Referral Management RFP Template: 40 Questions and an Evidence Workbook
A reusable procurement workbook for asking precise questions, recording exceptions, and connecting vendor answers to a decision you can explain.
Start with a scope page vendors can price
Write the scope before the questions. A useful first paragraph says which team is buying, what work enters its queue, where the work must end, and which systems record the result. “Improve referral management” is too broad to estimate.
Specify inbound intake, outbound tracking, or both. List sites and EHR instances separately. An EHR brand name does not describe the number of connections, local configurations, or ownership arrangements.
Attach monthly unique referral counts, document counts, channel mix, active users, expected growth, and the source of each estimate. Mark unmeasured figures as estimates. Ask vendors to state which assumptions would change their proposal.
Write the boundaries explicitly: “The first release covers referral receipt, document association, administrative exception assignment, and status updates. Clinical decisions remain with designated staff under our approved policies.” Add other workflows only when their owners have agreed to participate.
For a broader capability map, use the referral management software buyer's guide. If the open question is visibility versus execution, settle it with the management-versus-tracking comparison before issuing the RFP.
Copy this response workbook into your RFP
The workbook and question bank are original proposed procurement tools. Adapt them to your organization; they are not a certification checklist or a substitute for your procurement team's review.
| Response field | Required vendor answer | Why the evaluator needs it |
|---|---|---|
| Question ID | Repeat the numbered requirement | Keeps clarifications attached to the right claim |
| Availability | Available now, configuration, custom work, roadmap, or unavailable | Prevents future capability from appearing delivered |
| Workflow description | Trigger, action, output, exception | Shows what the proposed capability does |
| Evidence | Document, demonstration, test result, or relevant reference | Separates an assertion from inspectable support |
| Dependencies | Access, system version, third party, buyer work | Makes conditional answers visible |
| Ownership | Vendor role and buyer role | Identifies work retained internally |
| Commercial effect | Included, additional charge, or not quoted | Connects capability to scope and cost |
| Open issue | Missing fact and promised response date | Creates a follow-up record |
Require one response row per question. Permit an honest “not available” answer. It may support a narrower, workable scope. A vague “yes” does not.
Keep privacy, security, and contractual review with the responsible specialists. NIST describes its AI Risk Management Framework as voluntary guidance for AI evaluation and use. It does not certify a vendor. NIST AI RMF
The 40-question referral management RFP
A. Intake and document handling
- Which intake channels are included, and who performs any remaining manual collection?
- How does a received item retain its original identifier, timestamp, and source document?
- How are incomplete packets and uncertain extracted fields assigned for review?
- How are repeated transmissions distinguished from an additional document or a new request?
- Show how staff correct an extraction error while preserving the original source.
B. EHR connections and record updates
- Which connection methods are proposed for each EHR instance in our inventory?
- Which records and fields can be read, created, and updated in the quoted scope?
- What evidence confirms that an attempted write reached the intended destination?
- What happens if an update succeeds but the acknowledgment is delayed or lost?
- Who investigates interface failures, maintains mappings, and pays third-party connection charges?
C. Scheduling and communication
- Which administrative scheduling rules can be configured, and who approves changes?
- What happens when availability changes between displaying a slot and confirming a booking?
- How are outreach eligibility and approved contact preferences checked before an attempt?
- How do manual callbacks, bookings, and patient requests stop pending automated attempts?
- What context accompanies a handoff, and how is receipt by the assigned staff queue confirmed?
D. Administrative dependencies and exceptions
- How does the platform represent a dependency owned by another team?
- Can an item remain visible while waiting for that team's decision without being marked complete?
- What evidence and authorizer are recorded when staff release an exception?
- How are changed instructions linked to the earlier version of a referral?
- Demonstrate reassignment when an owner is unavailable, including the next-action deadline.
E. Reporting and exports
- Define receipt, acceptance, scheduling, visit confirmation, and documentation receipt separately.
- Which timestamps support those definitions, and which are inferred or unavailable?
- Can we reconcile a dashboard total to the underlying individual records?
- What happens to historical reports when duplicate records are linked or a status is corrected?
- Which identifiers, documents, events, configurations, and reports can we export?
F. Review evidence and operational access
- Which current security and privacy materials can our designated reviewers inspect?
- Which outside service providers participate in the proposed workflow, and what do they handle?
- Demonstrate access for a site coordinator, central supervisor, and support user.
- Show the history available when investigating one disputed record change.
- Who receives an incident report, what information accompanies it, and what response commitments are offered?
G. Implementation and ongoing service
- Provide a phased delivery plan with outputs and buyer effort for each phase.
- Distinguish the first live workflow from completion of the entire contracted rollout.
- Which realistic exception tests must pass before the proposed release?
- How are new employees trained, and which materials remain available after launch?
- Who owns support, configuration changes, and performance reviews after the implementation team leaves?
H. Pricing and transition
- Define every billable unit, including duplicates, retries, unsuccessful attempts, and test activity.
- Quote the same base, growth, and low-volume scenarios using our supplied assumptions.
- Separate implementation, integration, usage, support, change requests, and external charges.
- Describe transition assistance, export format, charges, and remaining work at the end of service.
- Identify each material proposal assumption that still requires discovery or written agreement.
The questions in section D deliberately ask about administrative handoffs. If your purchase includes authorization or other regulated workflows, the accountable specialists should add the relevant requirements. Do not convert a general purchasing questionnaire into an unreviewed clinical or coverage rulebook.
Turn three answers into a demonstration script
Written answers tell you what to inspect. Choose a normal referral, a repeated transmission, and an incomplete packet. Use synthetic records that reflect your formats without exposing actual patient information in an unapproved setting.
For the repeated transmission, ask the presenter to receive a packet, record its destination identifier, and submit it again. Then add a new attachment to the same request. The demonstration should make clear which receipt events exist, which request they belong to, and whether staff must reconcile them.
For the incomplete packet, require the reviewer to identify the missing item, assign the request, receive a correction, and resume the workflow. Observe the history, not just the final green status.
Capture the test environment and setup. A demonstration with prepared data establishes less than a test against your own configured environment. Record that distinction without treating the demonstration as worthless.
For cross-site scenarios, use the dedicated multi-site evaluation guide. Keep its detailed variation tests outside this general question bank.
Score evidence without letting averages hide a missing requirement
Agree on mandatory requirements before proposals arrive. Examples might include an acceptable export path, a necessary connection, or completion of specialist reviews. Your organization decides which are mandatory and why.
For preferences, use a simple evidence scale: zero for unanswered, one for an unsupported claim, two for a specific documented answer, and three for a successfully demonstrated requirement in the agreed setting. Record evidence relevance alongside the number.
Hypothetical scoring example: A team assigns 40 points to workflow fit, 35 to integration evidence, and 25 to implementation support. Candidate A receives normalized category scores of 80%, 60%, and 80%. Its weighted result is 32 + 21 + 20 = 73 points. Candidate B scores 70%, 80%, and 60%, giving 28 + 28 + 15 = 71 points.
A two-point difference does not settle the purchase. Candidate A still has an unverified mandatory export requirement. Record it as unresolved and obtain evidence before a decision. Do not let its higher total erase the missing requirement.
Ask evaluators to record their scores independently, then reconcile disagreements by inspecting evidence. “I liked the demo” is not an evidence reference. Neither is a customer's success story from a materially different deployment.
Finish with a decision record, not another presentation
Create one page with the recommended scope, satisfied mandatory requirements, open conditions, evidence links, estimated retained work, and reasons for accepting tradeoffs.
Send pricing questions to the quote-normalization worksheet. Keep expected benefits in a separate model so uncertain savings do not disguise actual contractual costs.
Have procurement document which representations belong in the final agreement and record unresolved differences. Keep a traceable record from requirement to answer to agreed scope.
When your team is ready to evaluate referral coordination software, bring the scope page and the unanswered questions.
FAQ
Can a small practice use this without a formal RFP?
How many vendors should receive the questionnaire?
What if a vendor will only show evidence under an NDA?
Should a roadmap commitment receive partial credit?
What if two evaluators score the same answer differently?
Sources
- NIST AI Risk Management Framework, voluntary AI evaluation and risk-management context.