AssessmentAlly — Requirements

Goals

• Cut the time to produce a defensible property-tax appeal package from a multi-day research project to under 20 minutes.
• Make the math visible: at every step, the user should see how each input changes their estimated annual and 3-year savings.
• Produce a county-ready output (narrative + comp set + condition worksheet) that a homeowner can submit without paid help.
• Convert latent intent ("I should appeal someday") into action before the deadline expires.

Primary user

Owner-occupant, age 35–70, who has just received a reassessment notice that materially raises their assessed value. They are financially literate but not a real-estate professional. Context of use: at home, on a laptop or large phone, in the evening, with the assessor's notice next to them. The job-to-be-done is "help me decide if I have a case, and if I do, hand me a package I can submit."

Functional requirements

• FR1: User can select a sample parcel on the landing screen and enter the wizard at Step 1.
• FR2: A persistent header shows the parcel's appeal deadline as a relative pill ("12 days left") and turns red within 7 days.
• FR3: Step 1 shows assessor-of-record property facts (sqft, beds, baths, year built, lot size). Each is individually editable; edits are flagged as "homeowner correction."
• FR4: Step 2 lists at least 6 nearby comparable sales sorted by relevance score; user can include/exclude each with a single click.
• FR5: Including/excluding a comp updates the proposed value (a weighted price-per-sqft of selected comps × subject sqft) live, with no page reload.
• FR6: Step 3 presents a checklist of common condition issues (roof age, HVAC age, foundation, kitchen, water damage, deferred exterior, structural). Each item has a default downward adjustment that the user can edit.
• FR7: Step 4 shows a side-by-side comparison of "current assessed × mill rate" vs. "proposed × mill rate," annual savings, and 3-year projected savings.
• FR8: Step 5 renders an appeal-letter preview that includes property facts, the selected comps, the condition adjustments, and the resulting proposed value.
• FR9: A progress bar shows wizard step state and supports forward/back navigation without losing entered data.
• FR10: All data persists in localStorage per parcel, so the user can come back later and pick up where they left off.
• FR11: All numeric inputs validate as positive numbers; obviously invalid entries (e.g., negative sqft) are rejected with an inline message.
• FR12: The app loads its sample data from an inline <script type="application/json"> block so it runs from file:// without a local server.

User stories

• As a first-time appellant, I want to start from a sample parcel similar to mine, so that I can see what a finished appeal looks like before I commit my own data.
• As a homeowner who's busy, I want each step to take under 3 minutes, so that I can finish the whole flow in one sitting.
• As a skeptical user, I want to see the math behind the proposed value, so that I trust the recommendation enough to put my name on it.
• As a homeowner with a worn roof, I want to claim a condition adjustment with a justification, so that the assessor's reviewer takes my appeal seriously.
• As a deadline-pressured user, I want a visible countdown to my filing deadline, so that I know how urgently to finish.
• As a careful user, I want to come back tomorrow and pick up where I left off, so that I don't have to re-enter everything.
• As a money-motivated user, I want to see annual and 3-year savings up-front, so that I know whether the effort is worth it before I begin.
• As a non-technical user, I want one screen at a time with clear next/back, so that I'm never lost in the flow.

Non-functional requirements

• Runs from file:// with no build step and no remote scripts that require auth.
• All third-party assets must be loaded from public HTTPS CDNs or be fully local; the prototype includes only Google Fonts via HTTPS.
• Mobile-first responsive layout — fully usable down to 360px wide.
• No PII leaves the browser in the prototype: all state is in localStorage, no analytics, no network calls.
• Color contrast meets WCAG 2.2 AA; all interactive controls are keyboard-accessible.
• First contentful paint under 1.5s on a mid-tier laptop opening from disk.

Out of scope (for the prototype)

• Real county-portal integrations (parcel lookup, e-file, bill pull).
• Live MLS / ATTOM / Zillow comp feeds.
• Server-side PDF generation (the wizard renders an HTML preview only).
• Photo upload and image storage for condition evidence.
• Authentication and multi-device sync.
• Payment processing for the $49 / $99 / success-fee tiers.
• Per-county appeal-form template library (US has ~3,000 counties; production would need a curated subset).

Open questions

• Which county should the first-county-integration MVP target — Travis (TX), Wake (NC), or Cook (IL)? Each has high parcel volume but very different appeal mechanics.
• Should the success-fee option be flat 15% or tiered by savings band?
• How do we source initial comp data without an MLS license — is county recorder + Zillow scraping viable, or should we partner directly with ATTOM?
• What's the right voice for the generated narrative — formal/legalistic, or plain-English homeowner-narrated? A/B test candidate.
• Does the brokerage white-label channel cannibalize direct sales, or does it expand the pie because brokerages reach customers we wouldn't?