Database Reference
In-Depth Information
Gutierrez:
The data sets you ingest seem pretty straightforward other than
the movement data. Where does this data come from and how do you use it?
Lenaghan:
The sources of our movement data are ad-request logs. So even
though our geospatial analytics platform is very horizontal, the main vertical
we've been working in is mobile advertising. We're now moving into consumer
insights and building up our geospatial analytics platform vertical-by-vertical.
Gutierrez:
So the idea is that, as I'm walking around with my mobile device,
I should get the right ad for the right product or service at the right moment
and location.
Lenaghan:
That's exactly right.
Gutierrez:
Could you describe a particular product you helped build?
Lenaghan:
The first product we launched into the ad tech space was
called Audience Now. It contextualized what was happening within a single
100-meter-square tile so that our database could identify its primary demo-
graphic and psychographic characteristics with high confidence—so we could
say, for example, “This tile index is very high for the 'shopper mom' type.”
Then, if your mobile device moved into that tile, it would be subject to “shop-
per mom” targeting. Audience Now was purely location-based. We didn't do
any device tracking or anything like that.
Gutierrez:
And how has this product evolved?
Lenaghan:
Subsequently we started to ingest a large amount of ad-request
data tagged with location, time, and device ID. Device IDs are obfuscated
through persistent hashes—so even though we can't tie back a given device to
any PII [Personally Identifiable Information], we can contextualize the move-
ment data. We can see that a particular pseudonymous device ID has been on
a golf course a couple of times in the past month and tends to dwell in tiles
that have a high score for affluence. So we might tag this device ID as “golfer”
type. The big push over the past year at PlaceIQ has been to target devices
not only by location but also by the contextualized location histories.
Gutierrez:
This seems very different from algorithmic trading. Why did it
interest you?
Lenaghan:
When I first came to PlaceIQ, I wasn't necessarily interested in
the ad tech space per se, and my background in equity trading and physics had
given me no specific expertise in mobile advertising. The mobile ad market
does display certain systemic similarities to the equity market. Both involve
low-latency, high-throughput systems, networks, and exchanges. I expected on
entry that my work in the ad-tech and equity market ecosystems would be
very similar, but the analogy broke down almost immediately. Instead of work-
ing on a lot of real-time bidding at PlaceIQ, I actually work more on geospatial
location analytics that use machine learning and heuristics to build audiences
to target mobile ads.
