Thanks for questioning my data. You are right, and I was almost going to give you the “winner winner chicken dinner” award.

While a good observation, here is the flaw in my own data that I was waiting for somebody to call me out on…

How in the heck can you (meaning me, Frank) use Seattle area data, average Realtor negotiation ability (% off list) and throw that number in when comparing Virginia average Realtor negotiation ability (% off list).

Those are apples and oranges. You can’t just cherry pick data points to make something look good.

For example in some states it is normal to have a high % off list. It doesn’t mean that the entire state negotiates better, it is just different. Some places bid up a ton. For example I looked at bank properties in Price William county. Half of them went over list (conveniently for my current point, not counting subsidy).

So my data, regardless of all the disclosures saying the numbers are rough, they are flawed. They are “correct” in one sense, but flawed in another.

Redfin’s claim of being 1% better negotiators, while disclosing that it doesn’t count subsidies, in Seattle *might* be “correct” in one sense, but in my opinion is it flawed.

I hope my message didn’t get lost. I will try to recap the guts.

1) When you claim to be better negotiators by 1%, when in reality that data is probably off by 0.75% NET due to a lack of seller subsidy data, it is irresponsible to use that data, even if you disclose “results omit subsidy.”

2) Data is too easy to manipulate to be useful. For example, if your company doesn’t do bank properties or short sales, the data will be way off. Regardless of disclosures of missing data.

For example, in theory, IF their data showed they were 10% worse… would they publish that data? Probably not.

Thanks for your comment, but you could have done a better job busting me. So I thought I’d do it for you.

Frank

People are reading (I hope).

Good thinking to clean up seller subsidies that are a result of typos. (ie putting in a subsidy of $100000 vs just 1,000)

I look forward to seeing your report.

Frank

Earlier this week, Redfin ran an analysis on every 2008 transaction in MRIS for Montgomery County, Fairfax County and Washington DC, excluding transactions where there was an obvious data-entry error (final prices 50% above or below list price; seller concessions that were greater than 15% of the final sale price).

This approach is better than the one taken by Frank because it excludes seller subsidies that are the result of a typo, and because it takes into account every seller subsidy, not a sample of seller subsidies from one day or another.

We compared the discount off list price negotiated by Redfin agents to the discount off list price negotiated by all other agents. Redfin negotiated a larger discount. We also calculated this discount net of any seller subsidies, for both Redfin agents and traditional agents, and Redfin still negotiated a larger discount. This negotiating advantage is over and above Redfin’s commission refund.

As Matt Goyer explained in responding to your original comment on our blog, we never previously claimed a negotiating advantage in the Washington DC area because we had not been open for business in DC for a full year when we published the 2008 results. In every market that we did analyze, data on seller subsidies was not available. We acknowledged this as a weakness of the analysis of those markets.

While our preliminary findings indicate that there is indeed a slight negotiating advantage for Redfin in the Washington DC area, we would prefer to publish these results in the context of the broader national report we prepare each April. By waiting, we can be more careful about which claims are statistically significant.

Hope this helps to clarify the issue, though I doubt at this point anyone is still reading the comments on this post.
Regards, Glenn

Glenn Kelman
CEO, Redfin

I don’t follow.

But that is ok. The point of the post is that numbers and averages like this don’t show much in terms of “good negotiator”.

Frank

- Not Robert