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
Lumi · Wednesday
Good morning, Niki.
Two showings · three leads need a nudge.
Showing · Passeig de Gràcia 84
I know my March
before March starts.
Most agents forecast by gut feel — “feels like a good month coming”. Gut forecasts are biased toward optimism and almost never bias-corrected. The protocol replaces gut with explicit pipeline math: stage-conditional close-rates × time-in-stage × deal value. The output is a calibrated range that's usable for actual finance planning.
Five rules. Calibrated forecast.
The discipline of pipeline forecasting that matches reality. Each rule prevents one of the failure modes that turns forecasts from useful into theatrical.
- 01
Stage-conditional close-rates beat overall close-rates.
An agent's '14% close-rate' is meaningless when applied uniformly to every pipeline deal. A deal in initial-contact stage has maybe a 4% chance of closing in the next 30 days; a deal under contract has a 75% chance. Stage-conditional rates compute these separately and produce a forecast that respects where each deal actually sits.
- 02
Time-in-stage is the most important secondary signal.
If a deal has been in 'negotiating' for 90 days when the historical median is 18, that deal is dying — even if it hasn't formally fallen through. Time-in-stage signals which deals to flag as at-risk, and adjusts the close-probability for each. Without this, the forecast over-counts long-stalled deals that look active but won't close.
- 03
The output is a range, not a number.
Single-number forecasts ('I'll close 4 deals in March') are wrong by definition — a single number can't capture the variance in real pipelines. The protocol's forecast is P25-P50-P75: 'best estimate 4 deals, likely range 3-6'. The range is what makes the forecast usable for actual planning; the single number is just a guess wearing better clothes.
- 04
Calibration delta is non-negotiable. Honesty is the value.
Every monthly forecast includes the previous period's predicted-vs-actual delta. If last month forecast 5 deals and actually closed 3, the current forecast either explains that gap or applies the bias correction. This honesty is what separates the protocol from agent-flattering forecasts that consistently over-predict and never get audited.
- 05
The forecast feeds into finance planning, not optimism.
The agent's actual use case for the forecast is concrete: 'Can I afford to take June off?' 'Should I onboard an assistant?' 'Is the new marketing budget reasonable against expected GCI?'. These decisions need calibrated bands. An aspirational forecast leads to over-spending; an honest forecast leads to good calls. The protocol's discipline is what makes the forecast usable for these calls.
Three forecasts that mislead the agent.
Each one is a real failure mode that turns the protocol from useful into a comforting story. The protocol's strictness — bands, calibration, pipeline-only — prevents each.
“March forecast: 8 deals, €120k GCI.”
Single number, no range, no calibration. The agent reads this and plans for 8 — and is shocked when March closes 4 deals. Without P25-P75, the forecast is a flattering guess. The honest version: 'P25 4 deals, P50 6, P75 8' is what the agent should plan against.
“[Each month for 6 months: predicted 6, actual 3-4. Forecast for next month: 6 again.]”
Six months of consistent 50% over-prediction. The protocol must surface and correct this — bias-adjusted forecasts move toward 3-4 deals. Hiding the calibration delta lets the agent over-spend monthly because the forecast keeps lying to them in the same direction.
“[Forecast includes 12 'pipeline' deals, 4 of which are vaguely-engaged leads with no active conversation in 60+ days.]”
Padding the pipeline with maybes inflates the forecast. The 4 zombie leads contribute close probability that won't materialise. The discipline: forecast on actual pipeline only — leads in active conversation, with stage and time-in-stage data.
What to feed Claude.
Sonnet recommended for the calibration reasoning and bias-correction logic. Run monthly on the 25th for the next-month forecast; quarterly for the next-quarter forecast.
You are a senior real-estate agent's
pipeline-forecast analyst.
INPUT
You receive: every active deal in the
agent's pipeline with stage (initial
contact / showing / negotiating / under
contract / closed-pending), days in
that stage, deal value, and the agent's
historical close-rate per stage from
the last 12 months (computed from
closed-vs-stalled cohort data).
OUTPUT
A monthly forecast object:
projected_close_count:
Best estimate (P50) + range
(P25-P75). Integer count.
projected_gci:
Best estimate (P50) + range
in agent's currency. Computed
from average commission per
closed deal in the same band.
high_confidence_deals:
Names of 2-5 deals likely to
close in this window with
confidence reasoning.
at_risk_deals:
Names of 1-3 deals stalling
(time-in-stage above historical
median) with intervention
suggestions.
calibration_note:
Last quarter's predicted vs
actual delta. Honest report.
RULES (non-negotiable)
1. The forecast is a range, never a
single number. P25-P75 is the
honest band; outliers happen.
2. Stage-conditional close-rates
matter more than overall close-
rate. A 12% overall close-rate
on negotiating-stage deals
becomes a 65% close-rate on
under-contract deals.
3. Time-in-stage is the early-warning
signal for stalled deals. A deal
that's been negotiating for 3×
the historical median is dying.
4. Calibration: the forecast must
include last period's
prediction-vs-actual delta. If
you over-forecast by 30% last
month, this month's forecast
accounts for that bias.
5. Never include 'maybe' deals
(vaguely-engaged leads with no
active conversation). Forecast
on pipeline only.
ANTI-PATTERNS (never produce these)
- Single-number forecasts
- Aspirational forecasts (the
number you want to hit, not the
number the data supports)
- Including hot leads not yet in
pipeline (they're the next
forecast, not this one)
- Hiding the calibration delta
(the honesty is what makes the
protocol useful)Run monthly on the 25th for next-month forecast. Compute historical close-rates and stage-conditional rates from CRM data; feed both as input.
Forecasting in ranges is step one.
Honouring the calibration delta 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
Lumi · Wednesday
Good morning, Niki.
Two showings · three leads need a nudge.
Showing · Passeig de Gràcia 84
Pipeline
Active
8
Warm
4
Cold
2
Clara Ruiz
Active€1.8M · 3BR
Passeig de Gràcia showing · 11:30
Andreas Moreno
Active€2.4M · 4BR
Send comps by 18:00
Dimitri Schneider
Warm€900K · 2BR
Contract review today
Silent 3d · last 3 days ago
Sarah Mitchell
Cold€1.2M · 3BR
Draft re-engagement
Silent 9d · last 9 days ago
A real-estate adaptation of the sales-pipeline forecasting discipline from B2B (Clari, Gong, Aviso). Our slice: the individual agent's monthly close + GCI range, calibrated against last quarter's prediction-vs-actual delta.
More guides like this on @lumi.estate. Follow if any of this was useful — it's how we know to keep writing.