An automated valuation model (AVM) uses property and transaction databases in mathematical models that calculate property values. Realtors, appraisers, mortgage companies, and title agencies use AVM in real estate to get an objective valuation of properties. 

  • It is not an appraisal 
  • It is a complex statistical model designed to predict the price of a house 

While many people treat these as low-cost alternatives to appraisals, they have limitations.  

Is an AVM in Real Estate an Appraisal? 

An AVM is not an appraisal. It is a mathematical tool to evaluate the value of a property. Appraisers use them, but an actual appraisal looks for qualitative as well as a quantitative measure to evaluate a house. AVMs: 

  • Are less equipped to handle intangible factors 
  • Cannot compensate for unknown factors that can either increase or decrease the value 
  • Are quicker and cheaper than a full appraisal done by a licensed professional 

They cannot be as comprehensive as an appraisal, but they are still a very valuable tool. Often, they are very accurate and give you a very good idea of what is going on in the market. In fact, most banks now use them as part of their underwriting processes.1 

Will AVM in Real Estate Replace Appraisers? 

AVMs probably will not replace appraisers soon. Statistical models have shortcomings that make traditional appraisals necessary.  

  • Cannot adequately evaluate qualitative measures 
  • Often lack “good” data 
  • Cannot compensate for property conditions 
  • Are unable to properly evaluate home improvements 

Good statistical models depend on large volumes of accurate data. This presents two problems. First, while there may be data for large subdivisions of nearly identical homes, how do you collect adequate data for a unique home in a unique location? Second, how do you quantitatively evaluate an upgrade, such as a new kitchen or bathroom? One person may love it and pay more for it; while another may hate it and want to pay less for it. Subjective criteria are very hard to model, and buying or selling your home can be very subjective. 

Further, these models depend on good modeling. Specifically, what variables are important, how do you weight them? Some examples are the: 

  • Age of the wiring and plumbing 
  • Type of roof, furnace, air conditioner, and hot water heater 
  • Presence of wood-destroying insects, mold, or radon 

These and other variables can be very hard to measure. Therefore, there are times when you still need a full appraisal. AVMs are an important tool in the process, and they can give you a very good idea about the value of a property, but they cannot give you the complete picture.2  

Is Zillow an AVM? 

Zillow is not an AVM, but they have a tool: Zestimate. Zillow is a marketplace where you can find professionals related to real property transactions. They have sections for buying, selling, renting and mortgages. Within these categories are several tools, including their online automated valuation tool. 

AVM in Real Estate Cost 

AVMs are cheaper than appraisals. Some simple ones are free, while other, more accurate ones do have fees.  

  • Many banks, realtors and mortgage brokers offer simplified, free tools online 
  • Companies, such as CoreLogic, have subscription services meant for banking professionals 

A free online tool is adequate if you want to know how much your house is worth, or if you want to know how much houses sell for in the area you want to buy. As you get more serious your agent should provide you with a more comprehensive report.  

How Does an AVM in Real Estate Work? 

An AVM is an algorithm that uses a hedonic statistical model and a repeat sales index. Sophisticated models use a complex analysis that takes these factors into account. The complex valuation inspects: 

  • Recent deed transactions 
  • The most recent price paid for the property 
  • Prices of similar properties 
  • Previous valuations 

These are very complex mathematical models, and they give very precise information. They do have limitations, however. They cannot account for a unique property, and they can’t deal with the nuances of the house.3  

Hedonic Statistical Model 

A hedonic statistical model uses regression analysis to estimate the influence of variables on property prices.  

  • Price is a dependent variable in the model 
  • Attributes of the property make up the independent variables 
  • The coefficients of the independent variables represent the weights buyers place on those variables 

What all of this means is that the model tries to estimate how important various aspects of the property are to you, so that it can calculate an accurate price for the property.4  

Repeat Sales Index 

A repeat sales index is another statistical model that tracks changes in price over time.  

  • It measures the same property over time 
  • It calculates the change in price over specific time periods 

This eliminates the problem of accounting for different characteristics in houses. It restricts data to houses that sold multiple times over a specified time period. It also does not account for the maintenance of the house. 

AVM Confidence Score 

An AVM confidence score signals how reliable the generated price valuation is. The reason that you would use a regression model is that you don’t really know what the price is going to be.  

  • Modelers use independent variables to predict the dependent variable, in this case, the price of the house 
  • Which variables to use is an important decision for the model 
  • Some independent variables will be significant and given weight, and others will be insignificant 
  • If you leave out a statistically significant independent variable then the whole model is flawed 

The confidence score tells you how reliable the statisticians believe their model is. A lot of factors go into this, but it boils down to how well they develop the model, and how good the data is. 

AVM Vs. Market Value 

AVM is a statistical tool that appraisers should use to determine market value. It can be an elusive goal to determine the value of a house.  

  • A unique property renders a statistical model almost useless 
  • Extensive renovation or lack of maintenance may also mean that you can’t use a valuation model 

Mathematical models get more accurate the more good data you can use, but even then, it can’t predict price with complete accuracy. An appraiser who understands all the tools available should always be able to predict value with greater accuracy than just the model alone. 

The Bottom Line 

AVMs are important new tools available to both buyers and sellers. There is value in an objective model for the housing market. This is because we tend to let our emotions play a big part in what house we want to call home. Limitations of the models dictate that they cannot be completely accurate, but they are can form the basis of a more objective analysis of the property. They can help you get a better idea of what a home is truly worth. 


  1. PennyMac  
  2. CoreLogic 
  4. Investopedia