Automate real estate valuation with predictive modeling

Manual Appraisals

Before introducing AI, this was the only means of appraisal, and even today, it is still being used in tandem with AVM. The main idea behind the manual valuation process is quite simple – a real estate agent personally evaluates the worth of a building using physical inspection. They do this by looking at every part of the building, its location, the state of the market at the time of the appraisal, etc. All of these factors are analyzed and collected. While this data is analyzed more subjectively, if only because it’s performed by a human being, meaning that there may be a slight margin of error in human-provided sales prices compared to the ”real value” of a given building.

What Is An AVM’s Purpose In Real Estate?

AVMs are used throughout the real estate industry.

Real estate agents, brokers and mortgage lenders each use AVMs as a starting point to evaluate properties. This service is used to provide accurate estimates on property values to commercial platforms and major consumer sites like iBuyers.

For example, iBuyer uses an AVM to generate an estimate of your property’s value. Once the estimate is created, you can sell your house without any need to hire a real estate agent or meet with potential buyers.

AVM reports are calculated based on several data points, including the sales history, the tax assessor’s value, the basic features of a property, and the sales history of similar properties. With that data, an AVM will use a mathematical model to create an estimate of the value.

It’s important to note that an AVM will not consider the condition of the property when determining its value. With that, it’s possible for an estimate created by an AVM to be inaccurate.


Pros and Cons of Automated Valuation Models (AVMs) in Real Estate

Despite their widespread use nowadays, AVMs spark some debate, especially over how they compare to traditional in-person appraisals.

Advantages of AVMs

The advantages of using AVMs over physical appraisals are similar to those of any automated system over human effort. Basically, they save time, money, and effort. They can make numerous calculations and comparisons in seconds, and they don’t need to physically drive out to view a property or similar properties (“comps” being a key factor when appraising and pricing a particular piece of real estate).

All of this lowers the cost of valuing a property or multiple properties. AVMs are particularly useful in assessing the worth of an entire real estate portfolio. Once set up, AVMs can be operated with little expense.

As well as being cheaper and faster, algorithms aren’t subject to human error—or malfeasance. As objective automatons, they remove bias and subjectivity from the equation. Therefore, there’s less risk of fraud or deliberate mispricing—although, of course, computer programs can be hacked or manipulated.

A 2017 conference paper, “Automated Valuation Models (AVMs): a brave new world?” by George Andrew Matysiak of the Krakow University of Economics, referenced other studies in addressing the strengths and the shortcomings of these models. “There is little hard impartial evidence on the accuracy of commercially available AVMs in the public domain,” the paper noted. “Despite high average accuracy levels, statistically based valuations may be widely off the mark and need to be augmented by professional judgment.”

Disadvantages of AVMs

For an AVM to work well, it needs high-quality data in enough quantity to be representative. That’s where its vulnerability lies.

The most oft-cited drawback against AVMs is that in determining value, they do not (and cannot) factor in the actual condition of the property. They just assume an average state, which may or may not be accurate. They can’t note details or variations in the condition.

AVMs are great at comparisons, but what if there is a dearth of comparable real estate (comps) or transactional data on record? For this reason, newly built properties are especially tough to value, and AVMs, being rather literal, tend to lack the imagination to come up with something to serve as comps. And because an AVM works based on known factors—the historic record—it misses out on intangibles that can raise or lower an estimate.

Finally, an AVM can only work with the data that it is given, and there is always the danger of data being entered incorrectly. Also, the information that it does have might not be up to date—making AVMs unreliable in fast-changing real estate markets.

4 is the official website of the Nationa is the official website of the National Association of Realtors. It won second place in Newsweek’s top three real estate databases for 2020. has a similar user experience to Redfin. In addition to the property overview of the number of bedrooms and bathrooms, it shares how the home value has changed over the past eight years and how it potentially will change in the next year.

Being able to see the trends in the price is really helpful to indicate if the neighborhood is improving. After all, if you buy a house, you will likely hold onto the property for many years before selling again. The website also lists similar properties in the area. They also have a visual timeline of the property’s history to indicate when the home was sold and at what price.

It will also show tax assessments and local schools. Also, the neighborhood’s average listing price, selling price, and how many days properties remained on the market are displayed above the listing prices for neighborhoods adjacent to the property you are viewing. This feature is great if you are unsure of what area you want to buy in. also includes a similar feature that shows properties at the price you enter. This feature is really helpful if you have a set budget and only want to look at your price range properties.

Our Verdict

We chose Zillow as a top option because it has a database full of listings that yields estimates with a 1.9-percent on-market and a 6.9-percent off-market median error rate. Redfin is a solid runner-up choice, with a 2.24-percent median error rate for on-market homes and a 6.74-percent rate for off-market homes and an easy way to get a free comprehensive analysis.

2. Redfin

Redfin’s estimate also draws on multiple lis

Redfin’s estimate also draws on multiple listing services to generate a property value with a 1.73% median error rate for homes on-the-market. Its off-market estimates have an average error rate of 5.84%. Redfin’s platform is easy-to-use, and it will also show the home value will potentially change over the next five years. In addition, it will show similar properties nearby.

Redfin’s estimated price seems to fall within the lower spectrum of the Zestimate’s range. It also shares local schools, rent estimates, and provides a score on the neighborhood’s walkability and bike-ability. Like Zillow, it also provides crucial information about the community from the median list price to the average number of offers each property received.

Redfin is as well known as Zillow and provides similar types of information. There are clearly some differences in the data they aggregate, which leads to the variance in estimates. One of the primary reasons for the difference in estimations from Zillow and Redfin, both leaders of online home valuation tools, is that Redfin takes into account the listing price of a property.

This intuitively makes sense as the home’s selling price will likely be close to the listing price. When comparing a few properties prices, Redfin’s listing is usually lower than the Zestimate. This listing highlights the importance of the data used to generate the estimation. Redfin has almost 75 million properties in its database. But since it uses the listing price, many people think Redfin provides a more accurate assessment than Zillow.


Designed for situations that require a high volume of accurate and quick property valuations.

Before You Use One of the Best Home Value Estimator Sites

Keep in mind that these home value estimator sites aren’t 100 percent accurate—they merely serve as a jumping-off point before taking a home to market or considering a loan. Many variables go into estimating a home’s value, and ultimately, real estate values can shift daily. With that being said, a user may want to get estimates from several of the top sites. If a homeowner has recently renovated or remodeled their home, the estimate may not reflect these updates. Therefore, a homeowner will have to use additional tools such as a home improvement calculator to help determine the return on investment (ROI) of their renovations. Keep in mind that not all upgrades will improve the home’s value, and those that do are unlikely to yield a 100 percent return on investment. Another thing to remember is that estimates from home value estimator sites most likely won’t be used to help list a home or refinance it. A more accurate assessment is needed, such as a real estate agent’s comprehensive analysis or an appraisal.

Related Resources

Viewing 1 – 3 of 3

Home Appraisal Amount Vs. Loan Amount: How An Appraisal Impacts The Selling Price And Mortgage Amount Home Buying – 7-minute read Hanna Kielar – June 27, 2022 Whether you’re selling or buying, you should learn how the outcome of the home appraisal affects the sale. Read More

What Is Fair Market Value (FMV) In Real Estate And How Is It Determined? Mortgage Basics – 4-minute read Lauren Bowling – May 23, 2022 Fair market value (FMV) in real estate is an assessment of a property’s worth in an open market. Learn how FMV is determined and what it’s used for. Read More

What Is An Appraisal Contingency? Home Buying – 7-minute read Victoria Araj – June 30, 2022 An appraisal contingency can give home buyers peace of mind about the purchase price of a new home. Learn how appraisal contingencies work, and if you need one. Read More

Investing and Servicing

Value support for live bidding Primary valuation for seasoned second liens Bulk refresh of portfolio values with low volatility


How does HomeLight’s free home value estimator work?

HomeLight’s Home Value Estimator is a fantastic option that includes a unique approach to valuing properties. We’ve done a lot of analysis and iterations to determine the right combination of publicly-available, proprietary, and user-submitted data to produce more accurate value estimates.

The process includes a short questionnaire with seven questions that are proven to increase the accuracy of your home valuation estimate. These include simple information such as the property’s condition and the year it was built.

Additionally, our free AVM includes your home’s Simple Sale price. With Simple Sale, you can request an all-cash offer for your home. If you accept, you can pick the move date and get your home sold fast without the hassle of buyers coming into your home. We estimate a home’s Simple Sale price to be around 90% of market value.

Try HomeLight’s Free Home Value Estimator now.

Having a full-time Realtor® with experience and access to the MLS (multiple listing service) is best for a reliable CMA. We aren’t appraisers, but we can usually get very, very close. Sandi Bates Real Estate Agent Close Sandi Bates Real Estate Agent at EveryUtahHome 5.0 Currently accepting new clients Years of Experience 23 Transactions 193 Average Price Point $373k Single Family Homes 129


While AVMs are still in the process of making a name for themselves in the realm of real estate, it seems inevitable that, as Big Data expands and algorithms improve, human real estate agents will rely more and more on them. Their results are already apparent even today, serving to streamline the process greatly. The few holes in their data, such as the current state of the building, may soon be filled with better access to data or even better integration into the process of human appraisers. 

Whatever the case, it looks like things might keep changing for the better in the real estate industry thanks to AI. If you need some advice, write to us – we can help your business!