business Problem   →  "So the real issue is X, not just needing new gear?"
Step 2  Cause Analysis     →  "And what's driving X — is it A, B, or C?"
Step 3  Negative Impact    →  "When X happens, what does it cost you?"
Step 4  Future State       →  "If X were solved, what changes for you?"
Step 5  Close              →  "Makes sense to get [next step] on the calendar?"

Step 1: Surface the Business Problem

Get past the feature request ("I need a lecture-capture box") to the business problem behind it ("faculty stopped recording and students are complaining"). Buyers open with solutions; your job is to walk them back to the problem.

Epiphan example (Higher Ed): Buyer says: "We're looking at replacing our capture appliances." You: "Before we talk boxes — what made replacing them a priority now? What's not working with what you have?" → surfaces "usage dropped below half and the provost is asking why."

What good looks like: the buyer describes a business outcome at risk (adoption, uptime, compliance, the record, the service) — not a spec.

Calibrated questions (ask at least one open one before any product talk):

• "What made this a priority now — what changed?"
• "Where do you tend to lose the most time — capture, switching sources, or getting recordings out the door?" (Tim's proven line)
• "If nothing changed here, what happens six months from now?"
• "How do you currently handle [the workflow they mentioned]?" (Voss calibrated)

Do NOT: pitch a product, name a SKU, or list specs. You're diagnosing.

Step 2: Cause Analysis

The most-skipped, highest-value step. Take the stated problem and find the root cause — because the root cause is what Epiphan actually fixes, and naming it is what makes the buyer trust you diagnosed correctly. Use reference/epiphan-root-causes.md to match the buyer's symptom to a known cause for their vertical, then confirm it back.

Epiphan example (Higher Ed): Problem: "faculty stopped recording." Cause-analysis: "What's making it hard for them — is it the setup per room, the different source types, or that nobody's around when something fails?" Buyer: "Honestly, one guy supports 50 rooms." → root cause = AV-staffing crunch (from the reference file).

Calibrated questions:

• "What's driving that — is it A, B, or C?" (offer the vertical's known causes)
• "What happens when [the failure] occurs — who gets the call?" (Voss calibrated)
• "Walk me through the last time this broke — where exactly did it fall apart?"
• "Is that a people problem, a gear problem, or a workflow problem?"

Do NOT: accept the symptom as the cause. "Recordings are inconsistent" is a symptom; "one tech covers 50 rooms so nobody's there on a failure" is a cause. See reference/epiphan-root-causes.md for the symptom→cause map per vertical.

Step 3: Negative Impact (Quantified)

Turn the root cause into a number the buyer says out loud. Cost of inaction is what creates urgency and justifies budget. Push gently for quantification — time, money, headcount, students touched, cases delayed, congregants reached.

Epiphan example (Higher Ed):

"If a faculty member stops recording because it's too hard, how many students does that touch?" → buyer: "A big lecture is 300 kids." Now the AV-staffing crunch has a number attached, and the problem is worth solving.

Calibrated questions:

• "If a faculty member stops recording because it's too hard, how many students does that touch?" (Tim's proven impact line)
• "How many hours a week does your team spend firefighting this?"
• "What does a failed [service / proceeding / all-hands] cost you — not just dollars, reputation too?"
• "How often does this happen — weekly, every event, once a semester?"

Do NOT: move to solution while the impact is still vague. If the buyer won't quantify, the problem may not be painful enough to buy against — note that.

Step 4: Future State

Have the buyer describe what "solved" looks like in their words. This gives you their Decision Criteria for free and makes the eventual demo a confirmation of their own vision rather than a feature tour.

Epiphan example (Corporate AV):

"If hybrid all-hands just worked every time — nobody sweating the stream — what would that free your team up to do?" → buyer describes the win, and you've got their success criteria and an emotional stake in it.

Calibrated questions:

• "If this were solved, what changes for you day-to-day?"
• "What would need to be true for you to consider this a win?" (Voss calibrated)
• "A year from now, if this worked, what does your team stop worrying about?"
• "Who else feels the relief when this is fixed?" (surfaces stakeholders → MEDDIC)

Do NOT: describe your future state ("with Pearl you'll be able to…"). Let them paint it; you map the product to their picture at demo time.

Step 5: Close to Next Step

Every discovery call ends with a committed next step — a date, an owner, and what happens. "I'll send some info" is not a close. Tie the next step to the problem and impact you just uncovered so it feels like progress, not a sales push.

Epiphan example:

"Based on the staffing crunch and the 300-student exposure, the fastest way to know if this fits is a 30-minute demo with the actual room sources. Does Thursday or next Tuesday work, and should your AV lead be on it?"

Calibrated questions / closes:

• "What's the best next step from your side to figure out if this is a fit?" (let them propose — Voss)
• "Makes sense to get a technical demo on the calendar — Thursday or Tuesday?"
• "Who else should be in the room when we show this?" (Decision Process + Champion)
• "What would stop this from moving forward?" (surfaces blockers early)

Do NOT: end open-ended. No date + no owner = no next step.

Calibrated Question Bank (by step)

Grouped for quick pull into sdr-dial-lists and epiphan-call-playbook. Voss calibrated forms ("How do you…", "What happens when…", "What would need to be true for…") are noted; they keep the buyer talking and non-defensive.

Per-vertical calibrated questions and root causes: reference/epiphan-root-causes.md.

MEDDIC Integration Points

Capture MEDDIC during discovery — don't run a separate interrogation. Each Orlob step naturally surfaces MEDDIC elements:

For deeper Economic Buyer validation and ATL/BTL classification, hand off to meddic-call-prep-auto-skill. A BTL-only call means you found Pain but not the Economic Buyer — note it and use Step 5 to ask who owns the budget.

Scoring Rubric (for sdr-call-coaching)

The discovery dimension is scored 0-3. This is the authoritative rubric that sdr-call-coaching grades against:

• 3 — Peeled past the surface to a business problem the buyer owns; asked calibrated/open diagnostic questions; surfaced a root cause (Step 2) and confirmed it back; ideally quantified impact (Step 3). Cause analysis happened.
• 2 — Identified a real pain point but stayed at symptom level — never traced it to a root cause. Some good questions, but no diagnosis.
• 1 — Asked some questions, but all surface-level; no business problem surfaced, no cause analysis.
• 0 — Pure feature dump. Pitched product with no discovery.

Coaching tie-in: when scoring below 3, name the specific Orlob move missed — e.g. "cause analysis — never asked what's driving the AV-staffing crunch" or "skipped impact — got the problem but never quantified what it costs them."

Common Mistakes

• Feature dump (score 0): leading with SKUs and specs before diagnosing. The buyer's opening solution-request is bait — walk them back to the problem.
• Staying at symptom level (score 2): accepting "recordings are inconsistent" as the problem and moving to solution. That's a symptom; find the cause.
• Skipping cause analysis (Step 2): jumping from problem straight to impact or demo. Step 2 is where you earn trust and where the right product surfaces.
• Unquantified impact (Step 3): "that sounds frustrating" is not a number. No number = no urgency = a deal that stalls.
• Painting the buyer's future state for them (Step 4): "with Pearl you'll…" Let them describe success; you map to it at demo.
• Open-ended close (Step 5): "I'll follow up" with no date/owner. Every call ends with a committed next step.

Emit Outcome Sidecar

After running discovery guidance on a call, emit an outcome sidecar so the analytics pipeline can track skill health. Write to the path below (outside the repo — required so the PreToolUse observer hook allows the write):

cat > ~/.claude/skill-analytics/last-outcome-orlob-discovery-framework.json <<'JSON'
{
  "ts": "<ISO8601 timestamp>",
  "skill": "orlob-discovery-framework",
  "version": "1.0.0",
  "variant": "control",
  "status": "success",
  "runtime_ms": <int>,
  "metrics": {
    "discovery_score": <0-3>,
    "root_cause_identified": <true|false>,
    "impact_quantified": <true|false>,
    "next_step_committed": <true|false>,
    "vertical": "<higher-ed|courts|government|healthcare|corporate-av|how|k12>"
  },
  "error": null,
  "session_id": "<session id>"
}
JSON

Set status to "partial" if discovery ran but no root cause was surfaced (score ≤ 2), and "error" with an error message if the skill could not run.