Start with the recent contact surge

• When did you most recently experience a sharp increase in contact volume and what triggered it?
• Which three contact drivers account for the majority of live-agent volume today? Options: Billing and charges, Account access and authentication, Provisioning and activation, Technical troubleshooting, Plan or product changes, Returns and refunds, Other
• How many customer contacts does your service organization handle in a typical month? Options: Under 100k, 100k to 500k, 500k to 2M, Over 2M
• Tell me your current end-to-end containment rate across web, mobile, and messaging channels, and whether that rate is steady, improving, or declining Options: Under 10%, 10% to 25%, 26% to 40%, Over 40%, Rate unknown / not tracked consistently
• If you could pick only one high-volume intent to fix first, which would it be and why?
• Which channel drives the most handoffs to live agents today? Options: IVR, Website chat, Mobile app chat, Email/ticketing, SMS/messaging, Social media

Where the current systems steal time and trust

• If a single backend failure is causing the most repeat calls, which failure would it be and what happens when it occurs?
• Who on your ops or engineering team owns the systems that most often cause customer escalations?
• How long do customers typically wait before they abandon self-service and escalate to a call? Options: Under 2 minutes, 2 to 5 minutes, 5 to 10 minutes, Over 10 minutes, Varies widely by intent
• Describe a recent example where automated self-service gave the wrong outcome and the downstream cost to support or churn
• Which single operational constraint, if unresolved in the next 6 weeks, would force you to pause or reject a pilot? Options: No API access to billing/provisioning, No sandbox environment available, Insufficient data for intent modeling, Security or compliance block, No local owner to run the pilot

Who needs to be in the room to make this move

• Who on your leadership team must sign off for the pilot to expand into production if targets are met?
• Which functions will actively participate during a pilot (choose all that apply)? Options: Digital CX, Contact center ops, Engineering / API owners, Security / compliance, Legal, Product management, Data science / analytics
• How quickly can procurement approve a standard vendor pilot agreement once the commercial terms are agreed? Options: Under 2 weeks, 2 to 4 weeks, 1 to 2 months, Longer than 2 months, Procurement timeline unknown
• If the pilot proves the containment gains you want in 60 days, who would be enabled to commit the budget to scale immediately?
• Which stakeholder objection typically kills projects like this before they start? Options: Security concerns, Integration cost, Ongoing support headcount, Fear of brand impact, Procurement rules

How customers feel when self-service fails

• Which type of automated mistake would cause material brand or regulatory harm for your business? Options: Incorrect billing adjustments, Unauthorized account changes, Failure to disclose fees, Misrouting of sensitive PII, Misleading troubleshooting advice
• How often do you receive formal complaints tied to self-service interactions versus live-agent interactions? Options: Mostly self-service, Mostly live agent, About even, Not tracked separately
• Describe the last time a bot or automated flow damaged customer trust and what remediation was required
• If a pilot produced even one high-severity error in the first month, would you pause the pilot pending review, continue while debugging, or stop entirely? Options: Pause and review, Continue while fixing, Stop entirely, Depends on error type
• Who on your customer experience or legal team would need to approve conversation templates or transactional language before go-live?

Pilot targets that make leadership sign

• Which single metric would convince leadership to move from pilot to production the week after success? Options: Containment rate uplift, Reduction in live-agent volume, Improved customer effort score, Cost per contact reduction, Improved average handle time for escalations
• What target containment uplift would you require for the pilot to be considered successful over a 60 to 90 day window? Options: 10% uplift, 20% uplift, 30% uplift, 40%+ uplift, Open to discussion
• Over the pilot period, which primary acceptance criteria will you use, and who will sign the acceptance? Options: Containment rate, Customer effort score (CES), Live-agent volume reduction, Error rate below threshold, Business owner signs
• How large a sample of interactions do you need to feel statistically confident in the pilot results? Options: Under 5k interactions, 5k to 25k, 25k to 100k, Over 100k
• If the pilot hits targets but requires additional backend work to scale, would you approve a phased rollout or require full integration before expansion? Options: Phased rollout, Require full integration, Depends on risk

The other paths on the table

• Which alternatives are you actively evaluating right now? Options: Incumbent vendor upgrade, Build internal solution, Point chatbot vendor, IVR redesign, Consultancy-led rework, Do nothing
• If you were to stay with your current approach, what evidence would need to appear to justify doing nothing for another 12 months?
• Has anyone on your team proposed an internal build to solve the top contact driver instead of engaging an external platform? Options: Yes, greenlighted, Yes, proposed but not greenlighted, No internal proposal, Not sure
• Which vendor or internal option would represent the lowest risk to your operations if it failed during a pilot? Options: Incumbent vendor, Internal team, New external vendor, Consultancy
• Is there an existing contractual or political relationship that would block running a pilot with a new external vendor? Options: Yes, contractual, Yes, political / internal preference, No blocking relationship, Not sure

Integration gates and who controls them

• Which integration dependency would derail your timeline if not available within four weeks? Options: Billing API access, Provisioning API access, Authentication/SSO access, Customer data feed, Sandbox environment
• Are your critical integration APIs documented and reachable from a third-party sandbox today? Options: Fully documented and sandboxed, Partially documented or no sandbox, Not documented or restricted, Unknown
• Who owns the API keys, credentials, or firewall rules that must be changed for a pilot and how quickly do they act?
• How many full time technical resources can you commit to integration and testing during a 60 day pilot? Options: None, 1-2 engineers, 3-5 engineers, More than 5
• Are there compliance or legal approvals that routinely add time to new integrations, and how long do they typically take? Options: No approvals needed, Under 2 weeks, 2 to 6 weeks, Longer than 6 weeks

Data, training, and keeping accuracy high

• If your products or policies change every quarter, what process ensures self-service intents and transaction rules stay accurate?
• Who is the owner of your canonical knowledge base or FAQ content and how often is it updated? Options: Product team, Support ops, Knowledge management, No single owner
• Do you have labeled interaction data available for intent modeling for the top three drivers? Options: Yes, well labeled, Partially labeled, Unlabeled raw logs, No usable data
• Which tool hosts your canonical customer data for use in personalized transactions? Options: Your primary CRM, Billing system, Order management system, Data warehouse, Multiple systems
• If there is no committed owner for ongoing training after the pilot, can you create one within the pilot timeline? Options: Yes, No, Need to discuss

If we hit the target, how fast can you move

• If the pilot meets your acceptance criteria in 60 days, what is the shortest realistic timeline to expand to a limited production rollout? Options: Under 30 days, 30 to 60 days, 60 to 120 days, Longer than 120 days
• Who will be the single point of contact for day to day pilot coordination and who will be the executive sponsor for decisions?
• What budget process or fiscal constraint could prevent scaling even if the pilot is successful? Options: Quarterly budgeting cycle, Annual budget lock, Requires capital approval, No budget constraint
• Which logistical dependency, if unresolved, would stop a production rollout after a successful pilot? Options: Long lead time for credentials, Unresolved compliance review, Lack of operations playbook, No staff for 24/7 support
• Are you ready to begin a time-boxed pilot within the next 30 to 60 days if commercial and technical terms align? Options: Yes, ready now, Yes, ready in 30 days, Ready in 60+ days, Not ready yet

• Solution Experience Session

• Confirm the current state and its cost to your team
• You confirm the demonstrated workflow eliminates the FAQ and IVR dead-ends that are forcing callers to escalate.
• Provide the top three contact drivers, 5-10 representative transcripts, and a high-level call-cost estimate to validate containment economics.
• You agree the scenario provides sufficient evidence to estimate containment rate and effort reduction for the pilot scope.
• Walk an end-to-end buyer scenario that proves containment
• Run the demonstrated billing transaction in a seller sandbox against a buyer-provided sample case and deliver the execution log plus an initial containment estimate before the pilot planning session.
• Show how backend actions and safety rules work
• You identify any remaining technical or governance gaps that must be closed before a time-boxed pilot.
• Confirm pilot success thresholds for containment rate, customer effort score change, and acceptable live-agent volume reduction, and commit to a target decision date for pilot acceptance.
• Surface measurable outcomes from this scenario
• Validation and decision mapping

• Solution Experience Session
• Solution Experience Deck
• Solution Brief

• meeting
• slides
• document

• Deploy Conversational Virtual Assistant
• Integrate Backend Transaction APIs
• Implement Guided Troubleshooting Flows
• Migrate FAQ Content to Knowledge Base
• Train NLU Model on Top Contact Intents
• Configure Channel Deployment (Web, Mobile, Messaging)
• Enable Secure Account Linking and Validation
• Configure Escalation with Contextual Transfer to Agents
• Configure Business Rules and Transaction Authorization
• Deploy Pilot Metrics Dashboard (containment, CES, volume)
• Establish Ongoing Model Retraining Pipeline
• Train Contact Center Agents on Handoff Processes

Deploy Conversational Virtual Assistant

• Which high-volume contact drivers (for example billing inquiries during a bill-cycle migration, payment failures, or plan-change requests) should the assistant handle in the pilot? Options: Billing inquiries, Payment failures, SIM provisioning/activation, Service outage/status checks, Plan upgrades or downgrades, Other
• How many concurrent assistant sessions must you support during peak billing-window traffic to reflect your worst-case contact surge? Options: Less than 100, 100-500, 500-2,000, 2,000+
• What channel UX elements must appear on web and mobile during billing and outage flows (for example proactive chat invite, pre-filled account context, inline CES prompt)?
• Will you require the assistant to execute live transactions inline (apply a billing credit, perform a SIM swap, change a plan) without routing to an agent? Options: Yes, No
• Describe any brand voice or regulatory wording constraints the assistant must use when discussing billing adjustments or payment disputes.
• Which of your teams will own conversational content updates and approvals after go-live (examples: Digital CX editorial, Contact Center ops, Product)? Options: Digital CX / Self-service, Contact center operations, Product or Billing team, Platform/Engineering, Other

Integrate Backend Transaction APIs

• Which backend transaction types must be callable from the assistant in the pilot (examples: apply billing credit, execute SIM swap, change customer plan)? Options: Apply billing credit, Execute SIM swap / provisioning, Change plan or bundle, Payment retry or settlement, Update billing contact details, Other
• Which integration endpoints expose these transactions (for example billing API base path, provisioning API path, CRM write API) and what authentication surface do they present?
• What acceptance criteria will validate API integration readiness for the pilot (for example transaction success >= 99%, API 5xx < 1%, end-to-end commit within 5 seconds)?
• Which authentication methods do your endpoints require (for example API key, OAuth 2.0 client credentials, mutual TLS) and do you support scoped tokens for per-transaction authorization? Options: API key, OAuth 2.0 (client credentials), OAuth 2.0 (authorization code), Mutual TLS, Other / custom
• Are there rate limits, throttles, or daily quotas on these endpoints we must design for during the pilot? Options: Yes, No
• Which engineer or integration owner on your side will accept test traffic and provide sandbox credentials for each endpoint?

Implement Guided Troubleshooting Flows

• Which top technical intents should guided flows address in the pilot (examples: no-data, slow throughput, device provisioning failure)? Options: No data / no connectivity, Slow speed / throughput, Provisioning or activation failure, Device configuration issues, Intermittent service drops, Other
• What diagnostic telemetry from your network and OSS/BSS can be surfaced in flows (for example error codes, signal strength, last provisioning timestamp, last billing event)?
• How many decision nodes per troubleshooting flow are acceptable before prompting a transfer to an agent or suggesting a manual fix? Options: 2-4, 5-8, 9+
• Will you permit remote diagnostic or corrective actions via backend APIs during troubleshooting (for example resend provisioning command, reset device session)? Options: Yes, No
• What is the maximum acceptable time-to-resolution per guided flow during pilot (target measured in minutes) so CES targets are preserved? Options: Under 3 minutes, 3-7 minutes, 7-15 minutes, 15+ minutes
• Which of your network or field ops SMEs will validate the technical accuracy of guided troubleshooting scripts before pilot launch?

Migrate FAQ Content to Knowledge Base

• How many FAQ articles covering your top three contact drivers need migration into the knowledge base for pilot? Options: Less than 50, 50-200, 200-1,000, 1,000+
• What format are current FAQ assets stored in (for example CMS pages, Google Docs, PDF handbooks, HTML pages)? Options: CMS, Google Docs / Sheets, PDF / Word documents, Static HTML, Other
• Do any FAQ articles require redaction or removal of personally identifiable information before migration (for example sample invoices containing account numbers)? Options: Yes, No
• What accuracy threshold for answer retrieval do you require for pilot (for example correct top result match rate for billing intents)? Options: 70%, 80%, 90%, Custom
• Which team will own editorial sign-off of migrated content (for example billing ops, legal, CX) and will they approve pre-production samples?
• Do you want migration to include structured metadata such as intent tags, product codes, and billing-cycle references to improve retrieval accuracy? Options: Yes, No

Train NLU Model on Top Contact Intents

• Which labeled intent set should be used for initial training (examples: billing_inquiry, payment_failure, plan_change, technical_troubleshoot)? Options: Billing inquiry, Payment failure, Plan change, Technical troubleshooting, Account access, Other
• How many historical annotated utterances per intent are available from your call transcripts or chat logs for training? Options: Less than 500, 500-2,000, 2,000-10,000, 10,000+
• What minimum intent recognition accuracy (for example F1 or top-1 accuracy) will you accept for pilot promotion? Options: 70%, 80%, 90%, Custom
• Are live-agent call transcripts available and cleared for use after PII redaction to seed intent labeling? Options: Yes, No
• Which of your teams will take responsibility for ongoing labeling and correcting NLU false positives during pilot? Options: Digital CX / Self-service, Contact center operations, Data science / ML, Shared responsibility, Other
• Do you require the model to detect sensitive intents (for example payment disputes, suspected fraud) and force escalation to a supervised flow? Options: Yes, No

Configure Channel Deployment (Web, Mobile, Messaging)

• Which channels should be active in the pilot (for example web chat widget on billing pages, in-app chat, SMS messaging, over-the-top messaging)? Options: Web chat widget, In-app mobile chat, SMS, OTT messaging (generic), Other
• Are there SDK or framework constraints for your web or mobile apps we must support (for example React, WebView, minimum iOS/Android versions)?
• Will you require single sign-on so conversations pre-fill account context on web and mobile channels? Options: Yes, No
• What branding, CSS, or accessibility constraints exist for the chat widget shown on billing and account pages?
• Which messaging carrier or aggregator categories will handle SMS or RCS traffic (for example carrier short code, aggregator API, telco-managed SMS gateway)?
• What peak message throughput per channel must be supported to match your busiest billing cycle events? Options: Under 500 per hour, 500-2,000 per hour, 2,000-10,000 per hour, 10,000+ per hour

Enable Secure Account Linking and Validation

• Which account identifiers will be acceptable for automated linking in self-service flows (for example account number, MSISDN / phone number, national ID)? Options: Account number, Phone number (MSISDN), National ID, Email address, Other
• Which verification methods will you permit during linking (for example one-time PIN via SMS, knowledge-based questions, OAuth identity provider)? Options: One-time PIN (SMS), Knowledge-based questions, OAuth identity provider, Biometric / device-based, Other
• What acceptance criteria will validate secure account linking for pilot (for example false-acceptance rate < 0.1%, OTP delivery >= 98%)?
• Do your security or privacy controls require tokenization or forbidding storage of full account numbers to remain PCI or GDPR compliant? Options: Yes, No
• Which role on your support side will handle identity-exception cases that fail automated linking?
• Are there customer segments that must be excluded from automated transactions (for example corporate accounts, flagged high-risk accounts)? Options: Yes, No

Configure Escalation with Contextual Transfer to Agents

• Which CRM or contact center integration surface will receive contextual transfers (for example ACD integration, API-based case create, or CRM case open endpoint)?
• What specific customer context fields must be passed on transfer (for example account_id, session transcript, last attempted transaction and error code)? Options: Account ID, Session transcript, Last attempted transaction, Error codes / debug logs, Other
• What maximum wait-to-agent time will trigger the assistant to offer a callback in billing flows? Options: Under 30 seconds, 30-120 seconds, 120-300 seconds, Over 300 seconds
• Describe the skills routing rules you need for escalations coming from billing intents versus technical intents.
• Will you require a soft handoff that includes full transcript and suggested resolution notes, or a hard handoff that only opens a ticket without transcript? Options: Soft handoff (transcript + notes), Hard handoff (ticket only), Conditional by intent
• Which of your teams will own agent enablement materials and Q&A for accepting contextual transfers?

Configure Business Rules and Transaction Authorization

• Which transaction authorization thresholds must be enforced in pilot (for example credits over $50 require supervisor approval, plan downgrades restricted for active promotions)?
• What audit trail and approval metadata must be captured for high-risk transactions (for example approver id, timestamp, justification)?
• Which authorization methods are acceptable for automated transactions (for example system-only flag, multi-factor auth, agent override with justification)? Options: System automated approval, Multi-factor authentication, Agent override with justification, Supervisor approval workflow, Other
• What latency threshold for authorization checks do you need to maintain target customer effort score for self-service flows? Options: Under 1 second, 1-3 seconds, 3-7 seconds, Over 7 seconds
• Are there regulatory constraints specific to telecom or financial services we must encode (for example number portability windows, bill-cycle lock periods, financial dispute holds)? Options: Yes, No
• Which role or committee on your side will approve business-rule exceptions during the pilot?

Deploy Pilot Metrics Dashboard (containment, CES, volume)

• Which KPIs must appear on the pilot dashboard to prove value (examples: containment rate for billing intents, customer effort score changes, live-agent volume delta)? Options: Containment rate, Customer Effort Score (CES), Live-agent volume, Average handle time, Transfer rate, Other
• What data sources will feed the dashboard during pilot (for example chat logs, ACD / contact center data, billing system transaction records)?
• What reporting cadence and roll-up levels do you require for pilot monitoring (for example real-time, hourly, daily; by org, queue, or intent)? Options: Real-time, Hourly, Daily, Weekly
• What acceptance criteria will confirm pilot success on those dashboard metrics (for example +30% containment on targeted intents, CES reduction >= 0.5 points, 25% reduction in live-agent volume for tested drivers)?
• Which person or role will own the dashboard and receive automated alerts for KPI breaches during pilot?
• Do you require anonymized customer-level event retention for analytics and model retraining and if so what retention window do you permit? Options: Yes, No

• decision_readiness
• success_criteria
• stakeholders
• current_state
• gaps
• desired_state
• current_state
• desired_state
• decision_readiness
• stakeholders
• success_criteria
• gaps
• stakeholders
• gaps
• decision_readiness
• current_state
• desired_state
• success_criteria
• decision_readiness
• decision_readiness
• decision_readiness
• decision_readiness

• Master Services Agreement (MSA)
• Order Form / Subscription Agreement
• Statement of Work (SOW)
• Data Processing Agreement (DPA)
• Regulatory Compliance Addendum
• Data Sharing Authorization & Connectivity Statement
• Success Acceptance & Pilot Sign-Off
• Deployment Readiness Acknowledgment
• Change Order Agreement

• Go-live Health Check
• First Measurement Review
• Pilot Acceptance Gate
• Operationalization and Incumbent Wind-down Check
• Quarterly Success Review

• Publish the production runbook and escalation matrix including alert thresholds and verification steps.
• Open remediation tickets for any failed acceptance criteria with target resolution dates and verification steps.
• Distribute the final evidence package to the operational teams for follow-through and audit.
• Sustained outcomes review
• Containment rate and customer effort score validated as sustained, or a clear remediation plan exists where they are not.
• Operational runbooks, escalation paths, and monitoring are documented and in use to preserve containment gains.
• Incumbent system decommission status confirmed, with data migration or archive actions logged and no ongoing fallbacks permitted.
• Re-confirm success criteria and owners
• Complete archival or migration of legacy data and confirm retention policies and read-only access if retained.
• Prioritize the enhancement backlog and publish a delivery schedule for the highest-impact requests.
• Rolling metrics review
• Confirm the solution continues to meet or is on track to meet containment rate and live-agent volume targets recorded in Pilot Evaluation.
• Ensure all high-severity incidents have an active remediation plan and a target close date.
• Maintain a prioritized enhancement backlog with target delivery quarters agreed for the top items.
• Update the enhancement backlog with business impact scoring and target delivery quarters.
• Close or reclassify stale operational tickets and publish the status for the next review.
• Publish the quarter's operational priorities and schedule the next metrics checkpoint.
• Deployment validated against the configuration details in the Deployment stage and critical integration endpoints confirmed live.
• All high-priority blockers captured with target resolution dates and accountable roles recorded.
• Analytics tracking for containment rate, customer effort score, and live-agent volume enabled and validated for reporting.
• Produce a post-deployment issues log with target resolution dates for each item.
• Enable and validate analytics events for containment rate, customer effort score, and live-agent volume in the reporting system.
• Publish an updated runbook describing escalation paths and verification steps for the next check-in.
• Present outcome data vs targets
• Determine whether containment rate and live-agent volume are trending to meet the numeric acceptance criteria recorded in Pilot Evaluation.
• Document the root causes for any shortfalls and the corrective action plan with target completion dates.
• Confirm the evidence package and timeline required for the Pilot Evaluation acceptance gate.
• Deliver a gap analysis report linking data shortfalls to suspected causes and estimated impact of proposed fixes.
• Schedule and log the implementation of corrective actions with target completion dates and verification steps.
• Prepare the acceptance evidence package that maps each Pilot Evaluation criterion to the supporting dataset and logs.
• Read back acceptance criteria and numeric targets
• Produce a documented acceptance decision for the pilot that records pass or fail per Pilot Evaluation criterion and captures a named signatory.
• For any failed criteria, have a remediation plan with target dates and verification steps that closes the acceptance loop.
• Ensure the acceptance record and evidence package are stored in the shared workspace for auditability.
• Publish the formal acceptance record including pass/fail outcomes and the named signatory for the decision.
• Deployment and configuration validation
• Operational handoffs and process changes
• Open issues and incident review
• Present outcome data and evidence
• Root cause diagnosis for gaps
• Corrective actions and timeline
• Document pass or fail and capture signatory decision
• Enhancement request prioritization
• Early adoption and usage signals
• Incumbent system wind-down verification
• Blockers and open issues triage
• Enhancement request and issue backlog review
• Next quarter operational priorities
• Confirm readiness for acceptance gate
• Agree remediation plan for any failed criteria
• Immediate remediation actions