KPIs & Reports

Two sides of one machine. THE DEAL ENGINE runs continuously — ingesting markets, scoring properties, surfacing deals. KPIs are the instrument panel that tells us the engine is sound. Reports are the products it emits — the surfaces an investor, a partner, or an operator actually touches.

KPIs prove it works. Reports are what it produces.

AVM err

the hero KPI — median % the model misses actual sales by. The number that proves the edge is real.

realized vs predicted

did funded deals perform the way we said they would? The honesty check that keeps us honest.

time-to-surface

raw data → ranked, underwritten deal. Speed is edge.

The point: Anyone can claim they "find deals." Almost nobody reports whether their model was right. We do — that scoreboard is the moat.

Part 1 — KPIs: the instrument panel

We track the engine across four buckets. Three of them are table stakes. The third — Model Accuracy — is the one almost no "deal finder" tool will show you, because most of them have no idea whether their numbers are any good.

Bucket 1 — Funnel Health

Is the engine fed, and is it moving? Coverage and throughput from raw market to ranked deal.

KPI	What it measures
Market coverage %	Share of the addressable market we actually see (parcels, listings, off-market)
Properties ingested	Raw volume flowing into the engine
Leads surfaced	Properties the engine flags as worth a look
Qualified opportunities	Leads that clear the buy-box and underwrite
Hit-rate	Qualified ÷ screened — signal density of the funnel
Time-to-surface	Data landing → ranked deal in your hands

Bucket 2 — Deal Quality

Are the deals the engine surfaces actually good? Pure underwriting economics, per opportunity.

KPI	What it measures
Cap rate	Net operating income ÷ price
Cash-on-cash	Annual cash return ÷ cash invested
DSCR	Net operating income ÷ debt service — can it carry the loan
Discount-to-value	Price vs model value — the margin we're buying
Projected IRR	Whole-hold return, time-weighted
Rent-to-price	Gross yield sanity check
Rehab estimate	Capital needed to stabilize

Bucket 3 — ⭐ Model Accuracy (this is the one that proves the edge)

Read this twice: Most "deal finder" tools never report this bucket. They'll show you a cap rate and a "score" and ask you to trust it. We measure whether the model was actually right — against real sales, real rents, real owner behavior, real performance. This bucket is the data edge. Everything else is downstream of it.

KPI	What it measures
AVM error	Median % our value estimate misses actual sale prices by — lower is the whole game
Rent error	Median % our rent estimate misses realized rents by
Motivation-score precision/recall	Of owners we flagged as likely to sell — how many did? How many sellers did we miss?
Deal-score calibration	Did high-scored deals actually outperform low-scored ones? A score nobody calibrates is a vibe.
Mispricing precision	Of listings we tagged "underpriced" — what share truly traded below value?

Why the third bucket is the moat

Score a deal → Deal closes / rents / sells → Compare prediction vs reality → Feed the miss back into the model

Fig. 1 — A competitor's "score" stops at step one. Ours completes the loop. Each closed deal makes the next prediction sharper — accuracy compounds where a static rules engine flatlines.

Bucket 4 — Outcome / Business

Does the engine produce money and learn from results? Where intelligence meets the P&L.

KPI	What it measures
Stage conversion	Sourced → underwritten → funded → closed, rate at each step
Realized-vs-predicted	Actual deal performance vs the model's projection — model drift, made visible
Cost-per-qualified-lead	Engine spend ÷ qualified opportunities
Offer win-rate	Offers made → offers accepted
Investor return vs projection	What partners actually earned vs what we told them they'd earn

Part 2 — Reports: what the engine emits

KPIs face inward — they tell us the engine is sound. Reports face outward — they're the deliverables. Each one is a packaged surface built from the same scored, underwritten data.

Report	What it is	Who it's for
The Deal Feed (flagship)	Daily ranked top-N opportunities — each with score, full underwriting, and the WHY behind the rank	Investors, partners, operators — anyone hunting
Deal Tearsheet	Per-property one-pager: comps, rent analysis, underwriting, rehab, owner/motivation, risks, recommendation	The thing you hand an investor or partner to make a call
Market Report	Opportunity index per metro/zip + trends — where the alpha is moving	Anyone deciding where to hunt next
Pipeline Report	The funnel + conversion at each stage, end to end	Operators, capital partners watching throughput
Model Health Report (internal)	Accuracy, calibration, drift alerts — the Bucket-3 KPIs, packaged	Us — keeps the edge honest
Outcomes Report	Realized vs predicted across funded deals — feeds the learning loop → following-the-money	Investors (transparency) + the engine (it learns)
Off-Market Lead Lists	Ranked motivated-seller lists, built for outreach	Acquisition / outreach operators
Alerts	Real-time: new mispriced listing, price drop, new distress filing — anything crossing the buy-box	Everyone, the moment it matters

One engine, two faces

Scored, underwritten data → KPIs (inward: is it sound?) → Reports (outward: here's the product)

Fig. 2 — The same scored data lands as both the scoreboard and the shipped surface. The Deal Feed and the Tearsheet are what you see; the Model Health and Outcomes reports are what keep them trustworthy.

The honesty checks

Two reports exist purely to keep the data edge real:

Model Health Report packages Bucket 3 — AVM error, rent error, calibration, drift alerts. If accuracy slips, this is where we catch it before a partner does.
Outcomes Report closes the loop with realized-vs-predicted across every funded deal. A model that never checks its predictions against reality isn't intelligence — it's a confident guess. This report is what makes the-intelligence-layer learn instead of merely score.

Both feed the learning loop. Every closed deal is a labeled example: prediction in, reality out, error measured, model corrected. That's the compounding mechanism — and it's traced end-to-end in following-the-money.

The takeaway: KPIs are the instrument panel; reports are the output. Model Accuracy and Realized-vs-Predicted are the two gauges most competitors don't even have — and they're exactly the ones that prove our deals are priced on a real edge, not a story.

Where this sits

The Deal Engine — the-deal-engine · the-intelligence-layer · signals-and-data · the-plays

See also — following-the-money · Research Pipeline · the-role-marketplace