Following the scent laid down by Redfin, the Jesuit bloodhound is hot on the trail and won’t let go. This looks like way too much fun to miss, being a numbers geek and all, so I’m happy to join in on what is otherwise a fairly pointless exercise: you can make numbers say just about anything you like. Plus, the only real numbers that count are those on Redfin’s P&L statement, which are still, by Glenn Kelman’s admission, distinctly in the red; his job as CEO, of course, is to eventually change the color of that ink to black and thus handsomely reward his investors.
At any rate, here we go…

I hereby stand by Redfin’s original claim, slightly modified per the obscure and eventually irrelevant mistake which I took great delight in finding — not because I have any particular need to prove or disprove what Redfin is saying, but simply because I take pleasure in futzing with numbers. An Excel summary shows that, with the data Redfin used, indeed, on average, homes in which Redfin represented the buyer were sold for, on average, a tad over 99% of the list price, whereas non-Redfin-represented buyers, on average, paid a tad over 100% of the list price.

This much is a fact, indisputable. What it means, however, is a completely different story. Some, including Glenn Kelman himself, would have us think that the difference is simply because Redfin agents are better negotiators. Alas, we have absolutely no way of proving or disproving that from the numbers. Correlation, if I recall correctly from my stats classes, in no way implies causation. The “Fasten Seat Belt” sign is highly correlated with take-offs, in-flight turbulence, and landings, but is unlikely to actually be the cause of any of them.Any of dozens of explanations could be behind the numbers: Redfin’s clients may tend to be bargain seekers who are content to wait till they find those hidden gems (or fallen crumbs) that happen to be available for less than list price; Redfin’s agents and clients may have a tendancy to bid on properties that they recognize as over-priced, enabling them to savor the cachet of having “got a deal”; perhaps the negotiating gods of Seattle smiled benevolently on all brokerages with unusual names; or — not impossible, but definitely impossible to prove — Redfin’s agents could simply be better negotiators.But back to the numbers themselves. Per Redfin’s report, they started by cleaning up some dirty data caused by agents who missed that 5th grade math class that explained the importance of getting that decimal point correctly placed.Greg Swann brought to our attention some other order-of-magnitude errors, so perhaps the data set I’m working with was uncorrected, or perhaps Redfin simply didn’t catch these errors.
This chart shows the most egregious violations of good taste in posting bogus numbers on MLS’s. A property sold for over 1000% of the list price? A full order of magnitude? I think not. I shan’t embarrass the agent in question except to say that perhaps the “Ryness Company” Windermere franchise should have a math screening exam for its agents. The list price on this property was $36,030 and the sales price was $384,990. Clearly either the first is missing a zero at the end, or the second has an extra one.The second highest violator is a bit less clear. The list price is $100,000 while the sales price is $700,000. The most likely reason is a careless transposing of the “1″ and “7″.

The third one is likely missing a “1″ in front of the alleged listing price of $99,500…and down we go.

At the bottom end, we have the following numerically-challenged gems:

A property sold for one-tenth the list price? I think not. Again, an errant 0. The next one is most likely explained by a transposed “6″ and “2″.

And on we go.

Instead of second-guessing dozens of possibly wrong entries, I thought it most fair to just purge the worst offenders, so I chopped off the top and bottom 25 entries. Now we re-run the numbers, and — unsurprisingly, given the size of the data set — we get only a modest change in results, narrowing the “Redfin advantage” by five-hundredths of a percentage point.

That was certainly much ado about nothing.

Inspired by the James Hsu approach, I decided to take a more detailed look.

Redfin had 170 sales (representing the buyer) during the time period in question. How well did similar-sized and larger firms do? I separated out all the firms with at least 150 sales and then sorted them by average sell/list price.

The results will gladden Glenn, give Greg goosebumps, and madden Marlow.
Redfin came in at the top of the list with its average of 99.33%. Next best was Prudential Northwest at 99.89% followed in third place by a tie between a Windermere franchise and Coldwell Banker Bain, both at 99.33%.

At the bottom of the bigger-broker list was another Windermere franchise at 103.55%

Why are my results different from James? If I understand his methodology, he used the median instead of the mean, included more than just King County, and only included single family homes, not condos.

What about if we include all firms, not just the ones with 150 or more sales? Now Redfin slips down in the rankings quite dramatically to about position 390 out of nearly 1200 firms represented.

The very best firm in Seattle, if you use Redfin’s reckoning, would have to be the one that scored an average of 82.84% for its sales/list ratio. The catch…they only had 1 sale.

The firms above Redfin in this ranking tend to all be smaller ones, which makes sense — if you’re a very small shop and had only one sale, and that sale happened to have a good ratio, then by this definition you’re a winner!

In fact — and you may have to read this statement a few times to understand it — of all the companies with a better average sales/list ratio than Redfin, the average number of sales was only 5. The biggest company that bested Redfin did 109 sales.

The bottom of the list, as you would expect, is also dominated by one-hit wonders.

If you’ve made it this far, you’re probably wondering what to make of all this. What do the numbers show? A few things:

1. Depending on which metric, geography, and property type you choose, you may get different results.
2. The whole average “sales/list ratio” thing is interesting, but not necessarily enlightening.
3. To the extent that you take that ratio seriously,
   • Redfin not only did well, but in fact beat all other relatively large firms.
   • Of the companies that beat Redfin, most were one-hit wonders.

Now, off to dinner…

Tags: Real estate, Redfin
Possibly related posts

• Redfin Launches Consumers’ Bill of Rights
• A Year’s Numbers Come In From The MLS: Agents Who Take More Pictures Negotiate Better
• The Lessons Of Redfin, Part II: Targeting A Demographic Niche