The general thrust - that you use non-locational keywords in a geotargeted campaign, and use location-specifying keywords for national advertising, is pretty much OK.

Where this idea will hurt you, is if the locality that you are trying to reach is a smaller part of the target - for example, a Pizza delivery place with a six mile delivery radius, in a 20-mile geotarget, sees 90% of its adverts being sent to people outside the delivery area (pi * radius squared, compared for six miles and twenty miles; 20 miles because that’s Googles recommended smallest target area). Where the service area is significantly smaller than the geotarget, then you might want to use locality names to get better precision and higher click to conversion rates.

In your case, you also need to consider Display URLs and effective CPM. That is, if you run a Californian geotarget advert, and a national advert, with the same Display URL, with AdGroups for California and New York locations (dentist reseda, dentist bronx, etc) then which advert shows depends on which earns Google the most - so if the Californian geotarget includes phrase matched “dentist”, which would match “dentist bronx”, then Google will need to pick whether to show the national advert or the state advert - that decision depends on the effective CPM. The advert with the highest effective CPM will win, AFAICS - it isn’t to do with where you are, but what you are prepared to pay. If you had different Display URLs, then potentially you could show two adverts for one search - one for the Californian geotarget using a phrase match on “dentist” and one for the national campaign, New York/Bronx AdGroup for the exact match “dentist bronx” keyword.

Does that make it clearer? It’s a pretty difficult area, I admit, because you have to consider so many different parts of AdWords to put the whole story together :)

Cheers, JeremyC.

I would love your thoughts on this! Thank you!

Cheers
Davin
“The Davinator”

Cheers, JeremyC.

Quite often one of the locations being displayed was 10 miles from me when in fact it was around 120 miles away.

This really is a problem for locating ads where they are needed. It is not always possible to find space in the ad to place location.

Too many people look at anything published by Goog to be “bible”.

I’d like to know how your company deals with proxies that are used
between different types of networks, e.g. between a company’s
corporate network and the public Internet, or between networks running
different types of protocols. In the case of the former, it is like
the AOL situation, where a large multinational corporation shows up as
coming from a few locations, probably much smaller than the actual
locations. In the latter, for example, people on IPv6 networks are
using proxies to get access to content on IPv4 networks, and vice
versa. Ignoring such proxies (assuming you can identify them) would
seem to be eliminating potential sources of income, especially since
IPv6 usage is growing in Asia, as are the amount of investment dollars
flowing to companies that have footprints in Asia, because of the
(presumed) economic growth. But including them runs the same risk as
with the AOL problem; the traffic could be coming from anywhere; it
just appears on the public Internet at a few locations.

There does seem to be correlation between some providers and the
geolocation data they provide. This happens when the providers go to
the trouble to make this data public and reliable. Unfortunately,
providers tend to change hands, and IP addresses tend to be reassigned
within the new companies, so this data can quickly become out of date.

I’m not completely sold on the idea of geotargeting, issues of click
fraud aside. There still seem to be too many holes. If someone I
know wants to use it, I advise that they do a lot of testing in
advance, and continually monitor their campaigns, to ensure that they
are getting results that are in line with their expectations
(including their geographic expectations of where the ads show up).

The point about Proxy Servers is that Google has not made it clear to advertisers what they do… Do advertisers risk sending adverts to non-local users, just because a proxy is in the geotarget? It’s great that gelocation companies are aware of this… but what happens to my clients adverts?

AOL is a reducing issue… but AdWords and Google Search are licensed to AOL. So European advertisers still see an unusual count of US searches, from AOL proxies. If AOL were not so closely linked with Google, the proportion would be closer to national ISP proportions.

Navizon - cool - thanks. Again, the articles’ point is that Google hasn’t positively indicated, anywhere that I’ve found, that they do, or do not, use geolocation for mobile adverts. Mobile adverts have a different format, differently served within AdWords - so they could be better targeted, potentially. I hadn’t realised that costs had fallen so far.

I’m not convinced by the WiFi location you suggest. IME, WiFi hotspot providers use a common cloud of IP addresses, at least in the UK. So the location is national, rather than anything smaller - unlike mobile phone location which can be reduced to at least the size of a cell (or smaller, with the right technology in the base stations or phones). There may be differences between the US and Europe here, again, caused by the history of how the telco/carriers developed.

Cheap geolocation problems - absolutely. Actually the serious problem is tying a good geolocation service that isn’t Google’s, to a web analytics package that offers advert tagging and tracking… Otherwise you can’t verify delivery. And that the third party web analytics and geolocation service needs to offer data about use of proxy services, historical mobile location, etc…

Mobile location is a very nasty problem for retrospective targeting verification, as would be packet-time-based location services for improving accuracy of dynamic IP address locations. This (retrospective analysis vs increasingly mobile IP addresses) suggests that as more users become mobile or use dynamic IP addresses, that location services for analytics must be run in real time, to get the current location.

Thought provoking comment - really helpful. Thanks again.

1) Proxy Servers are tracked and available for use by companies that wish to keep them from being used in geotargeting.

2) The AOL issue is becoming a non-issue. AOL has been losing subscribers at an amazing rate. I think they have only 12 million users now. Couple that with the fact that today many sufers connect to it through a broadband ISP (not dialup) that is easily geolocateable on a sub-country level and the issue becomes much less pronounced. At the end of the day, companies like ours identify AOL networks and other types of proxies so that mis-targeting can be avoided.

3) Mobile geolocation is on the rise for mobile devices, allowing providers a way of geotargeting ads on a sub-city level, regardless of the network/phone. Check out Navizon, which offered the first version to work with the iPhone, with query costs much lower than the .20 you mention. As WiFi phones begin to penetrate the mass market, this technology (combined with IP geolocation) will become more ubiquitous for geotargeting.

4) Cheap geolocation varies widely in quality (especially at sub-country level) and should not be compared to results done by the big guys in order to measure geotargeting accuracy. In truth, the only way of judging true accuracy is by running them against a truth set audited by a third party.