"How much does a pool add?"
I get some version of that most weeks. A pool, a second bathroom, a fourth bedroom, a garage conversion, a kitchen, a shop out back, an extra acre. Or the version agents ask, which is the same question wearing different clothes: what's the price per square foot around here. Sometimes it's an owner deciding whether to spend the money. Sometimes it's an owner who already spent it and is about to list. Sometimes it's an agent trying to price something, or to argue with an appraisal that came in low.
The honest answer is that nobody can give you a dollar figure that's good anywhere except the specific house, in the specific neighborhood, at the specific price point you're asking about — and that the figure changes when any one of those changes.
That sounds like a dodge. It isn't. Once you understand why the number moves, you can reason about your own situation well enough to make a good decision, and you'll be able to tell the difference between an answer that's supported and an answer that somebody made up.
This is where most of the confusion lives, and it is worth being precise.
Cost is what you paid to have the thing built. Materials, labor, permits, the contractor's margin.
Price is what somebody actually paid for a house on a particular day.
Value is what a typical buyer would pay, given a reasonable marketing period and no unusual pressure on either side.
Say you spend sixty thousand dollars on a pool. That is a cost. It tells you what a pool takes to build in the current labor and materials market. It does not tell you what a buyer will pay more for your house because the pool is there — and that second number is the only one that matters when you sell.
The gap between them is not a failure of the market or the appraiser. Those are simply two different questions. Cost looks backward at what was spent. Value looks forward at what someone will pay.
In appraisal the second number has a name: contributory value — what a feature contributes to the value of the whole property. It's the number every "how much does it add" question is really asking about, and it is almost never the same as cost.
When an appraiser compares your house to a sale down the street and the sale has a three-car garage while yours has two, the appraisal shows an adjustment. Most people assume that number came from a schedule. It didn't.
An adjustment is a measurement, extracted from the market. The cleanest way to get one is a paired sale: find two sales as alike as you can — same neighborhood, same era, same condition, same size, sold close enough in time that the market hadn't moved — where the meaningful difference is the one feature you're trying to price. The difference in what they sold for is evidence of what that feature contributed.
Do that once and you have an anecdote. Do it several times and a range starts to appear.
Here's the part people don't expect: the arithmetic is trivial and the pairs are the hard part. Houses differ in a dozen ways at once. A true pair — alike in everything that matters, different in one thing — is rare, and in a neighborhood with few sales it may not exist at all. Most of the work in supporting an adjustment is finding defensible pairs, or finding another credible way to get at the same answer when the pairs aren't there.
Which is why an adjustment that arrives without any of that work behind it is just somebody's opinion wearing a dollar sign.
Paired sales are the cleanest evidence, not the only evidence. In a neighborhood that turns over twice a year, waiting for a clean pair means waiting forever. There are several other supportable routes, and a competent appraisal will tell you which one was used.
Grouped data and sensitivity analysis. Instead of two sales, take a larger set and sort it by the feature. Do the sales with the feature cluster higher on price per square foot than the ones without, once you account for the obvious differences? It's blunter than a matched pair, but with enough sales the pattern is real, and it has the advantage of not resting on any single transaction.
Statistical analysis. With a large enough and clean enough dataset, you can model several variables at once and let the data separate the effect of the feature from everything correlated with it. This is powerful in an active subdivision with hundreds of similar sales. It falls apart quickly in a neighborhood with thirty transactions and no two houses alike, which describes a lot of Fort Worth.
Depreciated cost. What the feature costs to build new, less the wear it has taken, can support a contributory value — particularly for something recently added, and particularly when there's no market data at all. But this is the method that has to be handled most carefully, because it is exactly the substitution the first half of this article warns about. Cost is a starting point to be tested against the market, never a conclusion on its own. If the market won't support it, the market is right.
Rent differentials. If a feature reliably commands more rent — a shop, a garage apartment, an extra bath in a rental submarket — that difference can be converted into value. It's the natural method for investment property, and it occasionally rescues a question that sales data can't answer.
Market participants. Builders, agents who work that neighborhood every week, buyers who just chose one house over another. This is softer evidence and it can't carry an adjustment by itself. But it's often the thing that tells you which of the numbers above is believable, and it's worth more than appraisers sometimes admit.
Two things are worth saying plainly about all of them.
The first is that a supported adjustment can be zero. If the market shows no measurable difference between houses with the feature and houses without it, the honest answer is that it contributes nothing here — and that is a finding, not a failure to find one. Owners hear zero as the appraiser not trying. Frequently it is the most defensible number on the page.
The second is that the method is not the point. The point is whether whoever gave you the number can tell you where it came from. "Paired sales" is a good answer. "Regression on eighty sales in this subdivision" is a good answer. "Depreciated cost, checked against three sales" is a good answer. "That's what we use" is not an answer.
Three reasons, and each of them is something you can check yourself.
Price tier. A feature often contributes something closer to a percentage of the property's value than a fixed dollar amount. The same upgrade in a three-hundred-thousand-dollar house and a nine-hundred-thousand-dollar house is usually not the same number, and in the higher tier it may be expected rather than special. Pulling an adjustment from one price range into another is one of the most common errors I see, and it is almost never caught, because the number looks reasonable on its face.
What the neighborhood already has. This one is genuinely counterintuitive, so it's worth stating carefully. If every house on the street has a two-car garage, having one contributes close to nothing — it's the baseline. What costs you money is not having one. The contribution of a standard feature and the penalty for its absence are not mirror images, and the penalty is frequently the larger of the two.
So before you spend money to add something, the useful question isn't "what does this add." It's "what is normal here, and am I below it, at it, or above it?" Getting to normal is usually the best-returning money you can spend. Going past normal is where returns fall off.
The buyer pool. Features are valued by the people who actually buy in that market. A shop building is worth more where buyers have trucks and equipment. A pool is worth more where the comparable houses have pools and the buyers came looking for one. In the wrong neighborhood, the same pool is a maintenance obligation that narrows your buyer pool, and it can contribute nothing — or less than nothing.
Everything above is about differences between places. There's a fourth variable that catches people out, and it's the one I'd most want an agent to understand: contributory value changes over time, for reasons that have nothing to do with the property.
The clearest example any of us lived through is pools.
Before 2020, a pool in most of this market contributed what it had contributed for years. Then the world closed. Backyards became the whole vacation, the gym and the entertaining space at once, and buyers started reacting to pools far more strongly than they had the year before. The contributory value moved first — that was buyer behavior changing, visible in what people were willing to pay.
Cost followed. Once everybody wanted a pool, the builders had a queue, and the price of getting one built went up because demand for construction went up. Then the supply chains that had seized up caught up with it — materials, equipment, chemicals — and labor got more expensive on top of that. So within about two years, both numbers moved, in sequence, for different reasons.
Notice the order, because it's the opposite of what people assume. The market moved, and then cost moved. Cost did not drive value. Buyer behavior drove value, and later dragged cost along behind it.
Now run the same thinking forward, where it's genuinely useful.
Right now, in my market, solar panels and battery storage tend to contribute very little. Buyers here mostly don't pay much extra for them, and I have to report what buyers actually do rather than what seems like it ought to be true.
But look at what would have to change for that to stop being true. If the cost of power rose sharply — or if the grid got less reliable, which is a live question in a state adding data center load as fast as this one — buyers would start reacting to panels and batteries the way they reacted to pools in 2020. Not because the panels changed. Because the world around them did.
Generators are the same story with a twist. After the 2021 freeze, plenty of Texas buyers cared about backup power in a way they hadn't the previous winter. But that reaction isn't permanent either: if fuel got expensive enough that running one was prohibitive, buyers would quietly stop paying up for it, and the contributory value would fall back without a single generator changing.
Which is the whole point of this section. An adjustment is a measurement of how buyers behaved in a particular market at a particular time. Buyers change their minds, usually in response to something outside the housing market entirely — a pandemic, a freeze, a power bill, a fuel price. When they do, the number moves.
Two practical consequences.
For an appraiser: an adjustment extracted three years ago is not evidence about today. It has to be re-derived, and the assignments where that matters most are the ones involving features tied to something happening in the world.
For an owner deciding what to spend: the feature you add this year may be worth more or less when you sell, for reasons that have nothing to do with your house or your neighborhood. That's an argument for buying what you'll actually use, and treating any resale return as a bonus rather than a plan.
There's a term for improving a property beyond what the market will pay for: superadequacy. An over-improvement. It's the most expensive mistake I see homeowners make, and they almost never hear about it until they sell.
The pattern is always the same. Somebody puts eighty thousand dollars into a kitchen in a neighborhood where the houses sell for two-fifty. The kitchen is genuinely beautiful. It does not make the house worth three-thirty, because buyers shopping at that price point in that neighborhood won't pay it — and buyers who want that kitchen are shopping in a different neighborhood altogether. The money was spent. Most of it did not convert.
That is not an argument against renovating. If you're going to live there another fifteen years and the kitchen makes your life better, spend the money and enjoy it — you're buying a kitchen, not an investment, and that's a fine thing to buy. It's an argument against spending the money because you believe it will come back, without checking first whether the neighborhood supports it.
The check is not complicated. Look at what the best houses in your immediate area have actually sold for in the last year or so. That's roughly your ceiling. If your house plus the project lands above it, you should expect to eat the difference.
People arrive at this question having already found numbers somewhere. There are four common sources, and each is fine for what it was built to do and wrong for what it's usually used for.
Remodeling return surveys. Zonda publishes a Cost vs. Value report annually, and its figures get quoted everywhere. What it measures is what projects cost and what real estate professionals estimate they return, aggregated nationally and by region. It's genuinely useful for one thing: ranking project types against each other. Exterior work tends to outperform interior work, that sort of finding. It is not a prediction about your house, your street, or your price point, and it isn't built to be.
Automated valuation models. The estimates on the big portals are statistical models fit to an entire market. They are reasonable at the middle of the distribution and get worse as a property gets less typical. The model doesn't know your kitchen was redone, that your lot backs a greenbelt, or that the house two doors down sold cheap because it was a family transfer. For a common house in a subdivision full of similar houses, they're often in the neighborhood. For anything unusual, they can be badly off in either direction.
The appraisal district. The county's value is produced by mass appraisal, for taxation, on its own schedule. It is a different exercise with a different purpose and different rules. It is not an opinion of market value for your property, and neither a high number nor a low one tells you what your house would sell for.
Automated adjustment models. This is the newest one and the one worth slowing down for, because it doesn't have a settled name and most people meet it without knowing what they're looking at. You'll find tools — some of them built by appraisers — where you enter an address or a neighborhood and get back adjustment figures: what a bathroom is worth here, what the garage is worth, what the price per square foot runs.
Under the hood it's an AVM's close cousin. Same machinery, pointed at the components rather than the whole property: a model fit to a body of sales, applied to an address, with nobody having looked at anything.
The underlying technique is legitimate — it's the statistical analysis described earlier, and there's real software appraisers use to do exactly this as one input, disclosed, checked against the market before it goes in a report. Used that way it's good work.
Published as an answer, it inherits every weakness an AVM has. It doesn't know the condition of your garage, whether the feature is above or below what's normal on your street, or that your price tier behaves differently from the tier most of the data came from. And it carries a risk an AVM doesn't: a portal estimate announces itself as a computer guess, while an appraiser-branded number borrows a credibility the model hasn't earned. People discount the first and trust the second, which is exactly backwards from how much scrutiny each deserves.
The tell is the same as everywhere else in this article. Does it show you what it rested on — which sales, which neighborhood, which price range, how many — or does it just hand you a figure?
None of those four is dishonest. They're just answering questions you aren't asking.
If you're deciding whether to spend the money: find the ceiling first. What have the best houses within a few blocks actually closed at recently? Then ask whether you're currently below the neighborhood norm — in which case getting to normal is usually money well spent — or already at it, in which case temper your expectations about what comes back.
If you've already spent it and you're about to sell: the money is spent, and the only remaining question is how to price and present what you have. That's a different conversation, and it's a better one than relitigating the renovation.
If you're an agent pricing a listing, or arguing with an appraisal, the questions worth asking are:
Those four questions separate a supported figure from a confident guess faster than anything else I know. They work on me, too — that's the point of them.
You'll find tools that ask for an address and return adjustment figures. I've thought about building one and decided against it, and since that decision is really the argument of this whole article, I'd rather say why than leave it unexplained.
The first reason is the one above: a number keyed to an address, produced without knowing the neighborhood's norms, the price tier, the condition, or whether the feature is above or below what's typical there, would be wrong in exactly the way that costs people money. It would be wrong confidently, which is worse than being no help at all.
The second reason is that I'm licensed. When a Certified Residential Appraiser publishes figures about an identified property, that starts to look like an opinion of value regardless of what the fine print says — and in Texas, what something is depends on who performed it and under what standards, not on the label attached to it. I'd rather give you the method and a real conversation than a number I can't stand behind.
If you want the number for a specific property, that's an appraisal, and it comes with the analysis that supports it.
Terrence Bilodeau is a Certified Residential Appraiser (TX-1360232) and a Texas real estate broker. Clearfork Appraisals provides private-client appraisals in Fort Worth — estate and date-of-death, divorce, tax protest, partial interest, and pre-listing. This article is general information about how appraisers measure contributory value. It is not an appraisal and not a valuation opinion about any specific property.
Written fee quote, typically within one business hour. Questions about anything in this article are free, whether or not they turn into an assignment.