Opening the Conversation: what's top of mind this week

• What's the single retention question on your desk right now?
• Tell me about the last incoming cohort where first-year retention surprised your team, what happened and who noticed first?
• How long has this retention trend been visible in your dashboards or leadership conversations? Options: 1–2 terms, Less than one term, 2–4 terms, More than a year
• Which leaders are most likely to move budget or staff to address a retention gap—provost, VP enrollment, student affairs, or department chairs? Options: Provost, VP of Enrollment Management, Dean of Students/Student Affairs, Department Chairs, Other
• On a priority scale, where does solving first-to-second-year attrition sit against other institutional priorities this year? Options: Top priority, High priority, Important but not urgent, Low priority

Where student risk actually hides

• Which student segment do you most commonly discover too late to intervene—so we can test that story first? Options: First-generation students, Transfer students, Students with financial holds, Part-time students, Students on academic probation, Other
• Describe the exact reports or exports you rely on today to spot early academic decline, and how often those are refreshed. Options: Term grade rollups (weekly), LMS engagement dashboards (daily), Financial aid status reports (weekly), Advisor notes exports (ad hoc), Other
• How often do LMS engagement signals and SIS grades point in opposite directions for the same student, and what is your fallback when that happens? Options: Frequently, Sometimes, Rarely, Never
• Who on your campus is currently trusted to validate a risk signal before outreach begins, and could that role be delegated in a pilot?
• If you could change one data source so it reported earlier or more reliably, which would it be and why? Options: SIS grades, LMS activity, Financial aid disbursements, Housing/engagement data, Advisor notes, Other
• Which single data gap here would be a deal stopper for a pilot this semester?

Advisor workflows and the human bottlenecks

• If you had to name the advisor workflow failure that most prevents proactive outreach, what would it be?
• Name the screens, tools, or lists advisors open first when preparing outreach, so we can prototype on those surfaces. Options: Advising caseload list, At-risk dashboard, Student transcript view, Appointment scheduler, Email/communication tool, Other
• Walk me through a recent student case where outreach failed, who was involved, and what you wish had been different in the workflow.
• How many students does a typical advisor manage, and how many proactive check-ins per month are realistic today? Options: Under 100 / 10+, 100–199 / 5–10, 200–299 / 1–5, 300+ / less than 1
• What resource or small policy change would immediately free enough advisor time to change outcomes in a single term?
• Who will need to endorse changes to advisor scope and responsibilities for a pilot to proceed? Options: Department Chairs, Director of Advising, Provost Office, Union/HR, Other

Measuring success, the way your board will judge it

• Name the single measurable outcome this academic year that would make your leadership call the program a success. Options: Increase first-to-second-year retention by 1–2 pct, Increase retention by 3–5 pct, Reduce stopouts by X students, Improve on-time graduation rate by a percentage point, Other
• What baseline numbers do you have today for first-to-second-year retention, term-to-term stopouts, and on-time graduation? Please list values.
• How quickly do you need to see measurable improvement to satisfy accreditors or funders, within one term, one year, or longer? Options: Within one term, By end of academic year, Within two academic years, Timeline driven by external review
• If a pilot delivered the promised lift, what internal approvals would be required to scale—budget, cabinet vote, board approval, curriculum change? Options: Budget reallocation, Executive cabinet signoff, Board approval, Academic senate/curriculum approval, Other
• What single validation would make you comfortable signing a multi-year contract after a pilot? Options: Statistically significant retention lift, Advisor time saved metric, Operational readiness evidence, Positive student feedback, Other

Operational readiness, the practical gates we must clear

• Identify one integration or data dependency that would force you to pause implementation if it cannot be solved within the proposed timeline.
• Do your SIS and LMS provide programmatic access today, such as APIs, SFTP feeds, or regular exports, and who owns those endpoints? Options: APIs available and owned internally, APIs available but managed by vendor, SFTP or scheduled exports only, No programmatic access today
• Who owns data permissions and legal approvals for student records, and how long does their approval process typically take?
• Are there institutional compliance or privacy reviews that must clear before receiving student-level data for a pilot? Options: FERPA review required, IRB review required, General data sharing agreement only, No formal review required
• How many dedicated IT or data staff hours per week can you realistically commit to integrations during a pilot phase? Options: Under 10 hours/week, 10–20 hours/week, 20–40 hours/week, 40+ hours/week
• What single infrastructure or resourcing constraint would kill the timeline for a semester-long pilot?

The other options you are weighing

• Who or what are you most likely to turn to if you decide not to change vendors for retention analytics? Options: Keep incumbent vendor, Expand internal analytics team, Use point tools for alerts only, Partner with another vendor, Other
• Which vendors or internal solutions are you actively evaluating right now, and what draws you to each option?
• What would need to be true about your current approach for you to keep it instead of switching to a new vendor? Options: Improved accuracy without new integrations, Lower total cost, Clearer ROI within one year, Better advisor UX, Other
• Has anyone on campus proposed building an internal predictive model or workflow instead of buying? If so, who and what is their timeline? Options: Yes, IT/analytics team, Yes, academic affairs, No internal proposal, Other
• What single advantage would a new vendor need to demonstrate over your incumbent or internal build to win the project?

Decision triggers and next steps

• If a pilot proved the retention lift you expect, what would stop you from moving to a full contract within four weeks?
• What is your target decision date for selecting a partner, and are there external deadlines driving that timing? Options: Within 30 days, 30–60 days, 60–90 days, No fixed deadline
• Who are the final approvers for contracting and budget, and what procurement steps must we anticipate?
• How much budget has your team provisioned for a pilot and potential scale, or will you need to reallocate funds? Options: Pilot funded and budgeted, Pilot needs reallocation, Not yet budgeted, Other
• Would running a scoped, time-boxed pilot this term accelerate or delay your decision timeline? Options: Accelerate, No change, Delay, Unsure

Quick facts and contacts to move forward

• List the primary contact or owner for the pilot from your side, their role, and the best email to reach them.
• Which data owner can provide a representative sample dataset within two weeks, and can we expect anonymized extracts? Options: SIS data owner, LMS admin, Registrar, Financial aid office, Other
• What environments can we use for integration testing, sandbox or production read-only, and who controls access? Options: Sandbox environment available, Production read-only access available, No test environment, Other
• How soon could your team commit to a kickoff once scope and contracts are agreed, within 2 weeks, 4 weeks, or later? Options: Within 2 weeks, Within 4 weeks, 1–2 months, Unsure
• Finally, what single risk should we solve first to make a pilot successful for your campus?

• Solution Experience Session

• Orientation: sign-to-go-live overview
• You confirm the stated current state is accurate and quantify the cost to retention, tuition, and staff effort.
• Provide a representative sample dataset including SIS, LMS, and financial aid fields for a single cohort term.
• You confirm the demonstrated scenario reduces advisor manual triage and shortens time-to-contact for at-risk students.
• Confirm the current state and its cost
• Identify and share contacts for data owners and integration technical leads for SIS and LMS access.
• You confirm required data fields and access paths exist or identify specific gaps that must be closed for a pilot.
• Run a sample risk scoring run on the provided dataset and deliver an accuracy and feature-mapping report before the follow-up session.
• Scenario walkthrough: predictive scoring and early alerts
• You agree on the next evidence the seller must deliver and the buyer's timeline for a pilot decision.
• Advisor workflow proof: caseload routing to outcome
• Confirm the target retention metric and the acceptance criteria for pilot success.
• Schedule a pilot scoping meeting with operational owners to finalize scope and timeline.
• Data and integration readiness check
• Validation: does this map to what you need?
• Next steps and pilot commitments

• Solution Experience Session
• Solution Experience Deck
• Solution Brief — Predictive Student Success

• meeting
• slides
• document

• Integrate SIS student records
• Ingest LMS engagement data
• Import financial-aid and billing data
• Integrate student-services interaction data
• Build historical student-data pipeline
• Deploy predictive risk model
• Calibrate model to institutional outcomes
• Configure real-time risk scoring engine
• Enable early-alert triggers and notifications
• Set up advisor caseload management
• Activate appointment scheduling capability
• Implement intervention tracking workflows
• Deploy outcome analytics and retention reports
• Train advisors and student-services staff
• Pilot rollout and operational support

Integrate SIS student records

• How many SIS instances or campuses must be connected (single campus, multi-campus within one SIS instance, multiple distinct SIS installations)? Options: Single campus / single instance, Multi-campus within one instance, Multiple distinct SIS installations
• Which SIS export formats can you provide for term enrollments and course grades (flat CSV, XML, database dump, API access)? Options: CSV exports, SIS API (REST), XML export, Database dump / SQL
• Who in your registrar or data governance office will approve access to student records and provide service accounts for SIS extracts?
• Are student unique identifiers stable across terms (persistent student ID) or do we need a reconciliation strategy for multiple identifier types? Options: Persistent student ID available, Multiple identifiers require mapping (e.g., username + campus ID), Identifiers need reconciliation rules
• Within what timeframe can you produce an initial SIS data dump for backfill (days/weeks) and a cadence for ongoing incremental exports? Options: Initial dump within 7 days, Initial dump within 2-4 weeks, Initial dump >4 weeks, Ongoing incremental daily / weekly / monthly

Ingest LMS engagement data

• Which LMS event types are available from your LMS (page views, assignment submissions, quiz attempts, discussion posts, last access time)? Options: Page views, Assignment submissions, Quiz attempts, Discussion posts, Last access / session logs, Other
• At what frequency can you provide LMS event streams for ingestion (real-time webhooks, hourly batch, nightly export)? Options: Real-time webhooks, Hourly export, Nightly export, Weekly export
• Which field will you use to link LMS users to SIS student records (institutional student ID, username, email)? Options: Institutional student ID, Username/login, Institutional email, Other mapping
• Who is the LMS administrator contact for generating API credentials and reviewing event schema?
• Are course shell IDs and section numbers consistent across terms or do we need a crosswalk for archived/renumbered courses? Options: IDs consistent across terms, Course ID crosswalk required, Course shells frequently renumbered

Import financial-aid and billing data

• Which financial-aid data elements should be imported (FAFSA status, award amounts, disbursement dates, unmet need, hold flags)? Options: FAFSA/eligibility, Award amounts, Disbursement dates, Outstanding balance, Hold or restriction flags, Other
• Which billing or student accounts export methods are available for your bursar/finance system (API, nightly CSV, scheduled reports)? Options: API, Nightly CSV, Scheduled report extracts, Manual export
• Will you provide a mapping of institutional aid codes and award types, or should we infer semantics from sample files? Options: You will provide mapping, We should infer from sample files, Combination of both
• Who must approve access to financial records given FERPA or privacy constraints and do you require a data processing agreement signature?
• Are financial-aid snapshots versioned by term and date (allowing point-in-time reconstructions) or only current-state exports are available? Options: Versioned snapshots available, Only current-state exports, Partial historical snapshots

Integrate student-services interaction data

• Which student-services systems hold advising notes, case management records, tutoring referrals, or outreach logs that should be integrated? Options: Advising notes system, Case management, Tutoring/referral logs, Outreach/call center logs, Other
• How are interaction records linked to students today (student ID, email, appointment ID) and are those links reliable for automated joins? Options: Linked by student ID, Linked by email, Linked by appointment ID, Links require reconciliation
• Which interaction attributes should be prioritized for predictive features (appointment outcome, no-show flag, advisor risk flag, referral reason)? Options: Appointment outcome, No-show flag, Advisor risk flag, Referral reason, Case severity
• Who will own taxonomy mapping of interaction types to platform workflows and approve canonical event types?
• Are free-text advising notes subject to redaction or do you have a consent policy for ingesting unstructured notes? Options: Notes can be ingested raw, Notes require redaction, Consent required per student

Build historical student-data pipeline

• How many academic years of historical records (term enrollments, grades, financial snapshots, LMS logs) should be backfilled into the pipeline? Options: 1 year, 2-3 years, 4-6 years, Full archival history
• Which SIS and LMS tables or files must be included in the backfill (term enrollments, course grades, assignment-level logs, program changes)?
• Which acceptance criteria will define successful historical ingestion (for example: >=95% term-grade match rate, no schema errors, sample reconciliation report)? Options: >=95% grade match rate, No critical schema errors on import, Signed data quality report
• How many sample records or file size should we request for a pilot backfill to validate parsing and mappings (number of student rows, GB)? Options: 1,000 rows, 10,000 rows, 100,000+ rows, Specify sample size
• Who will review and sign off the historical data quality and reconciliation deliverable (registrar, institutional research, or data governance lead)?

Deploy predictive risk model

• Which risk model type do you prefer for initial deployment (off-the-shelf baseline model, institution-specific trained model, hybrid transfer-learning)? Options: Baseline generic model, Institution-specific trained model, Hybrid transfer-learning model
• Which feature groups should be included in the model (demographics, academic performance, LMS engagement, financial status, advising interactions)? Options: Demographics, Academic performance, LMS engagement, Financial status, Advising interactions
• How many training cohorts (graduation/completion cohorts by matriculation year) are available to build a calibration dataset? Options: 1 cohort, 2-3 cohorts, 4+ cohorts, Not sure / need IR help
• Which model performance metrics will you prioritize during deployment (AUC, precision at top decile, recall for probation students)? Options: AUC, Precision at top decile, Recall for at-risk cohort, Lift over baseline
• Who will own monitoring of model drift and schedule re-training triggers (institutional research, analytics team, vendor operations)?

Calibrate model to institutional outcomes

• Which calibration targets should we align to (first-year retention at term, 2-term persistence, program completion rate)? Options: First-term retention, First-year retention, 2-term persistence, Program completion
• Which sample size or minimum event counts do you require for calibration validation (minimum N per cohort)? Options: N>500 per cohort, N 100-500, N<100, No strict minimum
• How will you evaluate acceptable calibration performance (e.g., predicted vs observed retention within X percentage points by cohort)? Options: Within 2 percentage points, Within 5 percentage points, Custom threshold
• Who will provide institutional definitions for retention and stop-out used in calibration (institutional research or registrar definitions)?
• Are there subgroup fairness constraints to enforce during calibration (by program, race/ethnicity, Pell-eligible status)? Options: Yes, enforce subgroup parity, No subgroup constraints, Limited to select cohorts

Configure real-time risk scoring engine

• Which event sources should trigger real-time scoring (LMS submission failure, sudden drop in access, financial hold posted)? Options: LMS events, Grade posting, Financial holds, Advising notes
• Which API endpoints or webhook URLs will you expose for streaming events to the scoring engine?
• At what latency threshold must the scoring engine return updated risk (seconds, minutes) for advisor-facing alerts? Options: <5 seconds, <1 minute, <15 minutes, Near real-time (within an hour)
• Which risk score scale do you prefer for advisor UX (0-100 score, risk bands Low/Moderate/High, percentile rank)? Options: 0-100 numeric score, Low/Moderate/High bands, Percentile rank, Custom scale
• Who will approve production credentials and sign off moving the scoring engine to the live environment?

Enable early-alert triggers and notifications

• Which alert triggers should be configurable (sudden drop in LMS activity, missed grades, financial hold, advisor referral)? Options: LMS activity drop, Missed or low grades, Financial hold, Advisor referral, Other
• Which notification channels should be used for alerts (in-platform inbox, email to advisor, SMS for student, SIS task assignment)? Options: In-platform inbox, Email to advisor, SMS to student, SIS task assignment
• Which alert severity levels and escalation paths do you require for unresolved high-risk cases? Options: Low/Medium/High with advisor escalation, High triggers escalation to director, Custom escalation rules
• Which alert throttling or suppression rules are needed to avoid alert fatigue (one alert per student per day, aggregate by issue)? Options: One alert per student per day, Aggregate similar alerts, No throttling
• Who will own alert template review and sign-off for student-facing communications to ensure FERPA-compliant wording?

Set up advisor caseload management

• Which advisor roles and teams should be modeled (success coaches, faculty mentors, program advisors, centralized advising)? Options: Success coaches, Faculty mentors, Program advisors, Centralized advising
• Which caseload assignment rules do you use now that should be mirrored (student program, risk band, geography, last name range)? Options: By program, By risk band, By geography, By last name range, Manual assignment
• Which caseload size targets or maximums should be enforced per advisor (max students per advisor)? Options: <100, 100-300, >300, Custom target
• Which advisor queue sorting and filtering fields are required (risk score, next appointment date, probation status)? Options: Risk score, Next appointment date, Probation status, GPA
• Who will administer role and permission changes for advisor teams during rollout?

Activate appointment scheduling capability

• Which calendar systems must integrate for advisor availability (institutional calendar system, department calendars, group schedules)? Options: Institutional calendar system, Department calendars, Group schedules, Stand-alone booking
• Which appointment types should be configurable (drop-in, 15/30/60 minute, virtual meeting with link, phone)? Options: Drop-in, 15 minute, 30 minute, 60 minute, Virtual meeting
• Which student-visible scheduling constraints should be enforced (lead time, buffer time between appointments, max weekly bookings)? Options: Lead time required, Buffer time enforced, Max weekly bookings, No constraints
• Which reminder channels and timing do you want for appointments (email 24 hours prior, SMS 2 hours prior)? Options: Email 24 hours prior, SMS 2 hours prior, In-platform reminder
• Who will provide sample advisor schedules and recurring availability templates for configuration?

Implement intervention tracking workflows

• Which intervention types must be tracked (academic coaching, financial counseling, tutoring referral, probation plan)? Options: Academic coaching, Financial counseling, Tutoring referral, Probation plan, Other
• Which workflow states are required for interventions (opened, in progress, resolved, escalated) and any SLA targets? Options: Opened/In progress/Resolved/Escalated, Custom state machine, SLA targets required
• Which data fields must be captured for each intervention record (action taken, outcome, advisor notes, follow-up date)? Options: Action taken, Outcome, Advisor notes, Follow-up date, Student consent
• Which reporting requirements exist for intervention effectiveness (disaggregation by program, retention delta after X weeks)? Options: Retention delta after 12 weeks, Disaggregation by program, Custom reporting
• Who will own updates to intervention taxonomy and approve canned action templates for advisors?

• Subscription Agreement (Order Form)
• Master Services Agreement (MSA)
• Statement of Work (SOW)
• Data Processing Agreement (DPA) & FERPA Addendum
• Data Sharing Authorization
• Success Acceptance Criteria Addendum
• Operational Responsibilities & RACI
• Service Level Agreement (SLA)
• Change Order Agreement
• Termination and Transition Plan
• Invoice & Payment Schedule

• Go-live Health Check (weeks 1-4)
• First Measurement Review (weeks 4-10)
• Acceptance Gate Review (around day 90)
• Quarterly Outcomes and Continuous Improvement Review

• Add the top three approved enhancement requests to the delivery backlog with target delivery quarters.
• Adjust alert thresholds or routing rules and document expected impact on advisor workload.
• Schedule a focused advisor workflow training session and circulate attendance list.
• Restate acceptance criteria from Outcome Discovery
• Produce a single documented acceptance decision against the Outcome Discovery targets, with evidence attached.
• For unmet criteria, agree a remediation plan with clear owners, dates, and success metrics.
• Record the named buyer signatory or authorized approver for the acceptance decision.
• Publish the acceptance report with data annex and the recorded acceptance decision.
• If applicable, open remediation tickets with milestones and target completion dates for each failed criterion.
• Schedule the first Quarterly Outcomes Review to begin realization tracking from the acceptance date.
• Review retention and persistence metrics
• Confirm whether quarterly first-year retention rate and term-to-term persistence rate are meeting Outcome Discovery targets or require intervention.
• Identify which interventions produced measurable retention lift and agree which to scale or retire.
• Prioritize and time-box the enhancement and remediation backlog for the next quarter.
• Publish the quarterly outcomes dashboard and a short summary of intervention performance.
• Schedule the next model recalibration and data quality audit for the agreed window.
• Re-confirm success criteria and owners
• All production integrations required by Outcome Discovery are confirmed healthy or have remedial owners and dates.
• A baseline adoption snapshot is recorded, including number of advisors active and alerts fired in the first 14 days.
• A prioritized, time-boxed remediation plan for any blockers is agreed and scheduled.
• Publish health-check log files and integration error summary for async review.
• Deliver sample dataset extracts used by the model for validation within 3 business days.
• Open tracked tickets for each blocker with target resolution dates and owners.
• Present first measurement vs Outcome Discovery targets
• Determine whether first-year retention rate and percent of at-risk students contacted within 7 days are tracking toward Outcome Discovery targets.
• Document root causes for any shortfalls and a concrete set of corrective actions with delivery dates.
• Confirm timeline to the acceptance gate meeting and required evidence to demonstrate fixes.
• Run a model recalibration using the agreed local training window and publish performance metrics.
• Present outcome data against each criterion
• Intervention effectiveness analysis
• Integration and data validation
• Model and alert performance review
• Document pass/fail per criterion
• Early adoption and usage signals
• Open issues and enhancement backlog
• Operational adherence and workflow audit
• Operational data health and cadence
• Formal acceptance decision and signatory capture
• Open issues and blockers
• Root-cause diagnosis for gaps
• Agree targeted corrective actions
• Remediation plan for any failed criteria
• Agree actions for next quarter
• Agree immediate remediation actions