Your assessment is not a judgment about your house. It's a prediction — the output of a statistical model that valued hundreds of thousands of Philadelphia parcels at once, without anyone ever looking at yours. Understanding how that model is built, and where it breaks, is the whole argument for why so many individual assessments can be challenged.
The Office of Property Assessment (OPA) does not appraise homes one at a time. It uses mass appraisal: a single statistical model that ingests recent sales across the city, along with the recorded characteristics of every parcel — square footage, lot size, year built, number of stories, and so on — and learns the relationship between those characteristics and sale prices. It then applies that learned relationship to every property to produce a predicted market value.
This is the only practical way to value a city of this size on a schedule, and it has real virtues: it's consistent, it's repeatable, and it treats similar homes similarly. But it's important to be clear about what it is. Your assessed value is a model estimate built from patterns across groups of homes. It is not the product of anyone standing in your kitchen deciding what your particular house is worth.
Mass appraisal is genuinely good at the average case. On a block of near-identical rowhomes where several sell every year, the model has abundant, clean signal: it knows almost exactly what that kind of home fetches, because the market keeps telling it. For cookie-cutter housing with deep, recent sales data, the assessed value is usually close.
The model gets weak in four recognizable situations:
The model is a prediction about a type of home. The further your home sits from that type, the more the prediction is a guess.
The errors aren't random noise — they cluster in ways that push individual assessments off in predictable directions.
Square footage and prestige can outweigh condition. A model leans on the features it can measure cleanly, like size and location. It struggles to see that two similarly sized homes can be worth very different amounts because one is restored and the other hasn't been touched in decades. Condition is one of the largest real drivers of value and one of the hardest for a model to capture.
New-construction and flip comps drag up their neighbors. When a gut-renovated home or a new build sells at a premium nearby, that sale enters the data the model learns from — and can lift the estimated value of un-renovated homes on the same block that would never sell for the same price.
Stale characteristics sit on file. The record the model uses may describe a home that no longer exists: an overstated square footage, a wrong bathroom or story count, a finished-attic flag that's inaccurate, a condition rating years out of date. The model faithfully values whatever the record says, errors included.
Thin data forces extrapolation. Where genuinely comparable sales are scarce, the model reaches toward homes and neighborhoods that are less and less alike — and every reach widens the margin for error.
Here is the through-line. Your assessment was produced by a process that never examined your specific home, using data that may be incomplete or out of date, in a system whose errors concentrate exactly where homes are unusual, conditions vary, or sales are thin. An appeal doesn't ask the model to try again — it introduces something the model never had: the actual, comparable-based case for what your home is worth.
That's a favorable matchup. A careful argument built from genuinely similar homes — adjusted for size, lot, and condition — is grounded in the same market evidence the model relies on, but applied to your property with a specificity the mass process can't reach. Our comparable-sales guide shows how to find and adjust those sales defensibly, and the is-my-home-over-assessed self-check walks through the underlying comparison.
None of this is a guarantee. Some homes are assessed about right, and an honest process tells you so. But where a real gap exists between the model's estimate and what comparable homes actually sell for, that gap is the appeal.
Testing your own assessment is straightforward in principle: compare your assessed market value against what truly comparable homes near you have sold for recently, and adjust for the ways your home differs. If similar homes are trading for meaningfully less than your assessment implies, you likely have a case. If they're not, you've spent nothing to find out. At roughly 1.4% of assessed value (1.3998%), a correction on an over-assessed home recurs every year until the next reassessment, so the exercise is worth doing carefully.
Our how-to-appeal guide lays out the two-track Philadelphia process — an informal First Level Review and a formal Board of Revision of Taxes appeal — neither of which requires a lawyer. And the fastest way to see where you stand is the free check below: enter your address and let it run the comparison against the City's own recorded sales.
Mass appraisal is the practice of valuing many properties at once with a single statistical model rather than appraising each one individually. Philadelphia's Office of Property Assessment feeds recent sales and recorded property characteristics into a model that predicts a market value for every parcel in the city. It's efficient and consistent, but it estimates from patterns across groups of homes — it does not appraise your specific house on its own.
A model is only as good as the data it's trained on and the characteristics on file. Where recent sales are scarce, where homes are unusual, or where the recorded square footage, condition, or features are stale or simply wrong, the model has to extrapolate — and extrapolation is where individual homes drift away from what the market would actually pay. The value can be off in either direction, and the more your home differs from the typical parcel, the more room there is for error.
Generally, no. A citywide reassessment values hundreds of thousands of parcels, and no one walks through each one. The model works from data already on file and from recent sales nearby — not from a fresh interior inspection of your home. That's precisely why a specific, comparable-based appeal can beat the model: it introduces facts about your individual property that the mass-appraisal process never considered.
Mass appraisal is accurate for many homes — especially cookie-cutter properties on blocks with lots of recent, similar sales. The errors concentrate in predictable places: unique homes, thin-sales areas, mixed-condition blocks, and neighborhoods changing fast. We don't claim any particular share of assessments is wrong; the honest approach is to test your own against real comparable sales rather than assume.
Compare your assessed market value against what genuinely comparable homes near you have actually sold for, adjusting for size, lot, and condition. If truly similar homes are selling for meaningfully less than your assessment implies, that gap is the basis of an appeal. TaxAssessmentIQ runs that comparison against the City's own recorded sales for free, so you can see where you stand before deciding anything.
TaxAssessmentIQ gives you a free, honest verdict from the City's own recorded sales — just enter your address, no sign-up and no account. If your home looks over-assessed, you can get a ready-to-file appeal packet for a flat, one-time price, backed by a money-back guarantee. If it isn't worth filing, we tell you that too — for free, before you pay anything.