Before we start — who’s championing this?

• Who is the executive sponsor we should be aligned with for this initiative? Options: VP Maintenance / Reliability, Plant Manager / Operations Director, CIO / IT Director, Site General Manager, Head of Reliability Engineering, Other
• Please provide the sponsor’s name, title, and the best contact (email/phone).
• Who are the primary plant and IT decision-makers who must sign off before purchase and before deployment? Options: Plant Manager, Maintenance Manager, Reliability Engineer, IT/OT Architect, SCADA/Historian Admin, Procurement/Finance, Safety/Operations, Other
• How urgent is leadership about making a decision on this project? Options: Critical — must decide within 30 days, High — decide in 30–90 days, Moderate — 3–6 months, Longer horizon — 6+ months, Undetermined
• What are the sponsor’s top three expectations from a successful deployment (names, KPIs, or outcomes)?

Who will feel the heat if we miss the mark?

• If this program fails to deliver, whose performance or KPIs will be most exposed and why?
• Which roles will be held accountable for uptime, and which will own the results day-to-day? Options: VP Maintenance / Reliability, Plant Manager, Maintenance Foreman, Reliability Engineer, Operations Lead, IT/OT Team
• Which business units, production lines, or product families are most vulnerable to costly unplanned downtime?
• Can you share a recent event where downtime or asset failure had a material impact? Please describe the event, cost (if known), and who reacted.
• How do affected stakeholders usually respond emotionally when downtime spikes—frustration, urgency to spend, finger-pointing, or something else? Options: Frustration, Urgent escalation, Focus on temporary fixes, Budget reallocation, Search for external partners, Other

Why hasn’t this been solved already?

• What has blocked progress on reliability or predictive initiatives in the past—even when the problem was obvious? Options: Poor data quality, Integration complexity, No clear sponsor, Budget constraints, Operational disruption concerns, Technician resistance, Lack of internal skills, Vendor trust issues, Other
• Which one or two of those blockers have been most persistent for your site? Options: Poor data quality, Integration complexity, No clear sponsor, Budget constraints, Operational disruption concerns, Technician resistance, Lack of internal skills, Vendor trust issues, Other
• Tell us about a prior project that stalled: what was tried, where it stopped, and the root cause.
• How long have these blockers been in place (for example: months, years, since a merger, after an ERP upgrade)? Options: Less than 6 months, 6–12 months, 1–2 years, 2–5 years, 5+ years, Since a specific event (describe below)
• What small changes have actually helped mitigate these blockers previously—if any—and why did they stick or fail?

What would “better” actually look like for your site?

• Imagine six months after go‑live your team says: “We made the right choice.” What does that day-to-day look like?
• Which of the following outcome targets would make you feel confident this project paid off? Options: >50% reduction in unplanned downtime, 30–50% reduction in unplanned downtime, 10–30% reduction in unplanned downtime, Improve OEE by >5 points, Improve OEE by 2–5 points, Reduce reactive maintenance >30%
• What predictive model accuracy (precision/recall or lead time) would you need before you act on a recommendation? Options: Very high (>90%), High (80–90%), Moderate (70–80%), Lower but actionable (60–70%), Unsure — need to see pilot data
• Beyond metrics, what change in team behavior or capability would signal success (e.g., technicians trusting alerts, planning fewer emergencies)?
• Who on your team will be most energized by these improvements and who might resist them?

Which KPIs will prove the business case?

• Which 3 metrics will leaders use to justify continued investment if this program proceeds? Options: Unplanned downtime hours, OEE, Maintenance cost per unit, Mean time between failures (MTBF), Percent reactive work, First-time fix rate, Spare parts inventory days, Safety incidents
• Of those metrics, which are currently tracked in your systems (CMMS, historian, ERP) and where is each stored? Options: CMMS, Historian/PI, ERP, Spreadsheets, Not tracked / ad hoc
• Who owns each chosen KPI today and how frequently are they reviewed at site/executive levels? Options: Daily, Weekly, Monthly, Quarterly, Ad hoc / event-driven
• What specific targets or thresholds would trigger executive intervention or additional funding?
• Which of these KPIs do you expect to move first during a pilot, and why?

How will the decision get made (and who closes it)?

• When scope, timeline, or price are in dispute, who has final decision authority and what criteria do they use? Options: Sponsor (exec), Plant Manager, IT/OT Director, Procurement/Finance, Steering committee
• What procurement or legal milestones must we clear (e.g., security review, insurance, vendor approval board)? Options: IT Security Review, Data-sharing agreement, Procurement RFP, Insurance approval, Regulatory compliance signoff, Union/HR approval, Other
• What is your preferred approval timeline from pilot to enterprise roll‑out? Options: Pilot decision in 30 days, Pilot decision in 60 days, 3 months, 6 months, Depends on pilot results
• What are the budget constraints or funding windows we should know about (fiscal year closings, capital vs. OPEX)?
• What would make the procurement team say “yes” today—price, references, delivery timeline, or risk allocation? Options: Price, Proven references, Fast timeline, Shared risk model, Clear SLAs and acceptance criteria

Name the biggest risk—and how we prove it’s managed

• What single risk, if it happens after deployment, would make leadership wish they hadn’t approved the project?
• From this list, which three risks are most relevant to you right now? Options: Poor data quality undermines models, Integration delays with historian/SCADA/ERP, Technician adoption resistance, Security or OT policy blocks, Hidden integration costs, Vendor lock-in concerns, Insufficient business process change
• For each top risk you selected, what practical mitigation has worked or would feel convincing to you?
• Who must sign off on data access, and are there any known technical or policy constraints (segmentation, air-gapped systems)?
• How confident are you that your existing data sources (sensors, historian, CMMS) will support an initial pilot? Options: Very confident, Somewhat confident, Not confident — significant gaps, Unknown — need assessment

What small, persuasive step proves momentum?

• What low-risk pilot would convince leadership this is working (single critical asset, line, plant, or use-case)? Options: Single critical asset, Single asset class across site, Single production line, Short-term predictive-only pilot, Integration + workflow pilot (end-to-end)
• Who will be the named owner and the working team for that pilot (names/roles)?
• How long do you expect a pilot to run before we can report a reliable signal (days/weeks/months)? Options: 2–4 weeks, 1–3 months, 3–6 months, 6+ months
• What budget, people, and system access has already been committed to run a pilot? Options: Funding committed, Budget request in process, No funding yet, People/time allocated, System access pre-approved
• What would constitute a successful pilot (specific metric shifts, technician adoption targets, or acceptance criteria)?

Getting Comfortable — a quick site snapshot

• Which site or logical fleet are we focusing on in this conversation? Options: Single plant / site, Multiple plants (same region), Enterprise / cross-region program, Specific asset class (pumps/compressors/etc.), Other
• Who is the executive sponsor and the primary operational sponsor for this effort? (name & role)
• Roughly how many production-critical assets are in scope for this initial phase? Options: Fewer than 25, 25–100, 101–500, 501–1,000, More than 1,000
• What’s the most important outcome your leadership talks about when they approve reliability investments? Options: Reduce unplanned downtime, Lower maintenance costs, Improve safety/compliance, Optimize spare parts, Improve OEE, Other
• Tell us briefly about a recent maintenance problem that frustrated you or your team (what happened, who felt the pain, and why it stuck)

Show Me What’s Under the Hood

• If I asked an operator to sketch the asset train that most often causes shutdowns, would anyone on your team be able to do that from memory—and what would they get wrong?
• Which asset categories are highest priority for reliability improvements right now? Options: Pumps, Compressors, Motors/drive trains, Heat exchangers, Valves, Transformers/generators, Conveyors/rotating equipment, Other
• For each high-priority asset type, what percentage of failures are mechanical vs. electrical vs. process/control? Options: Mostly mechanical (>70%), Mostly electrical (>70%), Mostly process/control (>70%), Mixed (no dominant cause), Unknown / not tracked
• Do you have an asset registry or CMDB that maps serial numbers, locations, and BOMs for these fleets? If yes, where is it maintained? Options: CMMS (see list later), Excel/Spreadsheets, PLM/ERP, No centralized registry, Other
• What documentation or tribal knowledge exists today for failure signatures (photos, sound recordings, historical work-orders, tribal notes)? Options: Rich — photos/logs/work-orders, Partial — some work-orders & notes, Minimal — mostly tribal knowledge, None

Why Does This Keep Happening?

• When an asset fails today, do you find the team fixing symptoms (quick patch) more often than addressing root cause—and what does that cost you in repeat failures? Options: Mostly symptoms / frequent repeats, Mixed — sometimes RCA, sometimes patch, Consistently root-cause focused, Unknown
• Which failure modes produce the worst financial or safety impact at your site (name top 3 and an example incident for each)?
• How often do you perform formal root-cause analysis (RCA) after significant incidents? Options: Always for every incident, Usually for major incidents, Rarely, Never / informal only
• What is your current average time-to-repair (MTTR) and average time-between-failures (MTBF) for the highest-priority assets?
• How does unplanned downtime show up in your operating metrics and conversations (lost throughput, penalties, safety stoppages)? Give a recent quantified example if possible.

Where Does Your Data Trip You Up?

• If your predictive analytics had one glaring blind spot today, where would it be—missing sensors, noisy signals, disconnected work-orders, or something else? Options: Missing sensors, Noisy/no timestamped signals, Disconnected CMMS/work-orders, Incomplete asset metadata, Historic data retention gaps, Other
• Which systems hold your operation and maintenance data today? Options: IBM Maximo, SAP PM, Infor EAM, Oracle eAM, Azure/SQL data lake, OSI PI, Wonderware/AVEVA Historian, Custom historian / SCADA, No formal system, Other
• For signals you rely on (vibration, temp, pressure, current), what are typical sample rates and retention periods? Options: High-frequency (>=1Hz) retained long-term, Periodic (minutes) retained medium-term, Event-based or daily snapshots, Only alarm events retained, Unknown / inconsistent
• How reliable is the mapping between sensor tags and physical assets today (can you programmatically connect a tag to an asset record)? Options: Fully mapped and maintained, Partially mapped (~50–80%), Poorly mapped (<50%), Not mapped / manual only, Unknown
• Where do you experience the most data friction—permissions/security, network segmentation, field instrument calibration, or historical cleanliness? Please describe with examples.

Integration Reality Check (IT & OT)

• Which integrations feel like 'no-brainers' to your IT team—and which ones would they tell leadership are risky if asked bluntly?
• Do you have any of the following constraints that affect integrations or cloud access? Options: Air-gapped network, Strict firewall rules / VPN only, No external cloud access allowed, SIEM / strict logging requirements, Standard enterprise API access allowed, Other
• Which of these integration touchpoints must be included for a minimal viable deployment? Options: Historian time-series, CMMS work-order sync, ERP parts/master data, SCADA real-time points, Identity/SSO, Mobile workforce app, Other
• Who in IT/OT will need to sign off on network changes and data feeds, and how long does that approval typically take?
• Are there preferred protocols or middleware (OPC UA, MQTT, REST APIs, MQTT over VPN) that we should plan to use? Options: OPC UA, MQTT, REST API, Modbus/legacy RTU, Direct DB access (SQL), Custom middleware, Other

Who Owns the Problem (and the Solution)?

• If a high-confidence failure alert arrived at shift start, whose team is expected to act—and would that person have the authority and spare capacity to respond? Options: Shift technician/maintenance, Reliability engineer, Operations/Control room, Maintenance planner/scheduler, We'd debate it — unclear
• Who will be the day-to-day owner for validating model outputs and converting them into work-orders?
• How do technicians currently receive and complete work-orders (paper, mobile app, CMMS mobile, other)? Options: Paper printouts, CMMS web portal, CMMS mobile app, Paper + mobile photos, Email / PDF, Other
• What level of mobile adoption do field teams have today—are they comfortable using an app for guided inspections and closing tasks? Options: High — regularly use mobile apps, Moderate — intermittent mobile use, Low — prefer paper or radio, None — would require training & devices
• Who inside the organization will champion adoption and training, and what has worked for them in prior rollouts?

What Would Real Value Feel Like?

• If we reduced your unplanned downtime by 30% in 12 months, how would that change conversations with your CEO/COO—what would they celebrate or demand next?
• Which success metrics will your team use to judge a pilot: downtime hours avoided, work-orders reduced, OEE lift, MTBF improvement, parts-cost reduction, or something else? Options: Downtime hours avoided, OEE improvement, MTBF increase, MTTR reduction, Work-order volume reduction, Parts cost reduction, Operational safety incidents, Other
• What level of predictive accuracy or lead time would make an alert actionable for your team (e.g., 80% accurate with 7 days lead time)? Options: High accuracy, short lead (>=85% & <7 days), Moderate accuracy, longer lead (70–85% & 7–30 days), Lower accuracy OK for advisory use (<70%), Unsure — need benchmarking
• How do you want outcomes documented and reported (dashboard cadence, automated reports, integrated CMMS tickets)? Options: Live dashboard, Weekly automated reports, CMMS-integrated work-order evidence, Executive summaries monthly, Other
• What are realistic timelines you’d accept for measurable value (90 days, 6 months, 12 months)? Options: Within 90 days, 3–6 months, 6–12 months, Longer than 12 months, Unsure

Roadblocks & Deal Breakers

• What single technical or organizational issue would cause you to pause or cancel a deployment no matter the price or promise?
• Which of these risks worry you most about adopting a predictive platform? Options: Poor data quality undermines models, Lengthy, costly integrations, Technician resistance to new workflows, Vendor lock-in / lack of transparency, Regulatory or audit concerns, Budget overruns
• How strict are your validation & acceptance gates for new tooling (do you require POC acceptance, formal FAT/SAT, security penetration testing)? Options: Formal POC + acceptance tests required, POC encouraged but flexible, Security testing mandatory, No formal gates — flexible, Other
• Have you run pilots with other vendors recently? If yes, what specifically led to success or why they failed?
• Are there contractual or insurance constraints about analytics models, data residency, or third-party algorithms we should know now? Options: Data residency constraints, Prohibit external model hosting, Require explainable models, No special constraints, Other

Quick Wins & Low‑Regret First Steps

• Where could we demonstrate clear, measurable impact inside 90 days — which 1–3 assets or processes are the best low-risk bets?
• Which of the following is most commonly available for those candidate assets today? Options: High-quality time-series sensors, Partial sensor coverage + manual logs, Only alarm-based signals, Work-order history but no sensor data, No usable data
• What minimal data access would you approve quickly for a pilot (historian read-only, CMMS sample export, edge interface)? Options: Historian read-only, CMMS sample export, Edge gateway install, Manual CSV uploads, None — approvals required
• Who needs to be in the room for a 4‑week scoping sprint to unblock a pilot (names/roles)?
• What early success signal would change the political dynamic internally (reduced emergency work-orders, one avoided shutdown, technician adoption)? Options: Avoided shutdown, Reduced emergency work-orders, Technician positive feedback, Validated predictive lead time, Other

Let’s Commit to the Next Visible Step

• If we could guarantee one measurable pilot outcome in 90 days, would you be willing to commit the needed data and one asset group to prove it? Options: Yes — ready to commit, Maybe — need internal approvals, No — not ready
• What approvals or procurements must happen before data access and integration work can begin, and who owns obtaining them?
• Which communication cadence do you prefer during a pilot (weekly tactical, biweekly steering, monthly exec), and who should receive each update? Options: Weekly tactical + field team, Biweekly steering + reliability leads, Monthly exec summary + leadership, Ad‑hoc updates only, Other
• What would make you feel confident we’re the right partner after the first 90 days (transparency of models, clear ROI, documented runbook, technician buy-in)? Options: Transparent models & explainability, Clear quantified ROI, Operational runbook & SOPs, Technician adoption evidence, Other
• Is there anything else we should know that would change how we scope data ingestion, model selection, or pilot success criteria?

A Fast Siting Check — Getting Everyone on the Same Page

• Which site or plant are we preparing for this deployment (site name, region, and business unit)?
• Who is the reliability executive sponsor for this site, and who will be our primary day-to-day contact? Options: VP Maintenance / Reliability, Plant Manager, Site Reliability Engineer, IT Lead / Architect, Operations Superintendent, Other
• What is the target go-live window for the initial deployment (earliest possible date and any immovable deadlines)? Options: Within 30 days, 30–60 days, 61–120 days, More than 120 days, Not sure / TBD
• Which production lines, units, or asset classes are in scope for Day One (list top 3–5 priority assets or systems)?
• Is there an existing project or change freeze window we must avoid during cutover (dates or recurring blackout periods)? Options: No freeze, Known dates — provide details, Recurring monthly/quarterly blackout, Yes — ask operations for dates, Unsure

If Your Data Could Talk — How Honest Would It Be?

• How confident are you that historical sensor and historian data are complete and trustworthy enough to seed our predictive models? Options: Very confident — high-quality historical data exists, Somewhat confident — gaps but usable, Low confidence — frequent gaps/noise, Not confident — little to no historical data
• Which historian, SCADA, or time-series platforms store the sensor signals we need (select all that apply)? Options: OSIsoft PI / AVEVA PI, AVEVA Historian, GE Proficy / Historian, OSIsoft Cloud Services, ControlLogix/PLC local historian, In-house/custom historian, No historian / only edge devices
• How do you typically access historical data for analytics today? Options: Direct DB queries, Vendor APIs, Flat-file exports (CSV), Historian connectors (PI SDK/PI AF), Third-party ETL, We don't currently access it
• Can you share a recent example where data quality impaired decision-making or caused a false alarm? Tell us what happened and how it felt for the team.
• What are the most common data quality issues you expect us to encounter (select all that apply)? Options: Missing timestamps / gaps, Incorrect units or scaling, Duplicate or ghost tags, High-frequency noise / spikes, Drift or sensor bias, Inconsistent tag naming, Latency/streaming delays

Do We Really Have the Right Sensors Where It Counts?

• What would surprise you if we told you an asset lacked the sensor coverage required for reliable predictions? Options: That would be surprising — sensors are well-instrumented, Not surprising — many assets have minimal sensors, Some assets are well-covered, others are not, Unsure — haven't audited sensors recently
• For the top priority assets, which physical signals are available today (temperature, vibration, pressure, flow, current, etc.)? Please list by asset or asset group.
• Are there undocumented or 'zombie' tags in your systems that teams still reference informally? Options: Yes — several, A few, None that we know of, Unsure — need to audit
• How often are field sensors calibrated or replaced, and who owns that maintenance schedule? Options: Monthly, Quarterly, Annually, On failure, No formal schedule, Owned by third-party vendor
• If we find missing or mis-labeled sensors during mapping, how do you prefer we handle remediation planning? Options: Raise a work order for sensors, Document and proceed with available signals, Pause deployment until fixed, Create temporary data proxies / estimators

Who’s Holding the Rope — Roles, Responsibility, and Real Availability

• If production alarms start firing during a cutover, who has authority to pause or push the cutover back — and how fast can they be reached? Options: Plant Manager, Operations Lead, Reliability/Site Sponsor, IT/OT Manager, On-call Control Systems Engineer, Other
• Do you have named owners assigned for these areas: data access provisioning, tag mapping validation, security approvals, and technician enablement? Options: All owners named and confirmed, Some owners named — others TBD, No owners assigned yet, Owners exist but not available during deployment window
• How available will your SMEs (control systems, reliability, and operations) be for workshops, validation, and rapid triage during the first four weeks? Options: Full-time / dedicated, Part-time with scheduled blocks, Ad-hoc / as available, Limited availability — must schedule far in advance
• Describe the ideal escalation path for production-impacting issues during deployment (who gets notified, in what order, and SLA for response).
• Are there union, shift, or contractor constraints that affect who can perform cutovers or sensor work and when? Options: Yes — major constraints, Some constraints, No constraints, Unsure — need to check

If Deployment Fails — What Does That Cost (and Who Notices)?

• What is the material impact if a deployment mistake causes an asset shutdown or production loss during the cutover window? Options: >$250k/day, $50k–$250k/day, $10k–$50k/day, <$10k/day, Intangible/quality impacts only
• Which integration points are most sensitive and must be validated first (e.g., EAM/CMMS work-order creation, ERP inventory sync, historian writes, real-time control loops)? Options: CMMS/Work orders, ERP spare parts, Historian writes/reads, SCADA control loop interfaces, Mobile technician workflows, Alarming/notifications
• How closely does your test/staging environment mirror production for these systems and integrations? Options: Full parity — very close, Mostly similar with some differences, Limited parity — not reliable, No staging environment
• Do you have rollback or back-out procedures for integrations and cutovers? If yes, please summarize how long rollback takes. Options: Comprehensive — rollback < 1 hour, Documented — rollback 1–4 hours, Basic — rollback may take a day, None / ad-hoc
• Who signs the 'stop-the-press' approval if an integration test reveals a critical risk during deployment? Options: Site Sponsor, Plant Manager, IT Security Lead, Operations Lead, Joint decision meeting

How Will We Know Day One Worked — Defining Real Acceptance

• What are the three measurable acceptance criteria that would convince you Day One is a success (e.g., model uptime, data completeness %, number of validated alerts)?
• For predictive models, what minimum performance would you accept before we move from pilot to full use (precision/recall, lead time, or KPI improvements)? Options: High (>90%), Good (75–90%), Useful (60–75%), Baseline exploratory — prioritize learning
• Which reports or dashboards must be populated and verified on Day One for operations and reliability teams to trust the system? Options: Asset health dashboard, Alert / notification feed, Work-order recommendations, Model performance report, Mobile technician checks, Executive summary / KPI report
• How will you formally sign off on acceptance tests (who signs, format: email, ticket, signed doc)? Options: Email confirmation, CMMS/Project ticket sign-off, Signed acceptance document, Formal change control closure, Other
• If the first month shows model drift or higher false alarms than expected, what remedial actions are you willing to commit to (re-training cadence, additional sensors, operational tuning)? Options: Retrain models quickly, Add / repurpose sensors, Adjust alarm thresholds, Delay full rollout, Other

Practical Logistics — Access, Security, and Data Handshake

• What method will your IT/OT team use to grant us access to systems: VPN, site DMZ service account, API keys, SAML service principal, or on-prem agent? Options: VPN with jump host, DMZ-hosted API, Service account / API key, SAML/OAuth service principal, On-prem connector / agent, Other
• Which data transfer protocols and interfaces must we support or plan for (select all that apply)? Options: OPC UA, Modbus TCP/RTU, MQTT, REST APIs/HTTP, Database ODBC/JDBC, File drops / SFTP, Custom / Proprietary
• Are there formal security or vendor onboarding requirements we must complete before any credentials are issued (e.g., risk assessment, pentest, insurance, background checks)? Options: Yes — strict onboarding, Some documentation required, Minimal vendor checks, No formal requirements / fast-track
• What level of logging and auditability do you require for data access and model actions (retention window, SOC report needs, SIEM integration)? Options: Full audit + SIEM, Standard logs retained 90 days, Basic logging only, Unsure — need guidance
• Who in IT/OT will complete access provisioning and what typical lead time do they need for creating service accounts and firewall rules? Options: < 3 business days, 3–7 business days, 1–2 weeks, > 2 weeks, Not sure / depends

Training, Handover, and Making This Stick

• What style of technician enablement works best at your site: hands-on shadowing, short focused workshops, distributed e-learning, or train-the-trainer? Options: Hands-on shadowing, Short on-site workshops, E-learning modules, Train-the-trainer, Blended approach
• How many technicians and supervisors need access to mobile workflows on Day One, and what devices/OS do they use (iOS, Android, rugged devices)?
• What change-resistance have you seen historically when introducing new digital workflows to technicians, and what helped overcome it?
• Who will own post-deployment governance: maintaining tag mappings, model retraining cadence, and user access reviews? Options: Site Reliability Team, IT/OT, Operations, Vendor-managed, Joint governance
• What success communications would reassure leadership after cutover (example: 7-day status, 30-day KPI snapshot, executive one-pager)? Options: 7-day rapid status, 30-day KPI report, Executive one-pager, Weekly operations digest, Ad-hoc deep-dive as needed

Making a Real Plan — Who Signs, What We Need, and Next Milestones

• From your perspective, what are the non-negotiable artifacts we must have before starting cutover (access tickets, owner list, test plan, rollback plan)? Options: Access tickets / credentials, Named owner list, Test plan & scripts, Rollback/back-out plan, Acceptance criteria document, Security approvals
• Please provide the names and email addresses (or ticket numbers) of the individuals who will approve each of these artifacts.
• What is the single biggest risk that would make you pause deployment today, and what would it take to mitigate that concern?
• How quickly can your teams provide the first set of mapped tags and historical sample extracts for our initial model validation? Options: Within 3 business days, Within 1–2 weeks, 2–4 weeks, Longer than 4 weeks / TBD
• Are you willing to schedule a 60–90 minute readiness workshop this week to review access, mappings, and sign-off on the Day One acceptance criteria? Options: Yes — schedule now, Yes — schedule next week, Maybe — need to confirm attendees, No — prefer asynchronous review

Start Here: Scope, People, and Why We’re Talking

• Which site(s) or asset classes should we focus on for this discovery? Options: Single site/plant, Multiple sites/plants, Specific unit(s) (e.g., furnace, boiler, cracker), Specific asset class (pumps, compressors, turbines, conveyors), Enterprise-wide
• Who is sponsoring this initiative and who will be our primary operational contact? Options: VP Maintenance/Reliability, Plant Manager, IT/OT Lead, Operations Manager, Reliability Engineer, Other (please name)
• What is the immediate trigger that pushed this project onto your agenda now? Options: Recent major unplanned outage, Rising maintenance costs, Insurance/compliance requirement, Operational performance targets (OEE), Corporate digital transformation, Other
• Do you have a target decision date or event (budget cycle, audit, outage) driving timing? Options: Immediate (<1 month), 1–3 months, 3–6 months, 6–12 months, No fixed date
• Who will need to sign off for go-live and who controls budget approval (names/titles)?
• What would success look like to your sponsor in the first 6 months—briefly describe top outcomes?

What If We’re Underestimating How Much It’s Costing You?

• If your most critical unit went down for one full day, roughly how much would it cost the business? Options: < $50,000/day, $50k–$200k/day, $200k–$1M/day, > $1M/day, Don't know / need help calculating
• Describe the most recent unplanned shutdown for scope assets—what failed, how long did it take to recover, and what was the root cause?
• How many unplanned stops have those asset classes experienced in the last 12 months? Options: 0, 1–3, 4–10, 11–25, >25, Don't know
• Which of the following hidden costs occur when outages happen at your site? Options: Overtime labor, Expedited parts/shipping, Lost production / missed contracts, Quality rework, Safety incidents / near misses, Regulatory penalties, Other
• Who currently owns the calculation and reporting of downtime cost and production loss? Options: Operations / Plant Control, Reliability / Maintenance, Finance, Production Planning, No single owner
• How do these failures feel to the teams who respond—are they outraged, resigned, constantly firefighting, or something else?

What’s Actually Running Under the Hood (Assets, Systems, and Access)

• How complete is your asset registry for the target scope (tags, hierarchy, BOMs)? Options: Comprehensive (>90%), Mostly complete (60–90%), Partial (30–60%), Sparse (<30%), No registry
• Which systems currently hold the critical data we’ll need? Options: CMMS (SAP PM, Maximo, etc.), Historian (OSIsoft PI, Aspen InfoPlus.21), SCADA/DCS, ERP, Spreadsheets / local files, Third-party vendor systems, Other
• Do you have a single canonical asset ID mapped across those systems, or will we need to build and validate mappings? Options: Canonical IDs exist across systems, Partial overlap; mappings required, No consistent IDs; full mapping required, Unsure
• What sensor and signal types are available on priority assets (select all that apply)? Options: Vibration (accelerometers), Temperature, Pressure, Flow / Level, Motor current / drive telemetry, Acoustic sensors, None/limited
• How accessible are real-time and historical time-series feeds (streaming API, historian extracts, only manual reports)? Options: Real-time streaming APIs available, Historical batch extracts available, Both streaming and historical available, Only manual exports/spreadsheets, Unknown / need to check
• Who will be our technical data contacts for each system and how quickly can they provide credentials or extracts?

Where the Data Quietly Breaks (Quality, Labels, and Gaps)

• Which of these data issues most often undermines analytics accuracy at your site? Options: Missing tags or measurements, Irregular sampling / inconsistent rates, Time skew / timezone problems, Noisy or biased sensors, Inconsistent naming or metadata, Truncated historical windows, Insufficient failure labels
• Estimate the percentage of historical data you would trust for model training without significant cleaning. Options: >90%, 70–90%, 40–70%, 10–40%, <10%, Don't know
• What existing processes, if any, do you use to validate or clean historian/sensor data? Options: Automated QC pipelines, Periodic manual review by engineers, Ad-hoc fixes when investigating failures, No formal process
• How often do you discover sensors or tags were misconfigured only after a failure occurs? Options: Almost always, Often, Sometimes, Rarely, Never
• For supervised model training, approximately how many high-confidence labeled failure events can you provide for your top 5 asset types? Options: >100, 20–100, 5–20, 1–5, None, Don't know
• Are you willing to allocate engineering time to tag cleanup, historian retention changes, or sensor recalibration if it materially improves model outcomes? Options: Yes—dedicated engineering hours available, Yes—a short data-cleanse project, Only if vendor performs most work, Not currently resourced

Who Pulls the Emergency Lever? Decision Roles, Incentives, and Politics

• Which stakeholders’ KPIs are most likely to conflict during evaluation and procurement? Options: Maintenance / Reliability (MTTR, availability), Operations (throughput, OEE), IT/OT (security, support), Finance / Procurement (TCO, ROI), Safety & Compliance, Supply Chain / Spares
• Please list named decision-makers and a one-line summary of what success looks like to each (name, title, top priority).
• Which commercial model do you prefer or expect (so we can align proposals)? Options: SaaS / subscription, Perpetual license + services, Outcome-based pricing, Pilot-to-contract, Undecided / open
• What procurement gates and timeline should we plan for (legal, security review, steering committee)? Options: Fast-track (<1 month), Short (1–3 months), Standard (3–6 months), Extended (>6 months)
• Who will own acceptance testing and who signs operational acceptance at the plant? Options: Plant Manager, Reliability Lead, IT/OT Lead, Cross-functional steering committee, External auditor
• How do you prefer executive updates during the project (cadence and format)? Options: Weekly written summary, Biweekly steering calls, Monthly executive review, Ad-hoc as needed

If Predictions Were Right, What Would Actually Change?

• Select the measurable targets you want to achieve in year one. Options: Reduce unplanned downtime (%), Improve OEE (%), Reduce maintenance labor hours (%), Lower spare parts spend (%), Increase predictive alert lead time (hours/days), Improve safety / reduce incidents
• What minimum predictive performance would you accept before operations act on an alert (choose the best match)? Options: Precision >= 90%, Precision 70–90%, Precision 50–70%, Prefer defined lead-time guarantee over precision, Unsure—need vendor recommendations
• Which actions should be automated immediately on an alert versus require manual supervisor approval? Options: Auto-create CMMS work-order, Push mobile notification to technician, Supervisor approval before scheduling work, Trigger operator safe-shutdown, Advisory only (no automatic actions)
• What KPIs and reporting cadence will you use to declare the project a success (and how are those KPIs measured today)?
• If early pilot results don’t hit targets, which corrective actions would you consider acceptable? Options: Retrain models / tune thresholds, Add or recalibrate sensors, Adjust operational workflows, Extend pilot duration, Pause/terminate

What Could Derail This Before We Start (Risks, Constraints, and Past Lessons)

• Which single risk worries you most for a digital reliability project? Options: Poor data quality, Integration delays, Technician adoption resistance, Budget cuts, OT security/policy restrictions, Vendor lock-in
• How confident are you that we can complete integrations to historian / CMMS / ERP within your desired timeline? Options: Very confident, Somewhat confident, Unsure, Not confident
• What change-management or training resources have you allocated to ensure technicians adopt mobile workflows? Options: Dedicated change manager, Trainers + super-user program, Support from maintenance leads, No resources allocated yet
• Have past digital initiatives faced internal resistance—what happened, and what did you learn?
• Which constraints (budget, headcount, OT security, regulatory) would force us to pause work? Options: Budget, Headcount, OT security/policy, Regulatory constraints, Other
• Who will be responsible for ongoing model monitoring, triage of false alerts, and continuous improvement?

How We’ll Validate Together—and Who Signs Off

• What acceptance tests must pass before you will sign operational acceptance (pick all that apply)? Options: Accuracy benchmarks on holdout data, Live pilot performance over defined period, End-to-end integration & dataflow tests, Technician mobile workflow UAT, Security / penetration testing
• Which assets or production lines should be included in the validation pilot (please list names/tags)?
• What minimum sample size or duration will you accept for a valid pilot (events or weeks)? Options: >100 events, 20–100 events, 1–20 events, Time-based: >12 weeks, Time-based: 4–12 weeks, Other
• Who must provide formal sign-off at pilot completion and in what format (email, CFT approval, PO)? Options: Written email sign-off, Cross-functional team approval, Formal PO / contract amendment, Executive sponsor sign-off
• Which metrics should appear in the validation report and dashboard to convince your team to move to full deployment? Options: False-positive rate, True-positive rate, Average lead time to failure, Operational impact (hours saved), Cost avoidance / ROI estimate, Technician adoption and completion rates
• What are acceptable thresholds for mobile workflow UAT (for example: completion rate, data capture completeness)?