The Idea Funnel — Gathering, Registering, and Evaluating Ideas¶

Summary¶

Walks the first half of the course's idea-funnel spine: introducing the funnel concept itself, the one-hour AI-literacy training that seeds good ideas, idea generation, the submission form, the idea registry and its metadata, problem statements, evaluation criteria, feasibility and risk/benefit/cost scoring, scoring rubrics, expert review panels, feedback loops, and recognition awards. Readers can run a live idea-generation session and populate a registry for their own institution.

Concepts Covered¶

This chapter covers the following 17 concepts from the learning graph:

Idea Funnel
Idea Generation
AI Literacy Training
Idea Submission Form
Idea Registry
Idea Metadata
Problem Statement
Idea Evaluation
Feasibility Assessment
Risk Scoring
Benefit Scoring
Cost Estimation
Scoring Rubric
Expert Review Panel
Idea Feedback Loop
Idea Recognition Awards
Project Selection

Prerequisites¶

This chapter builds on concepts from:

Welcome to Chapter 5

Sage waving welcome Every great AI initiative in education started as someone's good idea — a teacher frustrated with paperwork, a counselor who wanted earlier warning about struggling students, a principal who noticed a pattern no one had time to act on. This chapter builds the system that catches those ideas and turns the best ones into funded projects. "Let's chart the course!"

The Idea Funnel — Your AI Operating System¶

The idea funnel is the central operational structure of this course — the repeatable, six-stage process through which an institution continuously discovers, evaluates, selects, and learns from AI initiatives. Think of it as the operating system for your AI strategy: just as an operating system manages the flow of resources in a computer, the idea funnel manages the flow of ideas, talent, and budget across your AI portfolio.

The funnel metaphor is deliberate: many ideas enter at the wide top; a much smaller number of funded, resourced projects emerge at the narrow bottom. This is not inefficiency — it is quality control. The purpose of the funnel is to ensure that the projects your institution actually funds are the ones with the strongest combination of feasibility, impact, equity, and strategic fit, not simply the ones proposed by the loudest voices or the most senior administrators.

The six stages of the idea funnel are:

Gather: Run AI literacy sessions and open submission channels to collect ideas from staff, teachers, students, and families.
Register: Record every submitted idea in a structured idea registry with standardized metadata.
Evaluate: Score each idea against criteria for feasibility, risk, benefit, cost, and equity.
Select: Choose a small portfolio of projects to fund, balancing quick wins and strategic bets.
Resource: Assign people, budget, and timeline to funded projects.
Evaluate outcomes: Measure project results against pre-defined success metrics and feed lessons back into stage 1.

This chapter covers stages 1 through 4. Chapter 6 covers stages 5 and 6.

Diagram: The Idea Funnel — Six Stages¶

Run The Idea Funnel — Six Stages Fullscreen

Interactive funnel diagram showing the six stages from idea to funded project

Type: chart sim-id: idea-funnel-stages
Library: vis-network
Status: Specified

Learning objective: Understanding (Bloom's) — readers trace the path of an idea through all six funnel stages and explain what happens at each gate.

Canvas: Responsive, full container width, 500px height.

Layout: Vertical funnel — wide at top, narrow at bottom. Each stage is a trapezoid layer, wider at top than bottom.

Funnel layers (top to bottom, each clickable): - Layer 1 (widest, light gray): "Stage 1: Gather" — Infobox: "Run one-hour AI literacy sessions. Open submission forms to teachers, staff, students, and families. Goal: many diverse ideas, no filtering yet." - Layer 2 (steelblue): "Stage 2: Register" — Infobox: "Enter every submitted idea into the idea registry with standardized metadata: problem statement, proposed approach, expected benefit, affected stakeholders, rough cost, and risk flag." - Layer 3 (teal): "Stage 3: Evaluate" — Infobox: "Score each idea on a rubric: feasibility (1–5), risk (1–5), benefit (1–5), cost (1–5), equity impact (1–5). Expert review panel scores ideas above a threshold." - Layer 4 (orange): "Stage 4: Select" — Infobox: "Review scored ideas as a portfolio. Select projects balancing quick wins (high feasibility, lower impact) and strategic bets (transformative but harder). Board or executive approval." - Layer 5 (deep orange): "Stage 5: Resource" — Infobox: "Assign team members, budget, and timeline to funded projects. Create project charters. Establish success metrics." - Layer 6 (narrowest, deep orange/red): "Stage 6: Evaluate Outcomes" — Infobox: "Measure results against pre-defined KPIs. Document lessons learned. Feed insights back into Stage 1 for the next cycle."

Arrows: Down the right side of the funnel, showing flow from one stage to the next. Side labels: On the left, show approximate quantities: Stage 1: "50–200 ideas/year", Stage 2: "All submitted ideas", Stage 3: "Ideas scoring above minimum threshold", Stage 4: "Top 10–20%", Stage 5: "Funded projects (3–10/year)", Stage 6: "Completed projects".

Interaction: Clicking a layer highlights it, dims others, and opens its infobox. Responsive: SVG-based funnel redraws on window resize.

Stage 1: Gathering Ideas — AI Literacy Training¶

The most common failure mode of idea-generation programs is that the ideas submitted are too narrow: staff submit ideas about tools they already know, rather than thinking creatively about problems they actually have. The antidote is a structured AI literacy training session — typically one hour — that precedes idea generation.

The one-hour AI literacy session has three objectives:

Calibrate expectations: Show staff what current AI can and cannot do with live demonstrations. Demonstrate AI drafting a lesson plan, generating differentiated reading passages, and explaining a concept in multiple ways. Equally important: demonstrate AI hallucinating confidently incorrect information, to calibrate appropriate trust.
Identify problem categories: Walk staff through the four broad AI-use-case categories introduced in Chapter 3 (instruction enhancement, teacher support, administration, student support) and prompt them to think about which problems in their daily work fall into each.
Model good problem statements: Show the difference between a bad idea submission ("we should use AI for everything") and a good one ("I spend 45 minutes per week writing substitute teacher plans — I think AI could draft these in five minutes"). The problem-first framing is the key skill.

After the training, staff are invited to submit ideas through a structured idea submission form — not an open-ended suggestion box, but a structured template that forces submitters to think through the problem before proposing a solution.

The Idea Submission Form¶

An idea submission form is the structured intake mechanism for the idea funnel. Its purpose is twofold: it forces submitters to articulate their problem clearly, and it captures the metadata the review team needs to evaluate and prioritize ideas without extensive follow-up interviews.

A well-designed idea submission form collects:

Your role and department: (Contextualizes the problem and affected stakeholders.)
Problem statement: In one to three sentences, describe the specific problem you are trying to solve. Avoid proposing a solution at this stage.
Who is affected: Who experiences this problem and how frequently?
Current workaround: How is the problem handled today, and what does that cost in time or resources?
Proposed AI approach: (Optional, and clearly labeled as preliminary.) What type of AI do you think might help?
Estimated benefit: If this problem were solved, what would improve? Be specific — fewer than 30 words.
Data involved: Would solving this require AI to access student data, parent data, or other sensitive information?
Urgency: Is this a nice-to-have, or is there a timeline driving the need?

The form should be available year-round, not just during an annual submission window. Problems do not follow a calendar, and some of the best ideas emerge in the middle of a busy semester when a teacher reaches a breaking point with a repetitive task.

The Idea Registry¶

Every submitted idea, regardless of its apparent quality, is recorded in the idea registry — a shared, searchable database that becomes one of the institution's most valuable strategic assets over time. The registry serves three functions:

Memory: Ideas submitted today may not be feasible today but could be highly feasible in 12 months when the AI landscape has shifted. A registry ensures good ideas are not lost.
Pattern recognition: When multiple staff members submit ideas about the same problem independently, that convergence signals a high-priority opportunity that might not be visible from any single submission.
Transparency: Staff who submit ideas and see them entered into a tracked, searchable registry feel that their contribution is taken seriously — even if their specific idea is not funded.

The registry can be as simple as a shared spreadsheet or as sophisticated as a purpose-built project management tool with workflow automation. What matters is consistency and completeness of idea metadata — the structured information fields captured for every idea.

Idea Metadata — What to Capture¶

Idea metadata is the standardized set of fields recorded for every submission. Consistent metadata is what makes the registry searchable, comparable, and analyzable. The following table shows the standard metadata fields, what they contain, and why each matters.

Metadata Field	What It Contains	Why It Matters
Idea ID	Unique identifier (e.g., IDEA-2025-047)	Enables tracking and reference throughout the funnel
Submission date	Date submitted	Enables trend analysis over time
Submitter role	Teacher, administrator, parent, student	Ensures all voices are represented in analysis
Problem statement	1–3 sentence description of the problem	The most important field; all evaluation starts here
Affected stakeholders	Who experiences the problem and how many	Scales the potential impact
Current workaround	How it's handled today	Establishes the baseline cost of inaction
AI approach (preliminary)	General category of proposed solution	Optional; helps route to appropriate reviewers
Data sensitivity	Student data involved? (Yes/No/Unknown)	Triggers FERPA/privacy review automatically
Feasibility score	1–5, assigned by reviewer	Set during evaluation stage
Risk score	1–5, assigned by reviewer	Set during evaluation stage
Benefit score	1–5, assigned by reviewer	Set during evaluation stage
Cost estimate	Rough order of magnitude ($1K/$10K/$100K)	Set during evaluation stage
Equity impact	Positive/Neutral/Negative, with notes	Set during evaluation stage
Status	Submitted/Under Review/Selected/Declined/Archived	Updated as idea moves through funnel

Stage 2: The Problem Statement¶

The problem statement is the most important element of any idea submission — and the hardest to write well. A good problem statement is:

Specific: "I spend 45 minutes writing sub plans for each absent day" is specific. "Our administrative processes are inefficient" is not.
Measurable: It implies a baseline that can be measured and a target that would indicate success.
Problem-focused, not solution-focused: It describes the pain point, not the proposed fix. The best problem statements generate multiple possible AI approaches, not just one.
Written from the perspective of the affected person: "Our counselors cannot identify at-risk students until they have already failed two or more classes" is better than "We need an early warning system."

A common mistake is conflating problem statements with solution proposals. "We should use ChatGPT to write lesson plans" is a solution proposal, not a problem statement. The underlying problem might be "teachers spend an average of 3 hours per week on lesson planning that does not differentiate for student needs" — a problem statement that might be addressed by AI lesson planning, but also by improved collaboration tools, better curriculum libraries, or a restructured planning period. Keeping problem statements solution-agnostic ensures that the best solution — AI or otherwise — gets found.

Sage thinks about problem statements

Sage thinking The best AI strategy question is never "how can we use AI?" It is "what problems are we trying to solve, and is AI the most effective tool for each?" Problem statements keep the focus on outcomes rather than tools. A good problem statement makes the success metric obvious before you've even proposed a solution.

Stage 3: Evaluating Ideas¶

Idea evaluation is the process of scoring submitted ideas against a consistent set of criteria so that they can be compared and prioritized. Without a structured evaluation process, idea selection defaults to whoever has the most political capital or the most compelling presentation — which is not the same as identifying the highest-value opportunities.

The five evaluation dimensions are:

Feasibility Assessment¶

Feasibility assessment scores how realistic and achievable the proposed solution is, given current technology, institutional capacity, and timeline. A feasibility score of 5 means "this can be implemented with existing tools and skills in a few months." A score of 1 means "this requires capabilities that do not yet reliably exist, or institutional capacity we do not have." Key feasibility questions:

Does mature AI technology exist for this use case?
Does the institution have the technical staff or vendor relationships to implement it?
Can it be piloted with a small group before full deployment?
Is there a clear pathway from pilot to full deployment?

Risk Scoring¶

Risk scoring assesses the potential negative consequences of the initiative — data privacy exposure, unintended bias, academic integrity concerns, vendor dependency, and student well-being impacts. Risk scoring is not about avoiding all risk but about making risk visible and deliberate. A score of 5 means "high inherent risk requiring significant mitigation measures." A score of 1 means "low risk, limited sensitive data involved, well-understood implementation." Key risk questions:

Does this involve student personally identifiable information?
Could this initiative have disproportionately negative impacts on specific student populations?
What happens if the vendor discontinues the product or raises prices significantly?
Could this reduce human oversight in a way that harms students?

Benefit Scoring¶

Benefit scoring assesses the potential positive impact of the initiative — on student learning outcomes, teacher effectiveness, administrative efficiency, equity of access, or community trust. A benefit score of 5 means "high potential to significantly improve outcomes for many students or staff." A score of 1 means "small quality-of-life improvement for a limited group." Key benefit questions:

How many students or staff does this affect?
How significantly does it affect them?
Does it address a problem that current approaches genuinely cannot solve well?
Does it create measurable, trackable improvements?

Cost Estimation¶

Cost estimation at the idea stage is intentionally rough — a cost estimate is an order-of-magnitude judgment, not a budget line item. The goal is to sort ideas into rough cost tiers:

Under $10,000: Typically software subscriptions, API access, and staff time for a small pilot.
$10,000–$100,000: Larger software deployments, professional development programs, or custom integrations.
$100,000+: Enterprise software platforms, significant infrastructure changes, or multi-year transformation initiatives.

The cost estimate at the idea stage should include not just the vendor cost but a rough estimate of implementation effort, training, and ongoing maintenance — a simplified total cost of ownership check.

The Scoring Rubric¶

A scoring rubric is the structured guide that ensures all reviewers apply evaluation criteria consistently. Without a rubric, two reviewers with different experiences will score the same idea differently in ways that reflect their backgrounds rather than the idea's merits. A scoring rubric defines what each score level means for each dimension.

Before examining the rubric structure, note that equity impact is scored separately — it is not a dimension to optimize but a constraint to check. An idea with an equity impact of "negative" (the initiative would disproportionately benefit already-advantaged students or harm under-resourced ones) should not proceed without a plan to address the disparity, regardless of its other scores.

Diagram: Interactive Idea Scoring Rubric¶

Run Interactive Idea Scoring Rubric Fullscreen

Interactive scoring tool — apply the rubric to a sample idea

Type: MicroSim sim-id: idea-scoring-rubric
Library: p5.js
Status: Specified

Learning objective: Applying (Bloom's) — readers apply the five-dimension scoring rubric to a sample idea and calculate a composite score.

Canvas: Responsive, container-width × 500px.

Controls (p5.js builtins, below canvas): - Five sliders (one per dimension): - "Feasibility" — range 1–5, default 3, step 1 - "Benefit" — range 1–5, default 3, step 1 - "Risk (lower = riskier)" — range 1–5, default 3, step 1 - "Cost-Effectiveness (lower = more expensive)" — range 1–5, default 3, step 1 - "Equity Impact" — select dropdown: "Positive (+1)", "Neutral (0)", "Negative (−2)" - Button: "Reset" — restores all to default.

Visualization: - Radar/spider chart with five axes (Feasibility, Benefit, Risk, Cost-Effectiveness, Equity). - Filled polygon showing current scores; polygon updates in real time as sliders change. - Score color shifts from red (low composite) through yellow to green (high composite). - Composite score displayed prominently: Composite = Feasibility + Benefit + Risk + Cost − Equity adjustment (where negative equity subtracts 2 points). - Interpretation band shown below score: 4–8 = "Low priority", 9–14 = "Consider for pilot", 15–19 = "Strong candidate", 20+ = "High priority".

Sample idea panel (static, left side): "A third-grade teacher proposes using an AI reading tool to generate differentiated reading passages for each student's level. Students would access these through the school's existing reading platform. No new student data would be collected beyond what is already tracked."

Responsive: updateCanvasSize() called in setup() and on window resize. canvas.parent(document.querySelector('main'));

Stage 3: The Expert Review Panel¶

Not all ideas can be fully evaluated by the idea funnel coordinator alone. Ideas above a threshold composite score — typically those that appear in the "Consider for pilot" or "Strong candidate" bands — move to an expert review panel for deeper assessment.

An expert review panel is a small, cross-functional group (typically 4–7 members) that meets quarterly to evaluate high-potential ideas. Panel composition matters: a panel of only administrators will underweight teacher feasibility concerns; a panel of only teachers will underweight privacy and equity considerations. Effective panels typically include:

A curriculum or instructional design specialist
A technology or IT representative
A data privacy or compliance officer
A classroom teacher from an affected grade or subject
A student representative (particularly for initiatives that directly affect student experience)
A parent or community representative (rotating)

The review panel does not make final funding decisions — that authority typically rests with the superintendent or a designated AI steering committee. The panel's role is to deepen the evaluation, flag concerns that the rubric scores may not capture, and prepare a summary recommendation for the decision-makers.

Idea Feedback and Recognition¶

Two practices turn an idea funnel from a one-way suggestion box into a sustainable innovation culture: structured feedback and meaningful recognition.

The idea feedback loop is the process by which every submitter receives a response to their idea — not just a generic acknowledgment, but specific information about how their idea was scored, why it was advanced or declined at this time, and what would need to change for it to move forward in a future cycle. A feedback loop has three benefits:

Submitters learn what makes ideas more or less fundable, improving the quality of future submissions.
Staff who receive honest, specific feedback feel that their ideas were genuinely considered, even when they were not funded.
Ideas that are declined because the technology is not mature enough can be automatically flagged for re-evaluation in six months.

Idea recognition awards acknowledge the submitters of ideas that are funded, implemented, and produce measurable results. Recognition does not need to be financial — a presentation to the school board, a feature in the district newsletter, or a certificate of recognition can be powerful incentives. What matters is that teachers and staff can see that submitting ideas has consequences — not just the hope of consequences.

Sage's Tip: Recognize the Problem-Finders

Sage giving a tip The most valuable contributors to your idea funnel are not the people who know the most about AI — they are the people who know the most about your institution's unsolved problems. Recognize problem-finders explicitly, not just solution-proposers. The teacher who articulated a problem clearly enough that someone else could solve it deserves as much credit as the person who built the solution.

Stage 4: Project Selection¶

Project selection is the decision process that determines which evaluated ideas receive funding and resources. Selection is not simply choosing the highest-scoring ideas — it is building a portfolio that balances several strategic objectives:

Quick wins: Projects with high feasibility and near-term impact that build confidence, generate visible results, and demonstrate that the idea funnel actually produces outcomes. Quick wins typically have composite scores above 15 and implementation timelines under six months.
Strategic bets: Projects with higher impact but greater complexity or longer timelines that move the institution toward its longer-term vision. Strategic bets are investments; they require more resources and more patience.
Equity priority: Projects that specifically address under-served student populations or reduce resource disparities, even if their composite scores are not the highest. Equity considerations should be explicit in selection discussions, not assumed to be captured in the rubric.
Portfolio diversity: A mix of initiatives across the four use-case categories (instruction enhancement, teacher support, administration, student support) so that no single department captures all available AI investment.

Selection should be documented. For every idea that is funded, document why it was selected over alternatives. For every idea that is declined, document why and under what conditions it might be reconsidered. This documentation is the institutional memory that makes strategy credible — and makes it possible to evaluate the quality of selection decisions over time.

Sage Celebrates Your Progress

Sage celebrating You can now run an idea-generation session, build a registry, evaluate ideas with a rubric, convene a review panel, and select a portfolio with deliberate balance. That is the first half of your AI operating system — fully operational. Chapter 6 completes the funnel with resourcing, execution, and outcome evaluation.

Key Takeaways¶

The idea funnel is a six-stage, repeatable process — Gather, Register, Evaluate, Select, Resource, Evaluate Outcomes — that is the operational backbone of an AI strategy.
AI literacy training (one hour, run before idea submission) calibrates what staff believe AI can do and prompts problem-first thinking.
A structured idea submission form and idea registry capture every submission with consistent idea metadata — the foundation for searchable, trackable, comparable ideas.
A good problem statement is specific, measurable, and problem-focused rather than solution-focused.
Idea evaluation scores each submission on five dimensions: feasibility assessment, risk scoring, benefit scoring, cost estimation, and equity impact.
A scoring rubric ensures reviewers apply criteria consistently; the expert review panel deepens assessment for high-potential ideas.
The idea feedback loop and idea recognition awards sustain the culture of contribution that makes the funnel self-renewing.
Project selection builds a portfolio balancing quick wins, strategic bets, equity priorities, and diversity across use-case categories.