DataSift

Deal Analysis

AI-Powered Comping Workflow

Get appraiser-grade property valuations in minutes, not hours.

The Claude comping skill runs a 9-step analysis, auto-detects disclosure vs non-disclosure states, and generates a complete Excel report with comps, adjustments, and ARV. Here is how it works, tab by tab.

15 min read
The Problem

Comping Takes Too Long

Traditional comping eats 45-90 minutes per property. Pulling comps, adjusting for features, calculating price per square foot, checking market sentiment. Most operators either rush through it and miss adjustments, or skip it entirely and guess.

The Claude comping skill compresses that into a single conversation. Give it an address. It pulls public records, finds comparable sales, sorts them into two buckets, applies feature adjustments, checks market conditions, and delivers a complete analysis with a confidence-banded ARV range.

The output is a 7-tab Excel workbook you can hand to a partner, lender, or contractor. Not a screenshot of a Zillow page. A structured report with the math trail visible at every step.

What It Produces

7-tab Excel workbook: Executive Summary, Subject Property, Comparable Sales, Adjustments Detail, Market Analysis, ARV Calculation, Sources and Notes. Plus a full in-context breakdown.

What You Need

Claude Pro ($20/mo), a property address, and optionally your own closed deal data for market calibration. That is it.

This skill took roughly 25 iterations to build and was peer-reviewed against real closed deals. It is intentionally conservative. You want your ARV estimate to be the floor, not the ceiling. A high ARV that does not hold up kills your deal. A conservative ARV that comes in low leaves room for upside.

Download the Claude Comping Skill

Upload this .skill file to Claude Co-Work or Claude Projects to start comping properties instantly. Download from Google Drive

See all available skills → Claude Skills for REI

How to turn on the comping skill in Claude

Turning on the comping skill in Claude. Click to zoom.

Core Framework

The Two-Bucket Method

Every comp goes into one of two buckets based on condition. The gap between them tells you what the market pays for renovation.

Bucket A: Unrenovated

Dated finishes, original condition, deferred maintenance. Similar age, size, and layout to subject property.

Bucket B: Renovated

Fully updated, flipped, or clearly modernized to current market standard. Verified via photos, remarks, or permits.

Bucket A Details

Select comps with similar size/plan, average or below-average condition, sold within 90 days (6 months max), arm's-length sales only. Calculate the Median PPSF_A (price per square foot for unrenovated properties).

Bucket B Details

Select comps that are flips or clearly modernized. Verify via photos, remarks, and permits. Quality-adjust if one comp is ultra-lux beyond typical. Calculate the Median PPSF_B (price per square foot for renovated properties).

MARKET PREMIUM FORMULA

Market Premium (%) = (PPSF_B - PPSF_A) / PPSF_A x 100%
This is not a rehab cost estimate. The Market Premium is a market-derived metric. It tells you what buyers in this micro-market are willing to pay for updated finishes. Typical spread is 10-30%. Below 5% or above 30%, re-examine your comp selection. Something is off.
In the Joyce Ann example, the renovation premium was 20.4%. That means renovated homes in Harrison Township sell for roughly 20% more per square foot than dated ones. That is a healthy, reliable spread. If you see 5% or less, the market does not reward renovation enough to flip profitably. If you see 35%+, you may have miscategorized a comp.
State Routing

Disclosure vs Non-Disclosure

The skill auto-detects which framework to use based on property address. Disclosure states have visible sold prices. Non-disclosure states hide them, requiring a triangulation method.

Two-Bucket Method (Standard)

Sold prices are publicly recorded. The skill pulls actual sale prices from MLS, county records, and aggregators (Zillow, Redfin). Comps are sorted into Bucket A and Bucket B using real sold data.

Confidence Band

±2-5% on final ARV. Tighter range because you are working with actual sold prices.

This is the framework used in the 5532 Joyce Ann example. Ohio is a full disclosure state.

Triangulation Method

Sold prices are not publicly recorded. The skill must derive estimated sold prices (ESP) using multiple methods before it can apply the Two-Bucket analysis.

Method A: Last List Price + DOM

Find the last list price before "Pending." Under 7 DOM = list price or 101%. 7-30 DOM = 97-100%. Over 30 DOM = 90-95%.

Method B: Deed of Trust Calculation

Find the loan amount from public records. Conventional: Loan / 0.80. FHA: Loan / 0.965. VA: Loan / 1.00.

Method C: Tax Value Ratio (Sanity Check)

Compare assessed value to list price ratios for active homes. Apply multiplier. Use only as confirmation, never as primary method.

Confidence Band

±5-7% on final ARV. Wider range because you are working with derived prices, not actual sold data.

Non-Disclosure States:

TXUTWYNMIDMTNDAKKSMSLAMO
Real Example

The 5532 Joyce Ann Comp Report

A 1961 brick ranch in Dayton, OH. 3 bed, 2 bath, 1,892 sqft. Dated condition, cosmetic rehab needed. The skill produced a 7-tab report. Here is every tab, broken down.

Executive Summary tab showing subject property details, ARV analysis results, and market snapshot

Executive Summary tab. Click to zoom.

What This Tab Shows

The executive summary is your one-page overview. Three sections at a glance:

Subject Property: Address, property type (Single Family, Brick Ranch), 1,892 sqft, 3/2, built 1961. Confirms the skill pulled the right property.

ARV Analysis Results: Final ARV of $265,000 with a range of $252,000 to $278,000. Moderate confidence. Post-renovation price per square foot of $150.53.

Market Snapshot: Balanced market. Median sale price $225,000, median PPSF $131. Average 34 days on market. 98% sale-to-list ratio. This tells you the market is not distressed and not overheated.

Subject Property tab showing full property details including basic info, characteristics, condition, and ownership

Subject Property tab. Click to zoom.

What This Tab Shows

Every data point the skill gathered about the subject property, organized into four sections:

Basic Information: Full address, city (Dayton), state (OH), ZIP (45415), county (Montgomery), subdivision (Harrison Township).

Property Characteristics: 1,892 sqft living area, 13,068 sqft lot, 3 bed / 2 bath, 1961 build, 1 story, 2-car attached garage, no pool, unfinished basement (mechanical/storage).

Current Condition: Dated. Cosmetic rehab needed. Renovation scope to be determined in a separate workflow.

Ownership and Legal: Individual owner, residential zoning, no HOA. Clean.

Cross-check your GLA. The Sources tab flagged a discrepancy: some public records show 1,652 sqft vs the 1,892 sqft used here. If actual GLA is closer to 1,652, the ARV drops to roughly $240,000 and the wholesale price falls to around $195,000. Always verify GLA with the county assessor before making an offer.
Comparable Sales tab showing 6 comps with Two-Bucket analysis

Comparable Sales tab. Click to zoom.

6 Comps, Two Buckets

The skill found 6 comparable sales within 2 miles, all sold in the last 90 days, all 3/2 brick ranches built between 1955-1975. Here is the full breakdown:

#AddressDatePriceGLAPPSFConditionDist.Adj.Adj. ValueBucket
15802 Sparkhill DrDec 12$190,0001,520$125.00Partial Update1.2 mi+$22,000$212,000A
26407 Woodville DrNov 15$210,0001,686$124.56Average1.5 mi+$8,000$218,000A
36012 Imperial Hills DrJan 12$208,7501,669$125.07Good/Maintained1.8 mi+$9,000$217,750A
46430 Oakhurst PlDec 19$240,0001,657$144.84Dated1.6 mi+$10,000$250,000A
5701 Fredericksburg DrDec 29$239,9001,464$163.87Updated0.8 mi+$17,000$256,900B
66029 Imperial Hills DrDec 29$210,0001,411$148.83Good1.8 mi+$19,000$229,000B

TWO-BUCKET RESULTS

Bucket A (4 comps): Median PPSF $125.03 | Avg $129.87

Bucket B (2 comps): Median PPSF $156.35 | Avg $156.35

Renovation Premium: 20.4%
Adjustments Breakdown tab showing line-by-line adjustments for each comp

Adjustments Detail tab. Click to zoom.

Line-by-Line Adjustments

Every comp gets adjusted to make it comparable to the subject property. The skill applied three types of adjustments across the 6 comps:

GLA Adjustments

+$78,000 total. The subject (1,892 sqft) was larger than every comp. Comp 6 was 481 sqft smaller, earning the biggest adjustment (+$19,000).

Condition Adjustments

-$3,000 total. One comp (5802 Sparkhill) had newer roof and windows than the subject. The adjustment accounts for that advantage.

Basement Adjustments

+$10,000 total. The subject has an unfinished basement. Comp 1 (5802 Sparkhill) only had a crawl space. Adjustment reflects the storage and expansion value.

TOTAL ADJUSTMENTS ACROSS ALL COMPS

GLA +$78,000 | Condition -$3,000 | Basement +$10,000 = Net +$85,000
Market Analysis tab showing market metrics, trends, and local notes

Market Analysis tab. Click to zoom.

Market Health Check

Before the skill calculates ARV, it reads the market. The Market Analysis tab has three sections:

Market Metrics

Balanced market. Median price $225,000, PPSF $131. 34 avg DOM. 98% sale-to-list ratio. 45 active listings, 22 pending, 3.5 months of inventory.

Market Trends

3-month: Stable (+1.2%). 6-month: Slight appreciation (+3.8%). DOM stable at 30-40 days. Inventory balanced.

Local color matters. The market notes section captured that Huber Heights / Harrison Township is known as the world's largest community of brick homes. Average home value around $204K, up 3.8% year-over-year. It also flagged that the subject has an R-22 A/C condenser that will need full replacement (R-22 refrigerant is phased out), and the new furnace (2024) is a strong selling point. These details affect both your rehab budget and your marketing strategy post-renovation.
ARV Calculation tab showing 9-step math from base PPSF to final ARV

ARV Calculation tab. Click to zoom.

The 9-Step ARV Math

This is where the skill shows its work. Every number traces back to a specific data point. No black boxes.

Base PPSF (Unrenovated)
$125.03
Renovation Premium
20.4%
Post-Reno PPSF
$150.53
Market Sentiment Adjustment
-2.0%
Adjusted PPSF
$147.52
Subject GLA
1,892 sqft
Base ARV (PPSF x GLA)
$279,108
Feature Adjustments
-$14,000
FINAL ARV
$265,000

Confidence Analysis

Moderate confidence. ±5.0% band. ARV Low: $252,000. ARV High: $278,000.

Why -2% Sentiment?

Balanced market with 98% sale-to-list ratio. Not hot enough for a positive bump, but the slight positive trend (+1.2% 3mo) prevents a deeper cut.

Sources and Notes tab showing data sources, search parameters, and recommendations

Sources and Notes tab. Click to zoom.

The Paper Trail

Every good comp report is auditable. This tab documents where the data came from and what constraints were applied.

Data Sources

Zillow, Sibcy Cline MLS, Coldwell Banker MLS, ATTOM Data, Montgomery County Auditor, Homes.com, and subject property due diligence photos.

Search Parameters

90-day window (Dec 2025 - Feb 2026). 2-mile radius. GLA range 1,400-2,150 sqft. Year built 1955-1975. Harrison Twp / Butler Twp / North Dayton.

Key Recommendations

The skill flagged several items for verification before making an offer:

  • Verify GLA with county assessor. Records show 1,652 vs 1,892 sqft discrepancy.
  • Get actual basement square footage and finish level. Unfinished adds $5-10K, finished could add $15-25K.
  • R-22 A/C condenser needs full replacement (refrigerant phased out).
  • Electrical panel should be evaluated (original 1959 panel).
  • Run title search. Check permit history for unpermitted additions.
  • Wholesale price target: $205,000-$215,000. Selling points: new furnace, 1,892 sqft, sunroom, fenced yard.
Your comps are only as good as the data behind them. DataSift's property records and SiftMap polygon search give you the transaction data the Claude skill needs. Create Your Account →
Reference

Adjustment Cheatsheet

The skill uses these ranges when adjusting comps. Values shift by price tier and climate. Expand each category to see the full table.

FeatureUnder $500KOver $500K
Bedroom+$10,000+$25,000
Full Bath±$10,000±$10,000
Half Bath±$5,000±$5,000
FeatureStandardHot/Cold Climate
Garage$10,000-$15,000$20,000-$25,000
Carport$5,000-$7,500$10,000

Climate note: Use high end in very hot (AZ, NV, TX) or very cold (IL, MN, WI) markets. Hail-prone areas (TX, OK, CO) make garage vs carport a major differentiator.

IssueUnder $500KOver $500K
Backing busy road-$10,000-10% to -15%
Fronting busy road-$20,000-20%
Commercial adjacency-$10K to -$20K-10% to -15%
FeatureUnder $500KOver $500K
Pool (hot climate)+$20K to +$40K+$20K to +$40K
Pool (cold climate)+$5K to +$15KNegative possible
Extra 5,000 sqft lot$5K-$10K$30K-$50K
Water view+$20K-$50K+$50K-$150K

Basements are not counted as GLA by appraisers. Value depends on finish level:

Finish LevelValue (% of Above-Grade PPSF)
Finished to same quality~50%
Finished with drop ceilings~35-40%
Partially finished~25-35%
Unfinished~10-15%

ADUs: Not separately deeded = ~50% of equivalent value. Separately deeded/titled = 100% at local PPSF.

IssueAdjustment
Minor cracks (cosmetic)-$5,000 to -$10,000
Moderate issues-$15,000 to -$25,000
Major repair needed-$25,000 to -$40,000+
Previous repair (documented)-$5,000 to -$10,000
Comp Boundaries

Draw Your Polygon in SiftMap

The comping skill works with any address. But if you want to constrain where it looks for comps, draw a polygon in SiftMap first.

SiftMap polygon drawn around Harrison Township area with investor transaction filter showing recent sales

SiftMap polygon with investor transaction filter. Click to zoom.

Why Draw a Polygon?

The skill's default search uses a radius from the subject property. That works in most markets. But in block-by-block markets where values shift street to street, a polygon gives you precision. You are telling the skill: "Only look inside these boundaries."

The golden rule of comping: do not cross major roads. Thick yellow lines on the map are value boundaries. The polygon lets you enforce that.

Investor Transaction Filter

Toggle the investor transaction filter in SiftMap to see what investors are actually paying in the area. These are real acquisition prices, not retail sales. Gives you ground truth for your wholesale price target.

Edit to Refine

The default polygon is accurate enough for most markets universally. But you can edit it to focus on a specific sub-area for more refined results. Narrow the polygon to 3-4 blocks around your subject for hyper-local comps.

Most operators never need to edit the polygon. The default radius-based search pulls solid comps in any market. But if you are working a neighborhood where values change block by block (older urban areas, mixed-use borders, waterfront transitions), tightening the polygon is worth the 30 seconds it takes.
Market Calibration

Customize with Your Own Deals

The default adjustment tables are national averages. Your closed deals are ground truth for your specific market. Feed them to the skill and it calibrates automatically.

Claude Co-Work skill editing interface showing the comping skill configuration

Editing the comping skill in Claude Co-Work. Click to zoom.

How It Works

Open the comping skill in Claude Co-Work (or Claude Projects). Add 3-5 properties you have closed in your target market. For each property, include:

  • Property address and basic specs (beds, baths, sqft, year built)
  • Your purchase price (what you actually paid)
  • Rehab cost (actual spend, not estimate)
  • Final sold price or appraised ARV (what it actually sold for or appraised at)
  • Any notable adjustments (pool added, garage converted, foundation repaired)

The skill uses your closed deals to recalibrate its adjustment ranges, market sentiment assumptions, and confidence bands for your area. Instead of using national averages for "garage adds $10-15K," it learns that in your market, a 2-car garage adds $18K based on your actual data.

Draft Your Calibration Prompt

Use the field below to draft your calibration data. Include 3-5 closed deals with the details listed above.

Do This

Feed real closed deals with exact numbers: purchase price, rehab spend, final sold price.

Include 3-5 properties from the same market or sub-market.

Note any unusual features (pool added, foundation repair, lot split).

Not This

Do not just tell Claude "my market is hot" without data to back it up.

Do not use deals from different cities or states as calibration data.

Do not mix flips with buy-and-hold rentals. Different exit strategies, different comps.

You do not need to re-enter your calibration data every time you comp a property. Save it inside the skill's project or Co-Work space. Claude remembers it across conversations within that project. Three solid closed deals from your market will dramatically improve accuracy for every future comp in that area.
Glossary

Key Terms

Click any card to flip and see the definition.

ARV

Click to flip

After-Repair Value

The estimated market value of a property after all renovations are complete. Calculated by applying the renovated PPSF to the subject's GLA, adjusted for market sentiment and feature differences.

PPSF

Click to flip

Price Per Square Foot

Sale price divided by gross living area (GLA). The foundational metric for comparing properties of different sizes. Always compare PPSF within the same micro-market.

Two-Bucket Method

Click to flip

Comp Sorting Framework

Sort comps into Bucket A (unrenovated/dated) and Bucket B (renovated/flipped). The gap between median PPSF values reveals the market premium for renovation in that micro-market.

Market Premium

Click to flip

Renovation Spread

The percentage difference between renovated and unrenovated PPSF. Formula: (PPSF_B - PPSF_A) / PPSF_A x 100%. Typical range: 10-30%. Not a rehab cost estimate.

GLA

Click to flip

Gross Living Area

Above-grade living space measured in square feet. Basements are excluded from GLA. This is the number used for PPSF calculations and comp matching.

Comp Set Tightening

Click to flip

Filter Constraints

Hard filters applied before selecting comps: 90 days max age, same subdivision, ±250 sqft GLA, ±10 years build, matching elevation style. Do not cross major roads.

Price Triangulation

Click to flip

Non-Disclosure Method

Used in 12 states where sold prices are hidden. Derives estimated sold prices using Last List Price + DOM logic, Deed of Trust reverse math, and Tax Value ratios as a sanity check.

Confidence Band

Click to flip

ARV Range Indicator

The margin of error around the final ARV. Disclosure states: ±2-5%. Non-disclosure states: ±5-7%. Wider bands mean more uncertainty in the underlying data.

Feature Adjustments

Click to flip

Comp Normalization

Dollar or percentage adjustments applied to comps to account for differences from the subject property. Categories: GLA, bedrooms, bathrooms, garage, pool, lot size, traffic, condition, basement.

Disclosure State

Click to flip

Public Sale Prices

States where sold prices are publicly recorded (38 states + DC). The skill uses the standard Two-Bucket method with actual sold data. Ohio, California, Florida, and most states are disclosure states.

Knowledge Check

Test Your Understanding

7 questions on the comping framework. No trick questions. No case study recall.

1. What does the Two-Bucket method separate comps into?

2. What does the Market Premium measure?

3. How does the skill handle non-disclosure states?

4. What is the preferred maximum age for comparable sales?

5. Why would you draw a polygon in SiftMap?

6. What is the best way to calibrate the skill for your market?

7. What confidence band applies to non-disclosure states?

Book Your Free Strategy Call 30 min. No obligation.