Quick intro — who’s on this mission?

• Who is sponsoring this initiative from your executive team? Options: Chief Procurement Officer (CPO), VP/Head of Procurement, Head of Category Management, Chief Financial Officer (CFO), Chief Information Officer (CIO), Other (please name)
• Who will be the day-to-day owner for the pilot (name + role)?
• Which stakeholders do you expect to be involved in decision reviews for this project? Options: Procurement / Category Management, Finance / AP, IT / Data Security, Legal / Compliance, Business Unit Leaders, Treasury, Other (please specify)
• How committed are each of the named stakeholders to allocating time and attention during the pilot? Options: Fully committed, Somewhat committed but limited bandwidth, Initially committed, may wane, Not committed yet / needs convincing
• Tell us about one prior analytics or classification project that succeeded or failed here — what was the outcome and who owned it?

Are decisions stalling where you least expect?

• When a purchasing insight or analytics recommendation is presented, how are decisions typically made? Options: Single owner signs off, Small steering committee, Functional consensus required, Requires cross-functional executive approval, Varies by decision type
• Which approvals have historically delayed projects like this (select all that apply)? Options: IT security signoff, Legal/Procurement contracting, Budget/Finance approval, Business unit buy-in, Data access approvals
• How long does it typically take to get a final commercial/go-no-go decision once a pilot completes? Options: <2 weeks, 2–4 weeks, 1–2 months, 2–3 months, Longer / unpredictable
• If a board-level question resurfaces mid-pilot (e.g., spend with top suppliers), what happens—who escalates and how quickly? Please describe.
• Which outcome would be a non-starter for your leadership (i.e., would cause them to reject the project regardless of other benefits)?

Who actually touches the data—and who’s supposed to own the truth?

• Which systems contain the spend data we’ll need for the pilot (select all that apply)? Options: ERP (AP) - specify system, PO module - specify system, Purchasing card program (p-card), T&E systems, Supplier master / MDM, Data warehouse / lake, Other (please specify)
• Who is the named data owner for AP/PO/p-card data today (team and contact)?
• How clean or fragmented is your supplier master today (e.g., many duplicates, inconsistent naming, missing tax IDs)? Options: Highly clean and governed, Moderately clean with known issues, Fragmented with many duplicates, Chaotic / unreliable
• How frequently are AP/PO/p-card extracts available for analysis (and in what format)? Options: Daily CSV/flat files, Weekly extracts, Monthly extracts, Ad-hoc/manual pulls, Accessible via direct DB query / API
• Describe one concrete example of where data mismatch or missing fields have caused a bad decision or rework recently.

If this went perfectly, what would your procurement team celebrate?

• Which measurable outcomes matter most for this pilot (select top 3)? Options: Classification accuracy (%), Supplier name normalization rate, New savings opportunities identified, Time saved vs manual Excel work, Number of categories rationalized, Speed to production
• What classification accuracy would make procurement trust the dashboards enough to stop the manual baseline work? Options: >95%, 90–95%, 85–89%, Below 85%
• What minimum sample size (months and business units) do you consider credible for the pilot’s validation? Options: 1 month / 1 BU, 3 months / 1–2 BUs, 6 months / multiple BUs, Other (please specify)
• Beyond accuracy, what else must be true for the pilot to be considered successful (e.g., normalized suppliers, traceable audit logs, clear handoffs)?
• Who will be responsible for signing off on pilot success—name, role, and acceptance criteria they own?

What’s IT’s posture—partner, gatekeeper, or a hardened firewall?

• What is IT’s biggest concern about allowing external classification/testing on procurement data? Options: Data security/compliance, Integration complexity, Maintenance burden, Vendor management policy, Other (please specify)
• Which security or compliance standards must we meet to access data (select all that apply)? Options: SOC 2, ISO 27001, GDPR, HIPAA, Internal data handling policy only, Other (please specify)
• Which data transfer methods are preferred/allowed by IT for a pilot? Options: SFTP encrypted extracts, Secure API, Direct DB read-only account, Hosted staging in customer cloud, No external transfer allowed
• How long does IT typically take to approve data access requests and run a security review? Options: <1 week, 1–2 weeks, 2–4 weeks, 4–8 weeks, Longer / variable
• Who in IT should we loop in now to avoid surprises later (name, role, and preferred contact)?

Money, contracts, and paperwork — who really calls the shots?

• Which team owns vendor contracting and commercial terms for analytics pilots? Options: Procurement, Legal, Vendor Management Office, Finance, IT, Other (please specify)
• Are there standard contracting vehicles you prefer for pilots (e.g., PO, MSA, SOW, proof-of-value agreement)? Options: PO, MSA + SOW, Short PO with T&Cs, Proof-of-value agreement, Requires legal review each time
• What internal budgets (or charge codes) would be used if this moves from pilot to production?
• Are there procurement rules that would block a quick pilot (e.g., mandatory RFP, approved vendor list)? Options: Yes — RFP required, Yes — approved vendor list only, No — flexible for pilots, Unsure / need to check
• Who must sign commercial terms for a production roll-out and what typical lead time do they require?

Timeline pressure — is it real, or a politely worded wish?

• What is the target date you’re working toward for having production-ready analytics? Options: <1 month, 1–2 months, 2–3 months, 3–6 months, 6+ months
• Is there an external or internal event driving that timeline (e.g., board meeting, category review, budgeting)? Options: Board meeting, Quarterly category review, Budget cycle, Vendor consolidation decision, No specific event, Other (please specify)
• If we miss the desired timeline, what are the concrete consequences for your team or for this initiative?
• How much calendar flexibility exists—are milestones firm or negotiable? Options: Firm (fixed dates), Some flexibility (+/− 2–4 weeks), Negotiable, Undetermined
• What would be an acceptable phased approach if full production by your target date is unrealistic?

Risk radar — what keeps you awake at night?

• Which of the following risks worry you most for this pilot (select top 3)? Options: Classification accuracy / trust, Data leakage or compliance, Disrupting business processes, Inadequate business adoption, Hidden integration complexity, Supplier disputes from reclassification
• Have you experienced vendor disputes or supplier confusion after reclassifying spend before? If so, what happened?
• If classification accuracy is lower than expected initially, what would your team need to regain confidence? Options: Additional labeling support, Iterative tuning period, Higher transparency / explainability, Fallback to manual review, Other (please specify)
• What controls or audit trails must be present to satisfy internal audit or compliance reviewers?
• Who will be our escalation contact if an issue arises during the pilot (name + role)?

Signal vs noise — how will you know we succeeded?

• Which specific KPIs will you use to accept or reject the pilot (select all that apply)? Options: Classification accuracy (%), Normalized supplier match rate, Net new savings identified ($), Reduction in prep time for category review, Number of categories with >X% confidence, Stakeholder satisfaction score
• Who will perform the validation checks—procurement analysts, finance, or a joint team? Options: Procurement analysts, Finance / AP, Joint procurement + finance, 3rd party auditor, Other (please specify)
• What acceptance thresholds must be met for each KPI (please specify metric + threshold)?
• How will we document and archive the pilot decisions so they are auditable for future category reviews?
• Who signs the handoff for production and what documentation do they require?

Immediate next steps — who does what in the next 7–14 days?

• Which three actions would move this pilot forward fastest right now?
• Who will provide the first data extract and by when (name, role, and target date)?
• Which stakeholder should we schedule a 30–60 minute kickoff with to confirm scope and success criteria? Options: CPO / Sponsor, Pilot Owner (Procurement), IT Data Owner, Finance / AP Lead, Legal / Procurement Contracting
• What information or assurances do you need from us before we can begin ingesting a data sample? Options: Security documentation (SOC2/ISO), Data handling agreement, High-level runbook, Pilot SOW, Other (please specify)
• What else should we know now to avoid surprises—hidden dependencies, upcoming freezes, or cultural dynamics?

Getting Comfortable with Your Data

• How often does your team run a formal category review that requires consolidated spend data? Options: Monthly, Quarterly, Bi-annually, Annually, Ad hoc / as needed
• Which data sources do you currently pull together for a typical category review? Options: ERP general ledger, AP ledger / vouchers, PO system, Purchasing card (p-card) transactions, Expense reports, Supplier invoices (PDF/image), Third-party procurement tool, Other
• Who on your team owns gathering and preparing those exports today? Options: Category manager, Procurement operations, Finance/FP&A, IT/data team, Shared across teams, External consultant
• What file formats and extracts are you typically able to produce for a review (be specific)? Options: CSV/Excel, Flat text/pipe-delimited, Database export (SQL), API access, PDF invoices only, Other
• Roughly how many rows/transactions do you hand off for a standard three-month sample across two business units? Options: <10k, 10k–50k, 50k–200k, 200k–1M, >1M
• If you had to summarize in one sentence what you wish your spend data would stop doing, what would it be?

If Your Spend Data Could Talk, What Would It Confess?

• What decision did you recently delay or make blind because the spend data didn’t inspire confidence?
• How often do you encounter spend records with missing or misleading vendor names (e.g., ‘ACME PO #1234’, payment processors, or acquirers)? Options: Almost always, Often, Sometimes, Rarely, Never
• Give an example of a taxonomy or commodity code that routinely maps incorrectly for your team—and what the downstream consequence was.
• When you compare a freshly classified spend cube against your manual baseline, what differences create the most doubt (classification accuracy, supplier normalization, spend mapping, currency/units)? Options: Classification accuracy, Supplier normalization, Mismatched PO/AP flows, Currency/date issues, Taxonomy gaps, Other
• Which business stakeholders get most vocal when the data isn’t trustworthy (CPO, category leads, finance, audit, lines of business)? Options: CPO, Head of Category, Finance/FP&A, Controller/internal audit, Business unit leaders, Other
• Tell me about a moment when the team almost gave up on a review because the data prep felt impossible—what happened and how did it land politically?

Where Things Break: The Taxonomy and Normalization Blindspots

• How confident are you that your current taxonomy covers the real spend categories you care about? Options: Highly confident, Mostly confident, Uncertain, Not confident at all
• Which of these taxonomy problems best describes your reality? Options: Too coarse—can’t separate strategic vs tactical spend, Too many custom codes—no consistency, Legacy ERP codes misused, No central taxonomy; team-specific labels, Other
• Approximately what percentage of transactions regularly land in an ‘uncategorized’ or ‘miscellaneous’ bucket? Options: <5%, 5–15%, 15–30%, 30–50%, >50%
• How many distinct vendor name variants do you see for your top 50 suppliers (order-of-magnitude)? Options: 1–2 variants, 3–5 variants, 6–10 variants, 10–25 variants, >25 variants
• What manual rules or heuristics do you rely on today to map items to categories (e.g., description parsing, PO matching, cost centers)? Options: Description keyword rules, GL-code mapping tables, Reference PO numbers, Supplier-based rules, Manual line-by-line review, Other
• How often do you have to create one-off mappings during a review because the model or rules failed? Options: Every review, Most reviews, Occasionally, Rarely, Never
• If you had a magic switch to normalize supplier names perfectly, what decision would you make differently first?

The Two Weeks No One Talks About — Your Hidden Manual Work

• When you say it takes two weeks of spreadsheet work before analysis can start—what are the specific steps during those two weeks?
• Which of these activities consumes the most calendar days in that effort? Options: Exporting and cleaning data, Vendor name reconciliation, Mapping GL/PO codes to taxonomy, Removing duplicates and refunds, Manual line-item classification, Stakeholder data signoff
• Who in your organization spends the most time on these steps and how many FTE-days does it add up to per review? Options: Procurement analyst, Category manager, Finance analyst, IT/data ops, External consultant, Shared
• How do you handle transactions that span categories or have split charges—do you adjust amounts manually, ignore splits, or use another approach? Options: Manual splitting, Assign full amount to one category, Ignore splits, Custom rules, Other
• Describe a recent reconciliation where a small classification change led to a materially different savings estimate—what changed and who noticed?
• What emotions surface for your team during that manual crunch—relief, dread, defensiveness, pride, skepticism? Pick all that apply and give an example. Options: Relief, Dread, Frustration, Pride, Skepticism, Other
• If we could automate a specific portion of that two-week process tomorrow, which single piece would create the biggest immediate ROI? Options: Vendor normalization, Accurate category classification, PO/AP reconciliation, Handling p-card spend, Automated dashboards

What Would True Visibility Change for You?

• If you had a spend cube you trusted at >90% accuracy tomorrow, what decisions would you accelerate in the next 90 days?
• Which success signals matter most for you in a pilot (choose up to three)? Options: >90% classification accuracy, Top supplier names normalized, New savings opportunities identified, Reduction in manual prep time, Dashboard self-service adoption, Seamless AP/PO reconciliation
• What acceptance criteria must the pilot meet for you to greenlight production (accuracy thresholds, sample coverage, stakeholder signoffs)?
• How would you measure the impact of improved classification—cost savings, time saved, fewer disputed invoices, audit readiness, or something else? Options: Cost savings identified, Analyst hours saved, Fewer reclassifications, Faster cycle time for reviews, Improved audit traceability, Other
• Who are the primary reviewers for pilot acceptance and what will convince each of them (procurement: accuracy; finance: reconciliation; IT: security)?
• Imagine the pilot reveals undiscovered savings—how do you want those surfaced (ranked opportunities, supplier-by-supplier view, category heatmap)? Options: Ranked savings opportunities, Supplier normalization dashboard, Category spend heatmap, Transaction-level drilldown, Custom reports

Tolerances, Risks, and Deal-Breakers

• What would have to go wrong in a pilot for you to pause the engagement—low accuracy, data leakage, inability to map your taxonomy, or something else? Options: Accuracy below threshold, Data security concern, Inability to access required extracts, Unresolvable supplier normalization, Poor stakeholder adoption, Other
• What minimum accuracy threshold do you need on day one for you to consider the pilot credible? Options: >95%, 90–95%, 85–90%, <85%
• Do you have any regulatory, privacy, or internal controls we must respect during data handling (e.g., PCI, PII rules, on-prem only)? Options: PCI concerns, PII/data residency, On-premises-only extracts, No special requirements, Other
• What access constraints exist—can we get raw transactional extracts, or only summarized exports or masked data? Options: Full raw extracts available, Masked/anonymized extracts only, Summarized exports only, API access only, Varies by business unit
• Who in your org needs to sign off on data sharing and pilot terms (legal, security, procurement, finance)? Options: Legal, Security/InfoSec, Procurement, Finance, IT/Data team, Other
• What SLAs around turnaround, accuracy improvement, or remediation would give you confidence to proceed?

Next Steps to a Fast, Confident Pilot

• If we committed to a four-week proof-of-value, what would you expect to see by week two that would reassure you we're on track? Options: Initial classification output, Supplier normalization preview, Data quality issues report, Mid-pilot checkpoint meeting, Other
• Which exact extracts can you provide for a pilot sample (please select all that apply) and include any format constraints in the next question. Options: AP ledger with invoice lines, PO lines with receipts, P-card transaction exports, GL exports with account codes, Supplier master file, Payment/clearing files
• Are there preferred date ranges, business units, or supplier cohorts you want included in the pilot sample? Options: Recent 3 months, Most recent fiscal quarter, Top 50 suppliers by spend, Tail spend under $50k, Specific business units, Other
• Who will be the day-to-day owner from your side for data delivery, labeling feedback, and signoffs? Options: Procurement ops lead, Category manager, Finance analyst, IT/data engineer, External partner
• What cadence do you prefer for checkpoints during the pilot (weekly demo, twice-weekly standup, ad-hoc)? Options: Weekly demo, Twice-weekly standup, Bi-weekly, Ad-hoc as needed, Single final review
• Are there any internal milestones or meetings (e.g., executive review, audit committee) that the pilot timeline needs to align with?
• Finally, what would make you say 'yes' to running a pilot with us right now—what’s the single most important condition?

Quick Check‑In: Who's in the Room?

• To make the pre‑deployment work smooth, who from your team will be actively involved during the next 4–6 weeks? Options: VP/Head of Procurement, Category Manager(s), IT/Integration Lead, Data Owner / Finance, Legal/Privacy, Other
• Which of those people are empowered to sign technical and data access approvals without escalation? Options: Procurement Lead, IT Lead, Finance Lead, Legal, Requires VP/CPO approval, Unsure
• What’s the preferred communication cadence and channel for rapid decisions during the pilot (daily standup, twice weekly, email, Slack, etc.)? Options: Daily standup, Twice weekly check‑in, Weekly, Ad hoc by email, Slack/MS Teams channel, Other
• Have you run an external classification or spend analytics proof‑of‑value before? If so, what went well and what prevented full adoption?
• What would feel like an obvious sign that we should accelerate the ingestion schedule versus pausing to remediate? Options: >90% accuracy on sample, Supplier normalization >95%, Discovery of new savings, No major PII concerns, Other

What If We Can't Get Full Access?

• If IT/Finance pushback limits our access, what are you most worried we’ll miss during the pilot? Options: Incomplete supplier list, Missing GL/PO linkage, P‑card transactions excluded, Line‑item details absent, Other
• Who technically owns the AP/PO/p‑card data extracts and who typically grants the access (role/title)? Options: AP Manager, ERP Admin, Finance Data Team, Procurement Ops, Third‑party provider, Unsure
• What security or compliance constraints should we plan for before asking for sample extracts (e.g., VPN only, no external storage, tokenized PII)? Options: VPN only, SFTP with IP allowlist, Tokenization required, On‑prem only, Cloud transfer allowed, Other
• If legal requests a data minimization approach, which fields would you be comfortable sharing in a first pass (e.g., supplier name, invoice amount, COA code, line description)? Options: Supplier name, Invoice amount, Invoice date, GL/COA code, Line description, No PII fields
• How long does your IT/security approval process typically take for a third‑party data ingest request? Options: <1 week, 1–2 weeks, 2–4 weeks, 1–2 months, Longer/Unsure

How Messy Is Your Data, Really?

• If you had to pick one word to describe your current AP/PO/p‑card data quality, what would it be and why?
• How many distinct source systems will we need to extract from for the pilot (ERP instances, card processors, AP tools)? Options: 1, 2, 3, 4+, Unsure
• Which of these common issues show up in your data today—pick all that apply? Options: Missing supplier IDs, Inconsistent GL/commodity coding, P‑card line‑level data missing, Suppliers with multiple name variants, Invoices with concatenated line descriptions, Other
• Can you provide a short example (paste or describe) of the kind of line description or vendor string that typically trips up manual classification?
• On a scale of 1–10, how confident is your team in the existing supplier master (1 = fractured, 10 = single source of truth)? Options: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
• Are there internal taxonomies or cost centers we must map to (i.e., non‑standard to our model)? If yes, who owns the mapping decisions? Options: Yes — Procurement owns, Yes — Finance owns, Yes — Shared decision, No — use standard taxonomy, Unsure

If We Start with Only Three Months, Will That Tell the Truth?

• Is a three‑month sample from two business units representative of your overall spend patterns, or do you expect material blind spots? Options: Representative, Some blind spots, Not representative, Unsure
• Which months would best capture typical spend versus seasonality for the categories we're testing? Options: Most recent 3 months, Same quarter last year, Peak season months, Would prefer 6 months, Unsure
• Do you have high‑value one‑off payments or projects in that sample window we should flag and exclude? Options: Yes — capital projects, Yes — one‑time vendor settlements, No, Unsure
• Would you prefer sampling by BU, by supplier, or by category to maximize signal in the pilot? Options: By business unit, By supplier (top spend), By category, Hybrid approach, Open to our recommendation
• How important is matching AP lines back to PO/receipts in this pilot for you to trust results? Options: Critical, Nice to have, Not necessary for pilot, Unsure

Who Will Own the Clean‑Up When We Find Problems?

• When the model flags inconsistent suppliers or unmapped categories, who on your side will be responsible for remediation? Options: Procurement Ops, Finance/AP team, ERP Admin, Shared responsibility, Unsure
• How quickly can your team address remediation tasks (e.g., supplier merge, GL recoding) once we surface them? Options: Within 48 hours, 1 week, 2–4 weeks, Longer/Depends on priority
• What tools/processes do you currently use for supplier master hygiene and change control (e.g., hub, spreadsheet, MDM tool)? Options: ERP master, MDM tool, Shared spreadsheet, Third‑party service, No formal process, Other
• If remediation requires IT/ERP changes, what is the typical lead time and approval path? Options: Days, 1–2 weeks, 2–6 weeks, 1–3 months, Unsure
• Would you like us to provide a prioritized remediation backlog (low/medium/high effort/impact) that your team can act on post‑pilot? Options: Yes — prioritized backlog, Yes — raw list only, No — we’ll handle internally, Unsure

What Would Make You Hesitate to Ingest?

• What are the top three fears or blockers that would make your team pause before allowing data ingestion?
• What minimum classifier performance (accuracy, normalization rate) would you require in the pilot to feel confident moving to production? Options: >95% accuracy, 90–95% accuracy, 85–90% accuracy, Other / undefined
• Are there legal, privacy, or retention policies that would prevent us from storing a copy of the sample outside your environment? Options: Yes — cannot store externally, Yes — only pseudonymized, No restriction, Unsure
• If an unexpected sensitive field appears in the sample, what is your preferred remediation: redact and continue, pause ingestion, or return sample and restart? Options: Redact and continue, Pause ingestion, Return sample and restart, Consult legal first
• What would be an acceptable rollback or containment plan if we discover critical classification errors after a large ingest? Options: Restore previous dataset, Isolate affected segments, Reprocess with corrected mapping, Stop and review with stakeholders, Unsure

Signoff: Whose OK Do We Need and When?

• Who must provide formal signoff before we begin production ingestion (list names/titles and their signoff authority)?
• Do you require a written data processing addendum, SLA, or security questionnaire completed before we access sample extracts? Options: DPA/contract required, SLA required, Security questionnaire only, No additional docs required, Unsure
• What is your desired signoff timeline from first sample delivery to final go/no‑go decision? Options: <2 weeks, 2–4 weeks, 4–6 weeks, Longer/Unsure
• Who should we notify immediately if a data or security incident occurs during the pilot (name/title and preferred contact method)?
• Would an interim checklist of technical, legal, and business signoffs before ingestion help accelerate approvals? If yes, who should own that checklist? Options: Yes — Procurement owner, Yes — IT owner, Yes — Joint owner, No

Small Wins Before We Turn the Key

• If we delivered three quick checks in 72 hours (sample schema validation, supplier duplicate heatmap, and top‑20 spend classification), which would you want first? Options: Schema validation, Supplier duplicate heatmap, Top‑20 spend classification, Other
• Which quick validation metric would give you the most confidence early on—classification accuracy, supplier match rate, or % of spend normalized? Options: Classification accuracy, Supplier match rate, % spend normalized, Other
• What minimal dashboard or deliverable would you consider a successful pilot milestone (e.g., spend cube, supplier cleanup list, savings opportunities)? Options: Spend cube, Supplier cleanup list, Savings opportunities report, All of the above, Other
• How would you prefer we present early findings to stakeholders—one concise executive slide, a working dashboard demo, or a walk‑through workshop? Options: Executive slide, Dashboard demo, Walk‑through workshop, Combination
• What immediate next step should we schedule after confirming access and sample scope (kickoff, data extraction window, legal signoff)? Options: Kickoff meeting, Schedule data extraction, Legal/IT signoff, Pilot roadmap review

Tell Me About Your Last Category Review

• Walk me through the last category review you ran—what data did you load, who was in the room, and how long did the data prep take?
• How often do you run formal category reviews (e.g., quarterly, biannual)? Options: Monthly, Quarterly, Biannually, Annually, Ad hoc
• Which business units and geographies were included in that sample?
• Who owns the final spend baseline and who typically validates it (roles/titles)? Options: VP Procurement/Head Category, Category Managers, Finance lead, IT/Data owner, Analytics team, Other
• On a scale from 1–10, how confident were you in the numbers you presented at the end of that review? Options: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
• What was the single biggest surprise or blind spot the team discovered after the review?

If Your CFO Asked Today, Could You Answer?

• When leadership asks 'How much do we spend with our top 50 suppliers and how much is under contract?', what typically makes that question hard to answer?
• Which of these is the primary blocker for that question in your organization? Options: Inconsistent ERP commodity codes, Unclassified p-card spend, Missing PO/AP linkage, Supplier naming inconsistencies, Lack of contract linkage, Other
• Who on your team feels the most pressure when executive reporting falls short (select up to two)? Options: CPO/Head of Procurement, VP Category Management, Head of Finance, Treasury/Controllership, Procurement Analytics, IT/Data Owner
• How does it feel when you have to tell leadership you can’t answer that core question—frustrated, embarrassed, resigned, or something else? Options: Frustrated, Embarrassed, Resigned, Motivated to fix it, Other
• How long has this visibility gap existed in your reporting? Options: Less than 6 months, 6–12 months, 1–3 years, 3+ years
• If we could reliably answer that executive question in under a week, what would change about how your team spends its time?

Where the Data Hides Its Mess

• How often do you discover new, unexpected data sources or feeds when you begin a review—bank extracts, shadow systems, or local ledgers? Options: Almost always, Often, Occasionally, Rarely
• Which of the following data sources are in scope for most category reviews? Options: AP ledger (ERP), PO system, Purchasing card (p-card), Corporate card programs, T&E, Supplier catalogs, Contracts repository, Other
• What are the common formats and handoffs you receive (CSV exports, APIs, PDFs, ERP reports)? Options: CSV/Excel exports, Database/SQL extracts, API feeds, PDF invoices, Flat files via SFTP, Other
• Tell me about taxonomy failures you see—examples of miscoded spend or categories that always look wrong.
• How often do supplier names split into multiple variants and cause spend to be scattered? Options: Almost every supplier, Many suppliers, Some suppliers, Rarely
• How long does your team typically spend reconciling or cleaning these source variations before analysis can begin? Options: Less than a day, 1–3 days, 1–2 weeks, 2–4 weeks, More than a month

How Much of Your Team’s Week Is Spent Fighting Spreadsheets?

• If you add up everyone involved in data prep for a review, how many person-days does that usually consume? Options: Under 5 person-days, 5–15 person-days, 15–40 person-days, 40+ person-days
• What specific manual tasks take the most time—supplier normalization, line-item classification, matching invoices to POs, or something else? Options: Supplier normalization, Line-item classification, PO/AP matching, Contract tagging, Cleaning duplicates, Other
• How often do manual errors from spreadsheets lead to wrong recommendations or missed savings? Options: Frequently, Sometimes, Rarely, Never
• When manual prep delays a review, what strategic activities get deprioritized as a result? Options: Supplier negotiations, Category strategy sessions, Sourcing events, Contract consolidation, Savings capture, Other
• Share a concrete example of a time manual data work changed an outcome—what decision was delayed or wrong?

If Your Spend Cube Could Tell the Truth

• Imagine your classified spend cube is 95% accurate and supplier names are normalized—what would that enable your team to do in the next 90 days?
• Which of the following measurable success signals matter most for your pilot? Options: >90% classification accuracy, Supplier normalization to single canonical name, New savings opportunities identified, Contract coverage % by spend, Timely production of dashboards
• What acceptance criteria would you require for the pilot to be considered a success?
• How would you quantify the business value you expect—hours saved, % improvement in contract coverage, $ savings identified, or speed to insight? Options: Hours saved, % contract coverage improvement, Identified $ savings, Faster decision time, Other
• If the pilot delivered those signals, who would need to be convinced internally to move to production? Options: CPO/Procurement leadership, Finance, IT/Security, Business Unit Leaders, Legal/Compliance, Other
• Which dashboards or views would you need to see first to feel confident (supplier concentration, savings pipeline, contract coverage)? Options: Supplier concentration, Savings pipeline, Contract coverage by supplier, Category spend trends, PO vs non-PO spend, Other

What Would Make You Skeptical—And How Could We Prove Otherwise?

• What would make you or your stakeholders say 'we tried AI once and it didn’t work'—what past experiences drive that skepticism?
• Which of these concerns matters most when assessing an AI classification partner? Options: Accuracy on first pass, Data security/privacy, Model explainability, Long-term maintenance, Integration effort, Vendor lock-in
• How important is it for you to see a human-in-the-loop labeling workflow during the pilot? Options: Critical, Important, Nice to have, Not necessary
• If initial accuracy is below expectations, how long would your team realistically give for tuning and iterative labeling? Options: 1 week, 2–3 weeks, 1 month, Longer than a month
• What audit or traceability requirements must the solution meet for compliance or internal controls? Options: Full lineage of classifications, Change logs and labeling history, Role-based access, Encryption at rest/in transit, Other
• Who on your side would own stakeholder trust during the pilot (who communicates wins and manages doubts)? Options: VP Procurement, Head of Category, Procurement Analytics, Finance Business Partner, Other

The Proof That Changes Minds

• What single piece of evidence would move your team from curiosity to commitment after the proof-of-value?
• What sample size do you consider convincing for a PoV (months of data, number of transactions, number of business units)? Options: 1 month / one BU, 3 months / one BU, 3 months / two BUs, 6 months / multiple BUs
• Which KPI would you prioritize in the PoV dashboard to demonstrate rapid impact? Options: Classification accuracy, Normalized supplier count, New savings opportunities, Time to baseline, Contract coverage
• What turnaround time do you expect from sample upload to first dashboard (realistic vs ideal)? Options: <1 week (ideal), 1–2 weeks, 2–4 weeks, 4+ weeks
• Who needs access to the PoV dashboards for the pilot to influence a go/no-go decision? Options: CPO/Procurement leadership, Category Managers, Finance leads, IT/Data team, Business Unit stakeholders
• If the PoV exposes a $X opportunity in tail spend, how quickly could your team act on that insight? Options: Immediately, Within 1 month, 1–3 months, Longer than 3 months

Operational Readiness: Who Will Own Production?

• When we move from pilot to production, who on your team will own ongoing data extracts and scheduling? Options: Procurement Analytics, IT/Integration, AP Team, Shared services, Other
• What constraints does your IT/security team typically impose on external platform integrations (e.g., no external hosting, strict IP allowlists)?
• Which of these access patterns is feasible for your team (select all that apply)? Options: SFTP file drops, API integration, Direct DB read-only, Manual CSV uploads, Third-party ETL
• What frequency would you expect for production refreshes (daily, weekly, monthly)? Options: Near real-time, Daily, Weekly, Monthly
• What internal SLAs or accuracy targets would you require before you’d allow this output to feed your strategic reports?
• Who will sign off on data privacy and sharing agreements on your side? Options: Legal, Data Governance, IT Security, Procurement leadership, Other

Money, Metrics, and A Yes/No

• If the pilot proves the value we discussed, what is the internal approval path for committing budget and a contract?
• What is the current budget posture for spend analytics initiatives? Options: Budget allocated and approved, Budget requested, pending approval, Budget needs to be created, No budget currently
• What ROI or payback timeline does leadership expect for analytics projects (months)? Options: <3 months, 3–6 months, 6–12 months, 12+ months
• What commercial model do you prefer for a pilot to production path? Options: Fixed-fee pilot then subscription, Success-fee or savings-share, Annual SaaS license, Phased implementation fees
• Who are the decision-makers that must sign commercial terms and SLAs? Options: Procurement leadership, Finance/Controller, Legal, IT/Security, CPO/CEO
• Are there procurement or compliance constraints (e.g., approved vendor lists, GSA schedules) that would affect contracting? Options: Yes, No, Not sure

Next Steps That Won’t Stall

• What immediate data extract can we request to start a PoV—three months of AP, PO, and p-card from one or two BUs—and who can provide it?
• Which timeline sounds realistic to you for starting a pilot after we receive initial extracts? Options: Start within 1 week, Start within 2 weeks, Start within 1 month, Longer than 1 month
• Who should be the single point of contact on your side to coordinate data, approvals, and stakeholder updates?
• What are your preferred meeting cadences and communication channels for the pilot (select all that apply)? Options: Weekly touchpoints, Biweekly reviews, Email updates, Slack/MS Teams channel, Ad-hoc calls
• What would make you immediately say yes to a pilot proposal (e.g., low/no cost pilot, clear acceptance criteria, short time-to-insight)?
• Is there anything we haven’t asked that would materially affect your ability to proceed?