Real estate agent matching, rating and ranking: the big data placebo pill, and quelling the angry mob

An incorrect picture is far worse than no picture at all

It seems that businesses worldwide are turning to big data for answers in an increasingly frequent fashion. Big data has been credited with a wide range of professional successes, including identifying tax fraud in healthcare systems, predicting retail buyer behavior and improving water management. While the victories lend support to expanded uses, the creators of data analytics programs need to remain open to the idea that the answers they produce might not fit into a crisp, singularly conclusive box. Oversimplifying data crunching to make sure the conclusions are neatly formatted answers can easily invalidate the entire process.

Agent matching, rating and ranking

shutterstock_128627147

This is a particular problem for the real estate industry. Rating and ranking systems for real estate agents are popping up more and more frequently, with realtor.com’s AgentMatch beta being the latest stick to stir the hornets’ nest. Each one is met by a fair share of professional pushback, often with good reason. As we attempt to simplify ratings and reviews of agents, we rely on data created by millions of individuals from hundreds of different organizations, all of them seeking to consolidate an enormously complex transaction into a few defined data points. The solution is often to lean heavily on a few easily compiled measures and hope that they’ll create an answer that can be clearly displayed to a consumer in a “Who is better and who is worse?” format.

The arguments for more consumer access to agent sales data from a single trusted source are clear. The data is already out there; other companies are already doing it; and we might as well create an authoritative system that consumers can rely upon, before another shoddy version becomes popular. Notwithstanding those points, the real estate agent population, much of which is open to a broader display of its sales history data, would offer two significant challenges to the process.

First, the measures used to compare agents in most rating systems are inconsistent and produce ratings of negligible value. Second, a simplified comparative analysis can’t address the widely divergent situations, needs and preferences of consumers who would be better served by a broad set of background data than a dumbed-down, simplistic answer.

Consumers want an agent who sells their home for the highest price possible and provides an excellent service relationship in the process.  While some stats can show that an agent has local sales experience, none so far can accurately measure those two biggest consumer concerns.

Sale-to-list-price ratios: foundation material necessary to build a house of cards

Let’s start with breaking down a dogma that pervades the real estate listing sphere. Sale-to-list ratio is a virtually worthless statistic in online agent comparisons. Unfortunately, it’s a cornerstone of many agent rating systems attempting to assign a simple grade to an agent. A consumer using this statistic in their evaluation of a real estate agent is just as likely to harm their financial outcome in the sale of their property as they are to help it. A high sale-to-list price ratio has no statistical tie to maximizing seller take-home proceeds or to the seller’s happiness with their agent relationship during the transaction. These are the two things consumers want, and it addresses neither.

A quick breakdown of the sale-to-list-value fallacy:

  • Listing agent A has a sale-to-list ratio of 96 percent. She sells properties for family and friends, and promises to maximize the take-home proceeds of her sellers, even if it takes some extra time and effort. Her vow to her clients is to never initially underprice their homes and leave money on the table. While it sometimes takes a price reduction and some time to find the right buyer at the maximum price, the seller is never concerned that they gave their house away.
  • Listing agent B has a ratio of 98 percent. His sales are comprised of 10 percent traditional resales and 90 percent bank-owned homes. The bank wants its properties sold quickly, and asks him to list them at 5 percent below market value to reduce their carrying costs. Every home sells quickly, and almost always at the list price.
  • Listing agent C has a ratio of 100 percent. She works in a quasi-pocket-listing group. Sellers are approached by the listing agent with a “buyer in hand.” The seller is given a price that the buyer will pay without the home ever going on the market. If the seller and buyer agree to the price, the listing goes immediately from active to pending on the MLS, and the agent retains a perfect full-price record. Open-market buyers might have paid more, but the seller’s home is never exposed to them.

It’s clear that these agents serve very different roles that can’t be explained by a simple statistic. The most damning conclusion from this scenario is that Agent C, whose practices could be seen as the least beneficial to the average home seller’s pocketbook, is purported to be the “best agent” if these ratios are given any weight. Even if a consumer was actively seeking an off-market sale, all three of these agents could provide those services, and the list-to-sale ratios given here would not help him or her know which of these three agents could make it happen more quickly or at a higher final sale price.

Days on market: further erosion of statistical support

The amount of time it takes a listing to go from active to pending is another big crutch for most agent rating services.  The “days on market” measure lacks the ability to capture even a portion of the complexity of why a home is being marketed for a certain period of time. Pocket listings sell quickly without being held up against market comparables. Short sales naturally take inordinately long time frames to close, and often involve agents who work far more hours for their clients than average. Quick sales by successful agents are often the result of outstanding marketing and pricing, and generation of open house traffic. Other times, the homes were merely underpriced and they sold right away. An agent who takes on challenging, unique properties often puts far more effort into those sales, but their marketing times will be much higher than average.

Remember the famous statistic that real estate agents, on average, keep their own homes on the market longer, and get higher prices, than they do for their clients? If we know that a short time on market isn’t all it’s cut up to be for our own homes, we shouldn’t push it on the public. It’s fairly clear that by combining these two horrendously overused statistics and building a product on top of them, the output is nothing more than a shaky house of cards. What we see with most agent rating systems is the attempt to make these kinds of statistics more palatable, easier on the eyes, and more shareable to the public. The use of data that is vaguely valuable on its own, and has no correlation to the consumer’s main goals, is merely a necessary casualty in the pursuit of an easy answer that drives traffic and popularity.

Transparency gets foggier the further we zoom in

As we continue to force complex data into ill-fitting metrics, we oversimplify the answers we’re giving to consumers so much that they can’t even see the real agent any longer. The focus of the consumer, whom we’ve told to compare agents based on these questionable stats, becomes myopic.

Meanwhile, a larger profile of each agent, combining their full sales history, unique expertise, background and verified reviews could be significantly more helpful to the consumer. Selling a home is a huge undertaking, and allowing the home seller to research local agents in a thorough, complex way would be the responsible direction to lead a consumer website that intended to inform the homeowner and support the agents’ outreach to them.

Why haven’t previous attempts at exposing more agent sales data gone this route? Either through the use of incomplete databases, inconsistent data sets, or just a lack of concern for the complexity of what they’re displaying, the creators succumbed to the zeal to put out a new shiny object as quickly as possible. It has inevitably created a simplistic caricature of the true agent profile. The resulting frustration from real estate professionals shouldn’t surprise anyone.

Dump the comparisons and rankings — just inform the consumer

The statistics used in attempting to rank agents rarely do it well. The goal of naming a “best” and a “worst” is always attractive, but the consumer is looking for more information, not beauty pageant results.

Imagine reading a travel guide that simply said Rome is an 8.5 and Prague is a 6.4. Why? For what reason? For which kind of traveler? It’s the same as attempting to compare Domino’s delivery and Uncle Anthony’s Little Italy. They both have pizza, but they serve different consumer needs, time constraints and budgets. There is no better or worse on a single scale. There is only better or worse for each individual’s needs.

Give consumers the red pill, and deliver the mess they deserve

Real estate is not a cut-and-dry business. There is no carbon-copy transaction. As much as data advocates decry this as a cop-out, and online traffic scrambles for one-touch buying and simplicity in all things, it’s a disservice to the consumer to pretend that a real estate sale can be packaged this way.

If we take the easy route, we can deliver the metaphorical blue pill, and the consumer will wake up in their bed the next morning with an agent who rated 9.5 and may or may not be a good match for them at all. On the other hand, we can deal with the serious nature of a home sale, offer them the red pill, and allow them to see the way real estate really works. This could include much more transparency about agent sales and other statistics, but it will be necessarily complex.

Consumers don’t know they’re getting the placebo

Consumers want answers: more transparent data about agents and sales. When they receive ranking data from a trusted source, they may very well make a decision to list their home with an agent based on those statistics. They will feel satisfied that they did their proper research. The resulting popularity of that rating system and the consumer’s feeling of accomplishment have no bearing on the question of whether it truly was a good decision for the consumer.

If we were to tell that same consumer that the statistics we used were highly unlikely to reveal the ability of that agent to get the best return on their investment, or to provide a successful agent-client relationship during the transaction, they would feel betrayed. The decisions they made, based on the purported answers we supplied, were merely a research placebo to make them feel confident that they were ready to make a decision. We didn’t actually help them make a wise decision.

Reporting real estate agent data in a heated sphere

Early creators of agent ranking systems stepped in it. They misrepresented thousands of agents in public spheres. They showed a lack of care for thousands of small businesses’ reputations, and the angry mobs that ensued were surprisingly unified for such a decentralized industry. This is the atmosphere that a company creating a new agent data display system is diving into. The water is boiling hot, and whether or not the newcomer is responsible for the buildup of distrust, they must be keenly responsive to it. There is a potential for a great consumer benefit in this space, but past transgressors have turned the path toward it into a tightrope.

It’s up to new agent statistical display builders to cover their bases before seeking the limelight.

  • Get your data in order first. Get all of the data. If you think you have it all, ask the local agents in that market. Without a full encompassing of the market’s data, any conclusions drawn will be incorrect. If that requires a custom data outreach effort in each distinct market, so be it. If you can’t get access to the local MLS data, don’t publish in that market. An incorrect picture is far worse than no picture.
  • Present a broad, data-rich display of each agent’s profile and history. Don’t dumb it down for the consumer. They need real statistics, not bite-sized buzz-stats. Treat them like adults making six-figure decisions. Reviews, unique expertise, and individual sales transactions allow consumers to ask agents questions that are relevant to their own unique situations.
  • Take your product back to the agents to see what they think. They will never all agree, but at least those who favor more transparency will be able to tell you if it’s truly ready for the public, or if it’s a time bomb waiting to go off. No matter how much pressure there is, don’t go public until you have some industry buy-in.
  • Skip the rankings and stick to educating. Consumers will learn far more by poring over a detailed report on an agent’s individual home sales in their neighborhood than they will by comparing the boiled-down stat of the day. Grouping agents by the communities they have sold in will allow for some level of clarity, but the minute you become the source of a “she is better than he is” conversation, you’ve overstepped your role and your support will flat-line.

This is no guarantee of success or of real estate agent support. It’s merely a few steps to keep your feet out of the fire long enough to potentially create a viable resource and give the consumer a more accurate look inside the real estate industry.

Sam DeBord is a managing broker with Coldwell Banker Danforth in Seattle, state director for Washington Realtors and a real estate/technology writer for numerous online outlets. You can find his team at SeattleHome.com.