On Hyperlocal Targeting
If you are interested in real estate and marketing, then you need to check out Real Estate Relativity. The author is an expert in matters of database marketing, social media, and real estate — and it shows.
In one of his recent posts, Mr. Relativity (I’ll let him reveal his identity himself… but he knows of which he speaks) wrote about hyper-targeting of enhanced listings:
Trulia partnered with 1020 Placecast to provide targeted ad services.
Once users input a location they want to learn more about on Trulia, Placecast will access that data and apply it as a key component along with common demographic data points like psychographic information to provide more targeted ads.
This process makes sense especially at the zip code level (see previous posts on zip code optimization) because demo/psychographic differences exist between zip codes–even contiguous zip codes. Accordingly, if I’m looking in a zip code that trends more affluent, Trulia can now serve ads that appeal to an affluent consumer (Jaguar advertisement). Alternatively, if I’m searching in a zip code that trends more middle of the road, Trulia can now serve an ad that appeals to a bargain shopper (Toyota Corolla advertisement).
I think this is right on, but I feel it would be appropriate here for us to once again preach the gospel of the Human Touch. The only gloss I would put on what Relativity wrote is to point out that human beings typically do not think in terms of zip codes. They think terms of neighborhoods, or towns, or real places.
Fact is, demo/psychographic differences exist intra-zip code as well, depending on whether there are multiple neighborhoods within a single zip code. But beyond that, when thinking about targeting ads, the geographic limitation imposed by arbitrary zip codes is pretty significant. Far more effective, we feel, to use neighborhood boundaries (which are themselves psychological in nature) or real places, such as towns.
But the real insight comes from Relativity’s next thought:
For real estate, I’d like to see a twist on this process: somehow also deduce from where a consumer searches so as to better deploy advertising resources with respect to select properties. For instance, let’s assume you’re a firm situated in a Utah ski resort community, and that you know based on previous dealings with out-of-market buyers that your to primary “feeder” markets are Chicago and Orlando, and that these primary markets are generally interested in purchasing luxury-oriented rental income properties.
It’d be a great service to be able choose which of your top properties to “enhance” that exist in a specific zip code and display the “enhanced” versions of these properties only when a consumer from either Chicago or Orlando conducts a search in the targeted zip code. Employing a scheme like this, one makes an ad buy based on a “known” marketing attribute (i.e., based on personal experience) along with hyper-targeting, which should translate into higher quality clicks to the “enhanced” properties and, thus, increase the potential ROI on those ad buys. (Emphasis added)
I have two thoughts about this.
First, consumer searches in real estate today cannot deduce anything at all. The reason is that today’s search — no matter how slick and how sophisticated — is still property-centric. It’s all about the property, its location, its features, and its price. The standard “zip, bed/bath, price” model is how a computer database stores information and processes queries. It has no relationship to how an actual human being goes about thinking through a search.
What is required then is a new model of search: a human-centric search that starts with the person doing the search, rather than the properties that are the targets of that search. We will have more to say on this topic.
If you start out with the focus on the person doing the search, then all of the requirements/factors above fall into place rather quickly. Rather than knowing from “previous experience” that the feeder markets of out-of-market buyers are Chicago and Orlando, the firm would know from the actual search trail that this particular consumer is hailing from Chicago. The firm would also know that this particular consumer’s search parameters dealt with skiing, horseback riding, and snowmobiling, because the search itself begins with a concern for the consumer’s preferences, rather than the “zip, bed/bath, price” paradigm.
Second, the process is inherently iterative. Because human beings do not search with “requirement X, Y, and Z” in mind. They start broad, then narrow down. They might start off thinking, “You know, I really want a chalet right on the mountain.” Then once they see the prices of such chalets, quickly change their minds to, “Nevermind — within 15 minutes driving distance is good enough.” Once confronted with a large range of choices, they could then refine their thinking even more: “It’d be great though to have a fireplace….” And so on.
The good news is that there is technology already available that does exactly this sort of iterative profiling of consumers. One such company is X+1, whose product, Site+1, is precisely this iterative consumer profiling mixed with psychographic data. [Full Disclosure: X+1 is not a client, not a partner, hasn't paid us, or me, and we have no business relationship with them whatsoever. But I do have a friend who works there, which is how I know about them.] While this technology is relatively expensive, and is currently used mostly by Fortune 500 companies, we know that it is possible.
What that means, then, is that hyperlocal targeting is not a one-and-done tactic, but a continuous, iterative thing for a website that is built to do it. Profile your $15m condo, together with two others that are more modestly priced. Depending on the consumer’s next move, you can showcase a whole new set of enhanced listings. Then based on the next input, showcase something else to boot.
This level of control is possible today. It remains for someone in real estate to put it all together.
I love that Real Estate Relativity is discussing issues like this, because it really opens the door to what consumer benefits — as well as increasing ROI — is possible leveraging the power of data and technology. My view is that the more the real estate web focuses on becoming more and more humanized, the better it will be for consumers and service providers alike.
-rsh
Tags: hyperlocal, neighborhoods, psychographics, X+1.
on Dec 4th, 2008 at 10:48 am
rsh, thanks for the compliments, and I can’t wait to see what Onboard will deliver to market next year. With respect to zip code specification, I have the privilege of tracking–at the national level–hundreds of thousands of search strings every month, and a surprisingly high number of searches include zip codes in both short and long-tails search queries. This is not to say that neighborhood, town, and city block specificity is lacking, though, as you point out. And in terms of “deducing” the originating location of an in-bound search, I was remiss in not clarifying my thoughts, which were that a firm could automatically perform a reverse IP look-up of the in-bound searcher and make an assumption as to the originating location–at the ISP level at least–and deliver a more targeted offering based on that assumption. Admittedly, this is a clunky process—and inaccurate at times–and it seems that a more elegant solution exists with Site +1…thanks for the lead on that product!
on Dec 4th, 2008 at 6:10 pm
[...] post on real estate brokerage future and this one on hyper-local targeting are two excellent discussions about the strategic decisions real estate brokers will face over the [...]
on Dec 4th, 2008 at 6:23 pm
Hi Eric,
Thanks for the comment — and the awesome article. Seriously, I have to think about your response on Relativity — some great thoughts there.
Quickly on the zip codes vs. neighborhoods, I’m certain that today’s searches are based on zips in a surprising way. I wonder, however, how much of that is because of the way consumers have been trained to think along zip codes, and by the time they get to conducting property searches, they’ve already selected the towns (and the attendant zip codes) they’re interested in.
Once search technology improves, I believe consumers will naturally gravitate towards natural language search that mimics the human mind more accurately. It can’t ever be perfect (I don’t think) but today’s search is based on databases, not people.
Just one thing about X+1. Based on what my buddy tells me, Site+1 goes beyond the reverse IP lookup. It does that, of course, but it then combines that with PRIZM data on customer psychographics, runs things through some black box with their proprietary targeting algorithms based on things like time of day, IP address, and comes up with anonymous user profiles. Critically, it then continually iterates the profile based on that anonymous user’s interaction.
I was totally jazzed about the product — pushed hard for it at Realogy when I was still there. But the downside is that the product is rather extraordinarily expensive. For now.
-rsh