Recently, I ran across a hilarious little parody website called ReallyRottenRealty.com. The site exposes some of the more compellingly awful business practices of some brokers and agents out there. Check it out, and be prepared to laugh while grimacing.
Turns out, the website is a marketing vehicle for a company called AgentHarvest, based in Dallas, TX. It’s basically a lead referral business, the likes of which we’ve seen before. But there are some interesting, unique things about AgentHarvest that make it worth discussing a bit.
Based on Finding an Agent
Most lead-generation websites in real estate are centered around properties. Homegain, for example, and “media” sites such as Realtor.com, Trulia, Zillow, and others are all focused on property search or home valuation that ultimately create a lead for a broker or an agent.
In contrast, AgentHarvest is centered around real estate agents, and promises to find its customers the best agents for their needs. (To be fair, HomeGain has something called AgentEvaluator that promises the same, but the selection criteria are likely to be slightly different. More on that below.) Most industry people right about now are shrugging their shoulders going, “Uh, so? Doesn’t like every single real estate website out there offer the same thing?” For example, Trulia has Find A Pro, and Zillow has its directory, and every brand website has some way to Find an Agent or Find an Office. So what’s the big honkin’ deal about AgentHarvest?
Enter Performance Statistics
A while back, the announcement by the Houston Association of REALTORS that it would begin allowing consumers to access the performance statistics of real estate agents was received as major news. Respected voices like Marc Davison of 1000watt applauded the leadership shown by HAR with its REALTOR Match tool. The HAR tool, in turn, was inspired by Agent Scouting Report, created in 48 hours by Diverse Solutions for the 2009 Connect Create contest.
At the heart of both of these widely-praised tools is the notion that a real estate agent can be evaluated in large part from his or her production statistics. For example, from the blogpost about Agent Scouting Report, we know that the Diverse Solutions team used an algorithm to take into account:
- Popularity: How many homes have they sold in the last 180 day, 1 year, and 2 year periods?
- Salesmanship: What was the average number of days on the market
- Knowledge of market: How many times did they drop the price from the initial listing?
- Experience: How long have they been a member of the MLS?
- Ability to negotiate: How close was the last list price to the final sales price?
- Diligence: On average per listing, how many open houses do they have, how many photos do they load in, and how lengthy are their descriptions?
In other words, number of transactions over a period of time, DOM stats, years of experience, sale-to-list ratio, and some other activities for which there are stats in the MLS data. There are opinions that such statistics are woefully inadequate to determine the quality of a real estate agent for a variety of reasons, nonetheless, it is a fact that industry people do take production statistics into account. For example, I have as yet to see an agent be named as a “top performer” by a brokerage or an Association or a franchise who didn’t have production numbers to match.
What AgentHarvest promises to do is to use performance statistics drawn straight out of the MLS to find consumers the best agent for their particular needs. The owner, Bill Petrey, is a licensed Realtor in Texas, and a member of the local Association. He, therefore, has access to the performance statistics from the MLS, as every member does. As AgentHarvest attempts to break into other markets, Mr. Petrey (or someone like him) will need to become subscribers to the local MLS in question.
Think of it as a human-powered version of HAR’s REALTOR Match or Diverse Solutions’ Agent Scouting Report tools. I got curious about how AgentHarvest goes about selecting agents for its customers, so I emailed Bill Petrey. Based on his answers, it appears that AgentHarvest basically uses the number of homes sold within last six months as the relevant statistic, then layers on both (a) familiarity, that is, has the agent done business with AgentHarvest in the past, and (b) what he calls the “naughty or nice” list based on consumer feedback:
Q: Do you rely on anything other than performance statistics when you make your recommendations? How is that information gathered? (For example, do you have an internal database of showing notes or consumer comments about an agent?)
A: We have a few processes that we go by, but the first round of elimination is solely based on statistics. I think that’s the only way to eliminate all forms of puffing, exaggeration and blatant lies. The first round in the selection process is a list of all agents that sold property, similar in size to the client’s house, within a 5-mile radius of the homeowner’s home (or buyer’s target area) within the last six months. If you aren’t on this list, you don’t get considered for the referral. From this list, we select several agents who have had the most sales and we also choose agents that we’ve worked with in the past and have received favorable remarks from our clients. From these selected agents, we choose three agents to present to the client. We also give the client some information to interview and research all of the candidates before making their selection.
The information we use on our initial round of elimination is gathered solely from the local MLS in that client’s area. Naturally this limits our service territory but we feel that it’s worth doing that way. I’d rather do a lot of business in a few places than a little business in a lot of places. If I’m cold-calling an area I’m not familiar with, like most agent referral services do, I’m not really doing justice to my clients. The information is gathered via MLS and we collect and build our own database of consumer comments regarding agents through our follow-up process after the agent interviews and during the agent’s listing. We call it our “naughty or nice list.” We do blacklist agents that receive bad remarks however don’t release names for OBVIOUS reasons.
The business model is straightforward. Since Bill Petrey is a licensed Realtor, he is able to collect referral fees (~25%) from the agent to whom he has sent the lead upon a successful closed transaction. Every Realtor in the country does this business, and does it all the time. It’s part of the industry’s practices.
So why is this interesting at all?
The Last Hidden Database
I find AgentHarvest interesting because it is (1) a business model I expected to see formally introduced for a couple of years, and (2) partially unveiling the last hidden database in real estate.
Perhaps there have been other companies that have done what Bill does: become licensed Realtors for the sole purpose of lead generation, leading to referral income. But Bill is the first I’ve seen who became a Realtor and will join other MLS’s for the express purpose of using the agent performance data to create a referral business.
In this, what AgentHarvest is doing is reminiscent of what Realtor.com and other property-centric websites have already done: taking information that was formerly the province of professionals alone and providing access to the consumer. We have heard for years that listings are everywhere, and that competing on the basis of listing data is foolish. That’s probably true. But what we also know is that while consumers can go to any number of websites and companies to do research on properties, neighborhoods, demographics, and so on, only the MLS holds all of the performance data on an agent. If property data is now everywhere, the last hidden database is agent performance information.
The only thing a consumer cannot research today with any degree of efficiency is agent quality. I have written and spoken on the topic of how a consumer could tell if a real estate agent is good or bad, and oftentimes, I come back to the conclusion that only other real estate professionals really know who’s good and who’s not. Because professionals, unlike consumers, deal with real estate and other professionals in real estate every single day, rather than once every seven years.
In AgentHarvest, we have a Realtor willing to tell you the consumer which of his peers are any good, using the hidden database of agent performance metrics that lies within his MLS system.
Seems to me, it’s just a matter of time before this kind of referral business pops up everywhere.
Imperfect, But Interesting
I think there are a number of gaps and holes in AgentHarvest as it is today. For one thing, it isn’t clear that number of sales standing alone tells enough of even a high-level story. It may be, of course, that Bill does use all the secondary statistics like DOM and price-to-list as well, and has some sort of algorithm that lets him recommend one agent over another. But deciding how to weight various statistics is in and of itself a real judgment call.
The larger issue is that a company whose entire experience with other Realtors is limited to sending them leads may not have enough direct transaction experience with them to make real quality judgments. Is the agent detail-oriented? Is she responsive? How is she in negotiations? How about managing the transaction? Consumer feedback helps, but remember that they’re consumers — they don’t have enough knowledge to compare.
Nonetheless, AgentHarvest is interesting precisely because it addresses a market need that is currently unfilled: finding the right agent. Automated tools like REALTOR Match and Agent Scouting Report may be the longterm answer, but they have even less exposure to the intangibles like patience, attention to detail, friendliness, and so on. For the time being, I would think that a slightly improved version of AgentHarvest-type of service would be invaluable to consumers.
Time will tell. This could be the start of something interesting.