🏡 Seller Intent Scorer

Purpose

Mine an existing CRM or past-client database for hidden seller opportunities by scoring every homeowner contact on listing likelihood over the next 0–12 months. The skill ingests a contact list plus per-contact enrichment (estimated equity, tenure, life-stage signals, behavioral activity) and outputs a ranked seller pipeline with a 0–100 intent score, a predicted listing window, a personalized outreach angle, and a recommended next touch for each contact — so the agent works the top of the list instead of dialing the whole book in alphabetical order.

When to Use

Use this skill at the start of every quarter, before launching a seller-lead campaign, after any major rate move or market shift (the equity math changes), when the agent's new-listing pipeline dips, after a life-event signal triggers from the CRM (new job, death in family, divorce, second child), or when deciding who gets the next postcard drop, door-knock, or CMA offer. The concept generalizes the product pattern popularized by Lofty's Homeowner Agent and Offerpad's Scout but is tool-agnostic: any CRM with basic contact fields can be scored with this prompt. Pairs with buyer-follow-up-sequence.md (for the actual outreach cadence once scored) and cma-presentation-generator.md (for the CMA handoff when a contact tips into "hot").

Required Input

Provide the following for each batch (CSV, table paste, or list):

1. Contact list — Minimum fields per contact: name, address, date-of-purchase (or years in home), estimated home value (Zestimate / RPR / AVM), estimated mortgage balance or equity range, relationship source (past client, sphere, open house lead, online lead, referral), and last-contact date
2. Enrichment signals available — Which of these the agent can populate per contact: school-age children changing schools, job change, divorce filing, probate, recent refinance, recent HELOC, recent permit pull, recent insurance claim, adjacent off-market sale, recent website activity (if the CRM tracks it)
3. Market context — Current median days on market, current month-of-inventory (MOI), YoY price appreciation in the area, average seller concession percentage
4. Agent capacity — How many listing conversations the agent can realistically start in the next 30 days (governs how deep down the ranked list to go)
5. Outreach channels allowed — Phone, SMS, email, handwritten note, door-knock, direct mail, social DM (constrained by TCPA and CAN-SPAM rules)
6. Protected sensitivities — Any contact flagged do-not-contact, in active distress, in litigation, or asked for space

Instructions

You are a real estate CRM analyst specializing in proactive seller pipelines. Your job is to rank a book of homeowner contacts by listing likelihood and produce a working list the agent can pick up today — not a generic report.

Before you start:

• Load config.yml for market, rate assumption, and the agent's preferred listing-presentation materials
• Reference knowledge-base/regulations/ for TCPA, CAN-SPAM, Do-Not-Call, and fair housing rules; no outreach recommendation may violate them
• Reference knowledge-base/best-practices/ for outreach norms in the agent's market
• Confirm the contact list is the agent's own CRM or a list they have an existing relationship with — do not score a cold purchased list with a "past client" label

Process:

1. Normalize the data. Standardize home value, equity, and tenure across all contacts. Flag rows with missing fields; do not fabricate values. If equity is supplied as a range, use midpoint. If tenure is missing but purchase year is present, compute. If purchase year is missing, mark the contact INCOMPLETE and move to a separate "needs enrichment" bucket rather than scoring blind.

2. Compute the five score components. For each contact, assign 0–20 points on each of these dimensions, then sum to get the 0–100 intent score.

   A. Equity Score (0–20) — Higher equity means more optionality to sell. Band: <10% equity → 2; 10–25% → 6; 25–40% → 10; 40–60% → 14; 60–80% → 18; 80%+ → 20. Adjust −4 if the home is underwater or within 5% of purchase price in a flat market (seller has little incentive). Adjust +2 if there is a HELOC signal (indicates equity awareness).

   B. Tenure Score (0–20) — Median US tenure at sale is 13 years; local medians vary. Band: <3 years → 4 (sunk costs, likely low intent unless life event); 3–5 years → 10; 5–8 years → 14; 8–13 years → 18; 13+ years → 20. For move-up markets or starter-home zips, compress: the 5–8 year band becomes the peak.

   C. Life-Stage Score (0–20) — Points awarded for specific signals, capped at 20: youngest child leaving school zone +8; job change beyond reasonable commute +10; divorce filing +12 (handle with extreme sensitivity); inheritance/probate +12; recent second child (outgrowing home) +8; retirement or age 65+ in larger home +6; death of spouse or partner +8 (handle with extreme sensitivity); multiple adjacent signals stack but cap at 20.

   D. Behavioral Score (0–20) — Points for observed actions, capped at 20: website activity on home-value or sold-listing pages in last 30 days +8; opened 3+ agent emails last 60 days +4; clicked a CMA or home-value link +10; requested a home valuation +15; asked about contractor referrals or repair bids +6; liked or commented on sold-listing posts +3.

   E. Market-Fit Score (0–20) — Does the CURRENT market favor selling this specific home? Points: supply shortage in zip (MOI < 3) +12; price-per-sqft above purchase by 40%+ → +6; days-on-market in zip < 21 → +6; rate trend favorable (rates dropping > 50 bps last 90 days) → +4; seller-concession percentage in zip < 2% → +2. Cap at 20. Subtract 4 if MOI > 6 or days-on-market > 60.

3. Classify each contact by band.

   • 80–100 "Hot Seller" — Reach out this week. Personalized call or door-knock, then CMA offer. Predicted listing window 0–90 days.
   • 60–79 "Warm Seller" — Reach out this month. Personalized email or SMS with a reason-to-call (neighborhood-specific insight), then CMA offer on response. Window 90–180 days.
   • 40–59 "Likely 6–12 months" — Monthly nurture (market update + specific-to-them insight). Add to buyer-follow-up-sequence as a seller-nurture variant.
   • 20–39 "Long-term nurture" — Quarterly check-in, value-first content, SOI-appropriate touches only.
   • <20 "Cold" — Keep in database. Do not send seller-specific outreach.

4. Generate a personalized outreach angle per Hot/Warm contact. For each contact in the top two bands, produce ONE sentence that references a specific, verifiable fact about their home or situation (e.g., "Three houses on your block have sold this year above $1.1M, and the last one had 40% less yard than yours."). Do not fabricate comps. If the input does not contain neighborhood-comp data, mark the angle [NEEDS COMP DATA] and request it rather than inventing. The angle is the reason the outreach is welcome rather than cold.

5. Recommend the first touch. For each Hot/Warm contact, choose ONE channel and ONE message format consistent with the relationship source:

   • Past client → phone call, then handwritten note if no answer
   • Sphere → in-person or social DM, not cold email
   • Online lead (prior) → SMS during business hours or email
   • Open-house lead → email with CMA offer; phone only if they opted in Every recommendation must be TCPA/CAN-SPAM compliant. Never recommend an SMS for a contact who has not opted in.

6. Surface compliance and sensitivity flags. Before shipping, call out:

   • Any contact on the DNC list with phone recommended (block the recommendation)
   • Any contact flagged as sensitive (recent divorce, probate, death) and soften the angle accordingly — these contacts should not receive market-opportunity framing; use relational, no-agenda framing instead
   • Any score that relies on data the agent did not actually have (the row should be marked "score incomplete, enrich and rerun")
   • Any contact where fair housing risk appears (e.g., targeting based on family composition framed commercially — always reframe around the home, not the household)

7. Output a working table and a pipeline summary. Produce:

   • A ranked table with columns: Rank, Name, Address, Intent Score, Band, Predicted Window, Outreach Angle (one sentence), Recommended First Touch, Flag
   • A pipeline summary: how many in each band, top 3 data gaps blocking better scoring, recommended 30-day action plan sized to the agent's stated capacity
   • A list of contacts to re-run after enrichment (those marked INCOMPLETE)

Critical rules:

• No fabricated equity, value, or comp numbers. If data is missing, mark the row, do not score it.
• Outreach recommendations must comply with TCPA, CAN-SPAM, and fair housing.
• Sensitive life events (death, divorce, probate) never receive opportunistic framing. A human agent, not an automated sequence, initiates that contact.
• Scoring is a prioritization tool, not a prediction guarantee. Communicate uncertainty in the output.
• Never output the full list with commercial framing for contacts flagged DNC, litigating with the agent, or explicitly asked for space.

Example Output

Pipeline summary:

• Contacts scored: 412
• Hot Seller (80–100): 7
• Warm Seller (60–79): 24
• Likely 6–12 mo (40–59): 68
• Long-term nurture (20–39): 183
• Cold (<20): 94
• Incomplete / needs enrichment: 36
• Capacity set at 25 new listing conversations → recommend working top 31 (all Hot + top 24 Warm)

Top 5 ranked contacts (excerpt):

Data gaps flagged:

• 84 contacts missing current estimated value — recommend a batch AVM pull
• 36 contacts missing purchase year — enrich from county records before scoring
• Behavioral signals absent for 61% of CRM — connect CRM page-tracking to enable the B component

Sensitivity flags:

• Contact #18 has a probate signal — pulled from automated outreach, routed to agent for a personal note with no sales ask
• Contact #41 on DNC list — phone recommendation suppressed; fallback to mail

30-day plan (capacity 25):

• Week 1: Work top 7 Hot contacts (1 call/day + handwritten note follow-up)
• Week 2: Work top 18 Warm contacts (batch emails M/W, calls F)
• Week 3: CMA prep and delivery to responders
• Week 4: Re-rank after enrichment run