It is tempting to treat comps as a search result. Pull the recent sales nearby, look at the average price per square foot, multiply. It produces a number fast, and it's wrong often enough to be dangerous.
Appraisers do something different, and it isn't more complicated — it's just more disciplined. They select fewer properties, adjust each one deliberately, and treat the result as evidence rather than an answer. Here's how that works, and how to apply it to a comp set someone hands you.
Selection is most of the job
A comp set is only as good as what's in it, and adding more properties makes a bad set worse rather than better. The screening priorities, roughly in order:
- Proximity. Same neighborhood, ideally same subdivision or the same side of a dividing feature. Value can change materially across a main road, a school boundary, or a rail line — physical distance understates how different two properties are.
- Recency. The more recent the sale, the less you're adjusting for what the market did in between. Older sales aren't useless, but every month adds an assumption.
- Similarity of the basics. Comparable size, bed and bath count, lot, age, and construction type. A ranch and a two-story of identical square footage are not interchangeable.
- Same market segment. A finished flip sold to an owner-occupant is not comparable to an as-is sale to an investor, even next door. Different buyers, different price.
- Arm's-length transactions. Family transfers, estate sales handled quickly, and foreclosure sales often trade below market. They tell you something, but not what a normal sale is worth.
Three tight comps beat twelve loose ones. If you need a wide radius or a long time window to find enough sales, that itself is information — it means the property is unusual, the area is thin, or both, and your value estimate deserves a wider range.
The adjustment idea
No two properties are identical, so raw sale prices aren't comparable on their own. The appraiser's move is to adjust each comp toward the subject: if the comp has something the subject lacks, subtract its contributory value; if the subject has something the comp lacks, add it.
The adjustment is always applied to the comp, never to the subject — you're answering "what would this comp have sold for if it were more like my property."
The categories that typically drive adjustments:
- Living area — differences in size between the subject and the comp.
- Condition and updates — a renovated kitchen and a 1980s kitchen are a real price difference, and one that is easy to overlook.
- Date of sale — the market moved between then and now, in one direction or the other.
- Location — the busy street, the backing-onto-commercial lot, the cul-de-sac premium.
- Site size, where lot value varies meaningfully in that market.
- Garage, basement finish, bed/bath count, and major systems — age of roof and HVAC included.
The contributory value of a feature is what the market pays for it — which is rarely what it cost to build. A $60,000 pool does not add $60,000, and in some markets it adds nothing at all.
Where adjustment values come from
The honest answer to "how much is a second bathroom worth" is: whatever this market pays for one. Appraisers derive that from evidence rather than assumption, and the classic technique is paired-sales analysis — find two sales that are similar in every meaningful respect except the feature in question, and the price difference indicates what that feature contributes.
Clean pairs are rare in practice, so the technique is often approximate. But the discipline behind it is the point: an adjustment you can't tie to market evidence is a preference, and preferences are how a comp set drifts toward the number you were hoping for.
This is also why national rules of thumb mislead. What a finished basement contributes varies enormously by region and by market segment, and a figure that's right in one place is badly wrong somewhere else.
Reading a comp set someone hands you
Whether it comes from an agent, a partner, or software, these are the checks worth running:
- Look at the spread, not the average. If adjusted values cluster tightly, the evidence agrees and you can trust the middle. If they scatter, the comps are telling you they don't know — and a single averaged number hides that entirely.
- Check the adjustment sizes. Large gross adjustments mean the comps weren't very comparable to begin with. A set requiring heavy adjustment is a weak set even if the final numbers look tidy.
- Ask what got excluded. A comp set is a selection, and selections can be shaped. The sale that didn't make the list is sometimes the most informative one.
- Verify condition, not just specifications. Records show bed count and square footage. They rarely show that a comp sold gutted, and that single fact can move a valuation more than every other adjustment combined.
- Watch for the same comp used differently. If a property appears in both an as-is and an after-repair comp set, something is wrong with at least one of them.
The two comp sets investors actually need
This is where investor comping diverges from a standard appraisal. You typically need two answers, and they come from two different sets of sales:
As-is value is supported by sales of properties in similar condition — other unrenovated houses, often investor purchases. After-repair value is supported by sales of finished, updated properties sold to retail buyers.
Mixing them is an easy comping error to make, and it's expensive in both directions: renovated comps inflate your as-is number, and as-is comps deflate your ARV until every deal looks bad. See calculating ARV step by step and how that number feeds your offer.
When the comps just aren't there
Rural properties, unusual layouts, converted uses, and thin markets all produce comp sets that can't carry a confident number. The right response is to widen your range and lower your offer to account for the uncertainty — not to reach for distant sales and pretend the evidence is stronger than it is.
An honest "I don't know within $40,000" is worth more than a confident number built on comps that don't support it.
Doing this at volume
Applying this discipline by hand on every lead is slow, and not every property deserves it. The practical approach is a consistent first pass on everything, then this level of scrutiny on the ones that survive.
Appraize shows the comps behind every valuation so you can inspect the selection, disagree with it, and adjust — which is the part that matters here. A comp set you can't examine is one you're trusting rather than reading.
The short version
Choose few, choose close, choose recent, choose similar. Adjust toward the subject, and tie every adjustment to something the market actually showed you. Read the spread rather than the average. Keep as-is and after-repair comps in separate sets. And when the evidence is thin, say so in your offer instead of in your assumptions.
