Comparative analysis: closed sales, not sellers' wishes

Three levels of numbers on the market

There is the price an owner wants. There is the asking price, which has already passed through an agent. And there is the price at which the property actually changed hands. On most markets the gap between the first and the third is wide, yet public analytics are built on the second — asking prices, where reasonable properties sit next to ones listed for a year. Comparative analysis only makes sense on closed sales.

Selecting comparable properties

Recency. The faster the market moves, the shorter your window. Sales older than six months describe a different market than yours.
Geography. Same block, same transport access, same school catchment. A building across a main road is a different market even if it is three hundred metres away.
Type and size. Area within a sensible band, the same building type, a comparable floor.
Condition. An unrenovated unit and a finished one are two separate lines and must not be averaged together.

Three to five good comparables give a truer picture than twenty random ones. Then come the adjustments: for floor, view, condition, area. An adjustment is not a guess — it answers how much a buyer will refuse to add because of that factor.

Comparables table
| Address | Area | Floor | Condition | Sale date | Sale price | Price per unit of area | Adjustments | Adjusted price |
Result: range from __ to __ , median __ , days on market for the comparables __

Insight. Days on market for your comparables tells you more than their average price. If similar properties went in three weeks and yours has been listed for six, the problem is not the market and not the photos.

Where AI fits

AI calculates adjustments, assembles the table, explains the logic of the calculation and writes a clear text for the client on why the number is what it is. You supply the data. Ask it to find the sales and it will produce plausible addresses and amounts that do not exist, delivered in a confident tone.

Common mistake. Pulling prices from open listings and calling it comparative analysis. That measures sellers' intentions: properties that will never sell at their asking price pull the median up and confirm your client's inflated expectations.

Pro tip. Bring the seller a table, not a number. Someone looking at five dated sales argues with the market rather than with you, and the pricing conversation gets weeks shorter.

Cheat sheet

Work from closed sales, never from asking prices.
Three to five precise comparables beat twenty random ones.
Days on market for comparables is the key sanity check on price.
AI computes and explains but does not supply sales data.

1. Why does analysis based on asking prices distort the picture?

2. Which indicator from comparables signals that your price is too high?

3. What does AI do in a comparative analysis?

Task — checked by AI

Take a property you currently have listed and build a table of three to five closed sales for it: address, area, floor, condition, date, sale price, price per unit of area. Apply adjustments and produce a price range with a median. Note days on market for each comparable. Then write how the resulting range compares with the price your seller wants.

← Back

🔒 Answer the question correctly to move on to the next lesson.