Lesson 5 of 6 · 48 min

Discovery & value storytelling

The customer-facing half of the job — the SE’s discovery stack (MEDDPICC, SPIN implication questions, Challenger reframe), the fixed five-pain stack of enterprise AI, value/ROI storytelling with credible 200–600% ranges, arming the champion, and the discovery round interviewers run.

You earn the right to demo, or you get shown the door

The discovery conversation is where a solutions engineer either earns the right to demo or gets shown the door. The architecture skills from Lessons 1–4 are necessary but not sufficient: the deals you win are the ones where the champion can re-tell your value story to their CFO when you’re not in the room. This lesson is the discovery-and-demo round of the SE interview — translate a vague business problem into a system spec in real time, and resist the urge to provide an immediate AI solution.

The SE’s discovery stack is five overlapping frameworks, and the senior skill is knowing which to lead with. MEDDPICC (MEDDIC + Paper Process + an expanded Competition field) is the qualification baseline that survives banking, healthcare, and government procurement — the three highest-leverage letters for an SE are I (Identify/Implicate Pain), C (Champion), and D (Decision Criteria, i.e. decode the written RFP so the demo speaks to every scorecard line). SPIN sequences Situation → Problem → Implication → Need-Payoff, and the Implication question is the SE’s specific contribution: it turns a complaint into a quantified downstream cost in the customer’s own voice. Challenger teaches the Reframe — change how the buyer thinks about the problem before showing the product. Command of the Message makes you “audible-ready” with Before Scenario / Required Capabilities / Positive Business Outcomes. ValueSelling talks to the economic buyer in financial terms.

code

1THE SE DISCOVERY STACK (which to lead with)23  Framework            SE's specific use                       Failure mode4  ------------------   -------------------------------------   ------------------5  MEDDPICC             qualify on I (pain), C (champion),      letter the deal but6                       D (decode the RFP scorecard)            can't deliver it7  SPIN                 Implication Q turns complaint -> cost   survey-style interrogation8                       in the CUSTOMER'S words9  Challenger           Reframe the buyer's mental model        contrarian with no insight10                       BEFORE showing product11  Command of Message   Before / Required Capabilities / PBOs   message not in champion's12                       audible-ready, champion-retellable      voice -> dies in committee13  ValueSelling         talk to economic buyer in $            script-following, no fluency1415  Off-stage: MEDDPICC artefacts. On-stage: the SPIN performance.

The Implication question is where you write the customer’s number on the whiteboard in their own hand. If the answer to “if your automation alerts are noisy, what does that cost downstream?” is “we lose 8–12 hours per engineer per week reconciling them,” you write 8–12 hours on the board — not a marketing claim — and the demo becomes an answer to a number the customer produced. The mechanism is anchoring plus loss aversion: by the time you demo, switching cost is psychological, not contractual. Interview angle. In the mock discovery, interviewers listen specifically for whether you ask Implication questions that quantify consequences in the customer’s voice, versus pitching features — the former is the SE’s signature move.

Discovery Masterclass: Everything You Need to Run a Perfect Discovery Call30 Minutes to President’s Club

AI discovery has a fixed five-pain stack

AI discovery differs from SaaS discovery in three ways: the data foundation is rarely ready, the integrator universe is fragmented, and the buyer’s mental model of “AI” is usually wrong. Geodesic’s 2026 research names the five enterprise-AI adoption pains, and a 25–35 minute discovery script built around them catches 80%+ of late-stage objections before committee:

code

1THE FIVE ENTERPRISE-AI PAINS (Geodesic) -> discovery Q -> demo anchor23  Pain                          Discovery question            Demo anchor4  ---------------------------   ---------------------------   ---------------------5  Data isn't ready (even        "What data exists today, in   ingestion + cleansing6  when they think it is)        which system, what quality?"  on a sample set7  Legacy doesn't integrate      "Which core systems does      API/webhook coverage8                                this process touch?"          of the NAMED systems9  Forward-deployed engineers    "Who owns the integration     declarative / low-code10  don't scale                   today, how many FTEs?"        vs hand-coding11  Agent security                "What can an agent access,    identity controls,12                                and who approves it?"         sandboxing, audit logs13  ROI unclear for agentic work  "How would you know it was    metrics dashboard +14                                working, cost of being wrong?" guardrails + rollback

Data readiness is the discovery filter that prevents a dead PoC. Because AI is gated by data, you must assess readiness in call one, not after the PoC. If you don’t, the PoC gets scoped on incomplete inputs and fails in week three — after the champion has spent political capital, which poisons the account. If the answer to “what data exists today?” is “we’ll pull it together for the PoC,” de-risk by scoping a fixed-scope data validation as the first milestone. And lead with use-case selection, not a platform pitch: Xebia scores each use case on feasibility, value, and risk before any tooling talk — the inverse of the platform-first demo 70%+ of AI vendors default to. Open with “Which business decision do you wish you could automate today, and how would you know it was working?”

Value / ROI storytelling: a credible range, not a single number

The ROI conversation is the SE’s most leveraged skill, and the rule is counterintuitive: anchor ROI in credible ranges — 200–600% is realistic, while 50% signals you’re losing money. A 50% ROI to an executive champion loses the room, because 50% is high enough to require project approval yet low enough to feel optimistic — the worst possible combination. A 200–600% range forces the buyer to challenge the assumptions, which is exactly where your value story lives. The strongest reps don’t wait for a value engineer; they run the Business Value Assessment themselves with structure and curiosity.

Attach at least one dollar figure per value lever, sourced from the customer, not from a vendor case study: productivity (hours saved × loaded hourly cost), revenue growth (revenue gained × conversion uplift), costs reduced (FTEs avoided × loaded cost — the economic buyer’s primary motivator), and risk mitigated (expected loss × probability). Each lever is a From → To → Gap worksheet: From = current state in hours/dollars/incidents; To = future state with a specifically defined metric; Gap = the dollar delta. Compose the Gap across levers and present a 3-year ROI with assumptions the champion can defend. You don’t need a finance degree — you need a framework that builds credibility.

code

1THE FOUR VALUE LEVERS (one $ figure each, from the CUSTOMER)23  Lever          Measurement                  Discovery question4  ------------   --------------------------   ----------------------------------5  Productivity   hours saved x hourly cost    "Hours/week on this manual task?"6  Revenue        revenue x conversion uplift  "What's +1pt conversion worth here?"7  Cost reduced   FTEs avoided x loaded cost   "Which roles get redeployed?"8  Risk           expected loss x probability  "Cost of one missed event/quarter?"910  Each lever = From (now) -> To (defined metric) -> Gap ($ delta).11  Anchor the 3-yr ROI as a RANGE: 200-600% credible; 50% loses the room.12  Frame headcount as REDEPLOYMENT, not FTE cuts -- it's political.

Arm the champion — the story they re-tell when you’re not there is the one that wins. Executives trust their own team; if the value story is in your voice, it dies on contact with the steering committee. So hand the champion the value proposition in their language (Command of the Message vernacular), and ensure they can deliver the BVA as if it’s theirs. The corollary: a BVA also serves to disqualify honestly — when 3-year ROI on realistic assumptions lands below ~100% with no strategic-risk component, walk; it protects pipeline integrity and spares the champion a fruitless effort. Set the “walk if” threshold with the AE before you start. Interview angle. Panelists score whether you communicate need over features by attaching each capability to a dollar/hour figure — e.g. “you have 70% retention; a 1-point lift is $2M” — and whether the champion could repeat it.

Which framework to lead with — and the tensions between them

The five frameworks overlap by design, and the senior skill is resolving the three productive tensions between them rather than treating any one as a script. (1) MEDDPICC wants completeness; SPIN wants progression. MEDDPICC says letter the deal; SPIN says don’t jump to Need-Payoff until the Implication lands. Reconciliation: the artefacts are MEDDPICC (kept off-stage), the performance is SPIN (on the call). (2) Challenger wants tension; Command of the Message wants fluency. Challenger’s constructive tension is uncomfortable on purpose; CMM’s situational fluency is comfortable. Reconciliation: CMM carries the message, Challenger delivers it — an SE who suppresses the tension leaves the conversation flat. (3) AI use-case discovery wants feasibility-first; CMM wants value-first storytelling. Reconciliation: discovery prioritizes feasibility, but every conversational moment also seeds CMM vocabulary, so the demo’s first sentence already matches the champion’s retelling.

There is also a clean division of labour with the account executive that interviewers probe: AEs optimize for economic-buyer continuity and MEDDPICC letter coverage; SEs optimize for process completeness and message re-tellability. The AE owns the qualification ledger and the close; the SE owns the process map, the technical win, and the demo. The successful pair matches the SE’s process map to the AE’s qualification ledger by the prospect’s terms, not product terminology — so when the AE references the SE’s discovery, they speak the customer’s language. Interview angle. “Where does the SE’s job end and the AE’s begin?” → the AE qualifies and closes; the SE owns discovery’s process map, the solution narrative, the demo, and the PoC — and the handoff is the artefact, named in the customer’s words.

The discovery round: from openers to a system spec

The discovery round runs as a 45-minute panel: a 5-minute intro, a 15-minute mock discovery call (often against a VP of Product and VP of Marketing for a tool you don’t fully know), then Q&A — and the 15 minutes are the entire round. Structure it in three beats: first third business outcome and stakeholders (“what does success look like in 90 days?”); second third the today-workflow and pain (“walk me through what happens when X fails”); final third constraints and risk (“what’s off-limits, who owns the data, what’s the worst case if we’re wrong?”). Only then summarize the spec back. Strong candidates ask scoping questions on end users, latency, auditability, and regulatory exposure before drawing a box; weak ones pitch the feature list or invent fake systems and APIs to plug into.

A practitioner truth that reframes the whole round: “having questions means you’re knowledgeable on the subject matter — and if they don’t have answers (they probably won’t), you’ll need to walk them through their own process to find them.” The SE’s job is diagnostician, not interrogator: demonstrate enough fluency to lead the buyer through their own workflow. Capture two artefacts in parallel — a MEDDPICC qualification snapshot (the AE’s) and a process map of people/process/technology/data/integration/security (yours) — and the joint narrative becomes the demo script for the next call.

Power Presenting in PreSales: Tips from a Master StorytellerPreSales Collective articleBusiness Value Assessment Overview (rep-first; 200–600% credible-range rule)Spotlight articleFive Enterprise AI Adoption Pain Points (and What’s Emerging)Geodesic Capital articleTechnical Discovery Best Practices (Vivun: 5 discovery phases)Vivun articleThe Best Storytelling Framework for Sales Engineers (hero/problem/resolution)Alpha Presales

Checkpoint

In a mock discovery the “VP of Product” says alerting is noisy and engineers waste time. What’s the highest-value next question?

A“Great — let me show you how our dashboard de-duplicates alerts.”BAn Implication question that quantifies the downstream cost: “If alerts are noisy, how many hours per engineer per week go to reconciling them, and what does that delay?”C“Which competitors are you evaluating?”

Checkpoint

A champion needs to defend the project to a steering committee you won’t attend. How do you build the ROI story?

APresent a polished 50% ROI in your own words so it feels conservative and credibleBBuild From→To→Gap across the four value levers with the customer’s numbers, anchor a 200–600% range, and hand it to the champion in their language so they can deliver it as their ownCSend the champion your standard vendor case-study deck

Checkpoint

On an AI deal, the buyer says “we’ll pull the data together for the PoC.” What’s the senior move?

AProceed to the full PoC quickly to keep momentumBDe-risk by scoping a fixed-scope data validation as the first milestone, surfacing readiness before the PoC depends on itCSwitch the conversation to model selection since data can be sorted later

Checkpoint

You have 15 minutes for a mock discovery against a VP of Product and VP of Marketing. How do you spend it?

AOpen the product and walk through the main features so they see the value fastBThree beats — business outcome/stakeholders, today-workflow/pain (with Implication questions), then constraints/risk — and summarize the spec back at the endCInvent plausible systems and APIs the case didn’t mention so you can draw an architecture

Checkpoint

A buyer pattern-matches your LLM agent to “just another rules-based bot we tried and it failed.” What does the SE do before demoing?

AReframe the buyer’s mental model first (Challenger), connecting their failed rules-engine experience to why a different approach is needed, then show the productBImmediately demo every feature to prove it’s differentCAgree it’s similar and compete on price

Interview prep

The discovery/demo round tests whether you translate a vague business problem into a system spec without leaping to a solution, quantify pain in the customer’s voice, reframe a wrong mental model, and tell an ROI story the champion can re-tell. Strong candidates drive buying logic through business outcomes and dollar figures; weak ones pitch features, invent fake APIs, or present a single optimistic ROI number.

01“Walk me through your discovery process.” → three beats (outcome/stakeholders → today-workflow/pain with Implication Qs → constraints/risk), then summarize the spec back.
02“What’s the SE’s signature question?” → the SPIN Implication question — it turns a complaint into a customer-owned number the demo then answers.
03“How do you qualify an enterprise AI deal?” → MEDDPICC’s I/C/D plus the Geodesic five pains (data readiness, integration, FDE scale, agent security, ROI clarity).
04“How do you anchor ROI?” → a credible 200–600% range from the customer’s four value levers (From→To→Gap); 50% loses the room.
05“How do you arm the champion?” → hand them the BVA in their own language so they deliver it as theirs; the story you can’t transfer dies in committee.
06“The buyer mis-models your AI as a legacy bot — what now?” → Challenger Reframe before the demo, or they pattern-match every feature to the old tool.
07“When do you walk?” → 3-yr ROI below ~100% on realistic assumptions with no strategic risk; set the ‘walk if’ threshold with the AE up front.
08“How do you handle a tough question in the demo?” → listen, summarize it back (“here’s what I heard — correct?”), then answer or say “I’ll verify against our docs.”

Going deeper, expect: “the case gave you no specific systems or constraints — what do you do?” (don’t invent fake APIs; ask scoping questions and walk the buyer through their own process — the documented strong move under ambiguity); “your champion loves it but the economic buyer is silent — what’s missing?” (you haven’t quantified the cost-reduction lever for the economic buyer, the primary motivator, or armed the champion to carry the number); and “defend your design when the buyer challenges an AI limitation” (strip the technical complexity, acknowledge the limitation honestly, and re-anchor on the quantified value — the FDE post-mortem’s critical bar). Tie each back to a number the customer produced.

Could you run a 15-minute discovery to a system spec, quantify pain with Implication questions, and arm a champion with a defensible ROI range?

New to itGetting thereConfident

Takeaways

Discovery earns the right to demo — translate a vague problem into a spec; resist the immediate solution.
The Implication question is the SE’s signature move: a complaint becomes a customer-owned number.
Run the Geodesic five-pain script; data readiness is the call-one filter that prevents a dead PoC.
Anchor ROI as a 200–600% range from the customer’s four value levers — 50% loses the room.
Arm the champion with the story in their words; one only you can tell dies in committee.
Reframe a wrong mental model (Challenger) before the demo, or features get mapped to the old tool.

Next: the capstone — run the full discovery → solution → demo arc end to end and defend it under pressure.

Sources

Free to read · better with Enzo

Learn it with Enzo

Save your progress, answer the checkpoints, and let Enzo quiz you on what you just read.

Continue with Enzo Sign up free