Use Case Worksheet
Score and prioritize AI opportunities across branches, programs, and back-office functions so leadership can launch with the lowest-risk, highest-impact pilots.
At a Glance
- Assess AI fit across circulation, reference, and program planning.
- Use a simple scoring model: impact, risk, effort, and measurability.
- Decide what should be piloted, paused, or rejected.
- Document dependencies: data sources, approvals, and staff owners.
- Identify quick wins for frontline relief and knowledge management.
- Align with procurement and privacy requirements before building.
- Create a 60-day pilot plan with success metrics.
Problem / Why It Matters
Many library teams collect promising AI ideas—automated program blurbs, volunteer onboarding helpers, or grant narrative drafting—but struggle to decide what to tackle first. Without a shared rubric, the loudest idea or latest trend wins, not the one that actually reduces staff load or improves patron experience.
Risk varies widely across use cases. Drafting a newsletter is low-risk; suggesting legal resources to patrons is not. A worksheet that weighs impact, effort, and governance keeps the team honest about what belongs in a pilot and what requires more policy work.
Stakeholders need clarity on ownership. IT wants to know about integrations, public services worries about tone, and communications teams care about brand alignment. A structured worksheet aligns teams early, preventing surprises when the pilot goes live at a busy branch.
Recommended Workflow
- 1
Collect candidate ideas
Gather ideas from front-line staff, program managers, and IT. Encourage examples like “draft reply for lost-and-found inquiries,” “summarize board packet,” or “generate book display themes.”
- 2
Define evaluation criteria
Use four scores (1–5): Impact (time saved or patron value), Risk (data sensitivity and reputational risk), Effort (data, integrations, training), and Measurability (clear metrics). Keep definitions visible on the worksheet.
- 3
Score collaboratively
Hold a 60–90 minute workshop with stakeholders from public services, programs, IT, and communications. Score each idea together to avoid optimism bias. Capture comments explaining the scores.
Decision point: If any idea involves PII, minors, or legal/health topics, flag it for extra governance review before piloting.
- 4
Map dependencies
List required documents (policies, scripts, handouts), systems (ILS, ticketing, website CMS), and approvers (legal, union, board). Identify whether data can stay in AskLib or needs an integration.
- 5
Prioritize into buckets
Group ideas into “Pilot Now,” “Research,” and “Hold.” Prioritize items with high impact, low risk, and measurable outcomes—like drafting policy-aligned email replies or summarizing workshop feedback.
- 6
Design the pilot
For each “Pilot Now” item, define the scope, owner, start date, and success metrics (e.g., reduce response time by 30%, create 10 reusable prompts, or cut meeting room complaints).
Decision point: If success depends on training, include a 1-hour staff orientation and a feedback loop every two weeks.
- 7
Prepare guardrails
Create prompt templates that include do/don’t instructions, banned topics, and escalation paths. Add keyword blocks for sensitive data and confirm review/approval steps.
- 8
Launch and measure
Run the pilot for 30–60 days. Track usage, sentiment, and policy questions surfaced by staff. Compare metrics against your baseline and adjust templates as needed.
- 9
Decide on expansion
If metrics meet targets and risks are managed, scale to more branches or programs. If not, capture lessons learned and update the worksheet scoring for transparency.
Best Practices
- Keep the worksheet short—one row per idea with links to source documents—so teams actually fill it out.
- Use concrete metrics: “minutes saved per patron email” or “number of reusable prompts created,” not vague satisfaction scores.
- Favor tasks with existing content libraries (policies, scripts, handouts); they ground AI outputs and lower hallucination risk.
- Include staff burden as an impact dimension—front-desk relief often outperforms flashy patron-facing experiments.
- Note accessibility needs in the worksheet so pilots bake in translation, alt-text, and plain-language requirements.
- Assign an owner for each use case; ideas without owners rarely progress beyond discussion.
- Review legal and union considerations early, especially for changes to workflows or volunteer roles.
- Document why ideas are rejected; it reduces re-litigation and shows staff their suggestions were considered.
- Pilot in one or two branches before district-wide rollout; collect feedback from diverse communities.
- Schedule a short retro at day 30 to adjust prompts, add guardrails, and refine metrics.
Common Pitfalls
Scoring in isolation
Avoid by: Score as a cross-functional group so risk and effort are realistic, not optimistic.
Ignoring data availability
Avoid by: List required documents and integrations; if you lack clean sources, move the idea to “Research.”
Overweighting novelty
Avoid by: Prioritize boring, high-volume tasks (circulation replies, program blurbs) over experimental ideas.
No path to measurement
Avoid by: Define baseline metrics before launch; otherwise you cannot prove success or win board support.
Unclear approval flows
Avoid by: Add approver names and timelines for each use case—communications, IT, legal—to prevent stalls.
Skipping safety reviews
Avoid by: Run sensitive ideas through privacy and equity checks; add keyword blocks for PII and protected classes.
Launching too wide
Avoid by: Pilot with a small cohort and clear feedback loops before scaling to all branches.
Leaving volunteers out
Avoid by: Include volunteer coordinators in scoring; their workflows often benefit from AI-assisted guides.
Examples
Circulation inbox relief
Use case: Draft replies for recurring questions about renewals, lost cards, or interlibrary loan timelines. Impact is high, risk is moderate, effort is low because policies already exist.
Program marketing accelerator
Use case: Generate bilingual event blurbs, social copy, and signage templates for recurring programs. Impact is medium-high, risk is low, effort is medium due to translation review needs.
Board packet summarization
Use case: Summarize lengthy board packets into 1-page staff briefings. Impact is medium, risk is medium (governance content), effort is medium because documents are available but need permissioned access.
Homework help guardrails
Use case: Offer policy-aligned homework help that avoids generating full answers. Impact is high for patrons but risk is also high; worksheet flags it for deeper governance before piloting.
Security & Governance Notes
What data should never be entered
- Exclude patron PII, HR records, and student education records from scoring descriptions.
- Avoid including confidential board or legal content unless approvers sign off.
- Do not paste vendor API keys or system credentials into worksheets or prompts.
Suggested keyword blocks
- Flag terms like “diagnosis,” “legal advice,” “SSN,” “immigration status,” and “disciplinary action.”
- Set alerts for structured numbers (card numbers, student IDs) to prevent accidental inclusion.
Role-based access guidance
- Governance or privacy lead reviews high-risk ideas before they move to pilot.
- IT validates integration feasibility and data isolation requirements.
- Communications approves any public-facing automation or templated responses.
Audit & logging
- Keep a versioned copy of each worksheet with scores and decision notes.
- Log rationale for moving ideas between buckets; transparency builds trust with staff and boards.
- Document which guardrails (keyword blocks, templates) were added during pilot design.
Retention suggestions
- Retain worksheets for at least one fiscal year to inform grant reporting and audits.
- Purge draft ideas that contain sensitive descriptions once decisions are finalized.
- Store pilot metrics separately from raw prompts to avoid keeping unnecessary data.
FAQ
How many use cases should we pilot at once?
Start with one to three. Focus on tasks with clear metrics and low to moderate risk, such as drafting circulation replies or creating program blurbs. Running too many pilots dilutes training and measurement.
Who should own the worksheet?
Assign ownership to a cross-functional lead—often the digital services manager or innovation lead. They coordinate scoring sessions, ensure approvals happen, and update the worksheet as decisions are made.
What if scores are tied?
Use tie-breakers: urgency (seasonal programming), staff burden (front desk relief), and governance readiness. If still tied, pilot the use case with clearer measurability first.
How do we include equity considerations?
Add a prompt asking whether the idea improves access for under-served patrons, supports language equity, or risks reinforcing bias. Reflect that in the Impact and Risk scores.
Do we need IT involvement for every idea?
No. Many pilots rely on documents and templates only. Bring IT in when integrations, authentication, or analytics pipelines are required.
How do we keep the worksheet current?
Schedule quarterly reviews. Update scores based on pilot results, new policies, or changes in staffing. Retire ideas that no longer align with strategic goals.
Can we share the worksheet with the board?
Yes. A transparent worksheet demonstrates due diligence. Redact sensitive details and include a short summary of why certain ideas are paused for governance reasons.
What metrics work well for pilots?
Measure turnaround time for patron communications, number of reusable prompts created, decrease in policy clarification emails, and staff satisfaction with clarity of templates. For analytics, use aggregated themes rather than storing full transcripts.
How do we handle staff suggestions that are rejected?
Document the reason in the worksheet (e.g., “requires student data; need district approval”). Share updates at staff meetings to show suggestions were considered and to revisit when conditions change.
Related Guides
Talk to us
Request a demo or ask for a customized version of this guide.