Lumi.
Lumi · private beta · EU · LatAm · MENA

Your second brain
for closing deals.

Speak after a showing. Forward an email. Pull up a client. Lumi captures the soft signals, fills the brief, and feeds Claude — automatically.

  • Voice → CRM, auto. No forms.
  • Works offline. Syncs when you're back.
  • Free for agents in EU · LatAm · MENA.

9-min read · Updated April 2026

9:41

Lumi · Wednesday

Good morning, Niki.

Two showings · three leads need a nudge.

Clara Ruiz
Tomorrow 11am showing at Passeig de Gràcia 84 with Clara Ruiz. She wants to bring her partner.
Got it — creating the showing.
Suggested event · 92%

Showing · Passeig de Gràcia 84

Thu · 11:00–11:45Gràcia
What’s the HOA for Apt 4?
€210 per month, covers elevator, concierge, and rooftop.DOC 12
Ask Lumi or speak…
Calendar
Todos
Lumi
Clients
Settings
agent toolkit · field guide

80% of your leads
aren't real.

Most agents work their CRM as if every lead deserves equal attention. The result: they spend Mondays on leads that won't close for 18 months while the actual hot lead goes cold over the weekend. The 4-factor daily score ranks every lead 0-100 and tells the agent exactly which 20% to focus on.

9-min readUpdated April 2026Pack 27 of 30 · @lumi.estate
the four factors

What goes into the score.

Four factors, explicitly weighted. The weights are calibrated to typical residential dynamics; tweak with reason and trade weights against each other rather than adding new factors.

factor
weight
Intent stage
40%
Where the lead is in the buyer journey. Active showings + multiple touches > saved-searching > form-fill > cold prospect.
Window
25%
How soon they need to move. <90 days = high; 90-180 = medium; 180+ = low. Window collapses scores faster than any other factor.
Budget alignment
20%
Whether the agent's current inventory matches the lead's budget. Misalignment = wasted effort regardless of intent.
Channel responsiveness
15%
Whether the lead replies on the channel they signed up through. Non-responders drop fast; responsive replies elevate fast.
the protocol

Five rules. Daily recompute.

The discipline of lead scoring that actually changes the agent's behaviour. Each rule prevents a specific failure mode that turns scoring from useful into theatrical.

  1. 01

    Daily recompute. Not weekly. Not monthly.

    The single biggest mistake in lead scoring is treating it as a one-time tag. A lead's score must update every day based on the day's signals — replied to a message, attended a showing, went silent for 5 days. Weekly scoring lets cold leads accumulate as 'warm' for too long; monthly scoring is fiction. Daily scoring is the protocol's foundation.

  2. 02

    The 4 factors, weighted explicitly.

    Intent stage (40%), window (25%), budget alignment (20%), channel responsiveness (15%). These weights matter and are calibrated to typical residential-real-estate dynamics. Tweak only with reason — for instance, in luxury markets you might raise budget alignment to 30% and lower window to 20%. Don't add factors without dropping others.

  3. 03

    Cadence is dictated by tier. No exceptions.

    HOT leads get daily touches; WARM weekly; WATCH monthly; COLD quarterly; DEAD archived. The temptation is to give every lead 'special attention' — and the result is the agent burns out chasing leads that aren't real. The discipline of the cadence by tier is what creates the time to deliver hot-tier service to the leads that matter.

  4. 04

    The score is for the agent. The lead never sees it.

    Hard rule: lead scores never appear in client-facing artefacts. Never in a message ('Marina, you're our priority lead this week'), never in a CRM share with the seller, never in a referral note. The score is a prioritisation tool for the agent's day; surfacing it to the lead either flatters them inappropriately or insults them depending on the tier.

  5. 05

    DEAD ≠ deleted. Archive with re-engagement triggers.

    Leads scored DEAD (0-19) are archived from active cadence but never deleted. Six months later, the relationship-radar protocol may surface them with a life-event signal — and the brief from the original interaction is still there. The score gets them out of the day-to-day; the data preserves the option to re-engage when conditions change.

anti-patterns

Three failures that destroy the scoring system.

Each one is what scoring becomes when discipline slips. Each one has consumed weeks of agent time before the system was rebuilt without it.

the demographic proxy

rationale: 'Score 35 — Marina is single, professional, 32 — typical low-intent demographic.'

Scoring on demographic proxies (age, marital status, profession). Wrong, biased, and legally exposed in most jurisdictions. Score on behaviour and stated intent only — not on inferred attributes that correlate with anything.

the leaked score

[Email to lead]: 'Hi Marina! You're one of our top-tier leads — let me make sure you get priority service this week!'

Surfacing the tier to the client. Either patronises them ('top-tier' as flattery), distracts from the actual value, or invites them to ask why they're tier-rated at all. The score is for the agent's prioritisation. The lead never knows.

the auto-archive

[CRM action]: 'Lead Marina automatically archived after scoring DEAD for 14 consecutive days. No agent review.'

Auto-archive without agent review. If the score's 4 factors miscalibrated for a niche scenario (luxury client, multi-year buyer, referral pipeline), the agent loses the lead entirely. DEAD-tier triggers archive, but archive is a queue for review, not a delete.

the prompt that scores

What to feed Claude.

Haiku is sufficient — the scoring logic is well-defined and structural. Run nightly across all CRM leads; surface tier changes (boundary crossings) in the morning review.

score_system_prompt.md
You are a senior real-estate agent's
lead-scoring analyst.

INPUT
You receive: per lead, the CRM brief
including intent_stage, window
(earliest/latest move date),
budget_alignment with current
inventory, recent soft_signals (call
log + showing capture), days since
last touch, and channel_responsiveness.

OUTPUT
A single integer score 0-100 per lead,
with a 1-sentence rationale showing
the key factor.

  Tier mapping:
    80-100 — HOT. Active conversation,
             window <90 days, budget
             aligned, channel responsive.
    60-79  — WARM. Engaged but window
             >90d OR budget gap OR
             channel friction.
    40-59  — WATCH. Meaningful soft
             signals but no active
             conversation.
    20-39  — COLD. Old leads with
             little active engagement.
    0-19   — DEAD. Bad data, wrong
             stage, never replied.

Cadence-by-tier rule:
  HOT  → daily touch, voice/SMS
  WARM → weekly touch, channel
         of preference
  WATCH → monthly touch via market
          brief or relationship-radar
  COLD → quarterly via newsletter
  DEAD → archive

RULES (non-negotiable)
1. Score is daily-recomputed. A lead
   that didn't reply in 14 days drops
   tier even if intent_stage was hot.
2. The 4 factors weighted: intent
   (40%), window (25%), budget (20%),
   responsiveness (15%). Soft signals
   are tie-breakers.
3. Rationale names the dominant
   factor, never lists all four.
4. NEVER use the score directly in
   client-facing messages. The score
   is for the agent's prioritisation;
   the lead never knows their tier.
5. Re-tiering happens in tier
   boundaries (40, 60, 80). A 79
   stays warm; a 80 jumps to hot.

ANTI-PATTERNS (never produce these)
- Scoring on demographic proxies
  (age, gender, employer)
- Multi-paragraph rationales
- Auto-archiving DEAD without agent
  review
- Using the score as a confidence
  on a single message ('80% likely
  to reply')
Open Claude →

Run nightly via cron job. Surface tier-boundary crossings in the morning review (the weekly-review protocol consumes the score).

built around this exact 4-factor lead scoring

Scoring nightly is step one.
Honouring the cadence-by-tier is step two.

Lumi is the app that runs this workflow for you. You speak after a showing — Lumi captures the soft signals. You forward an email — Lumi updates the constraints. You open the app at 8am — the brief is already there, ready to feed Claude.

  • Voice → structured CRM, automatically
  • No forms. No data entry. No copy-paste.
  • Free for agents in EU · LatAm · MENA
9:41

Lumi · Wednesday

Good morning, Niki.

Two showings · three leads need a nudge.

Clara Ruiz
Tomorrow 11am showing at Passeig de Gràcia 84 with Clara Ruiz. She wants to bring her partner.
Got it — creating the showing.
Suggested event · 92%

Showing · Passeig de Gràcia 84

Thu · 11:00–11:45Gràcia
What’s the HOA for Apt 4?
€210 per month, covers elevator, concierge, and rooftop.DOC 12
Ask Lumi or speak…
Calendar
Todos
Lumi
Clients
Settings

A real-estate adaptation of the lead-scoring discipline from B2B sales tooling (HubSpot, Salesforce Einstein, Apollo). Our slice: the 4-factor daily score and the cadence-by-tier rule that protects the agent's attention from 80% of leads that aren't real.

More guides like this on @lumi.estate. Follow if any of this was useful — it's how we know to keep writing.