I've worked with sites hosted in Canada and the UK that have great penetration into those markets but its only 30% to 50% of their site's traffic respectively. Even with good market penetration, the small market size can limit the amount of traffic available.
For these sites, US market share is less, but US traffic still accounts for another 30% of their traffic. Its a tough call. If you move the site to the US, you are targeting a broader market and have more room for growth. On the other hand, its more competitive and you'd almost certainly be throwing away the traffic from the country where you had been hosting.
If the site is UK hosted and the UK traffic is 5% and US traffic is 80%, the answer is clear: move to the US. For the clients I've had, it certainly isn't that clear.
The site is exclusively targetted at the US. And it ranks in the US, pretty well. Traffic from other countries is not a consideration, the product is only available in the US. So they don't care about losing non-US traffic. They are only concerned about how big of an increase they'll get on US traffic, and any risk of losing existing US traffic.
It's a reasonably big job to move the site so I'm trying to evaluate the benefits and risks for them.
The non-US sites I have worked with have market share 1/10th of the US hosted sites in the same vertical and 10 times the UK/CA market share compared to the US sites. All these sites are happy to service traffic from any English speaking country, so the the situation is a little different. In my experience hosting location (without setting webmaster tools targeting) can be a ten times factor.
So if they move to the US,they may expect a multiple traffic increase? Holy crackers!
I would qualify that in two ways:
1) The sites I worked with tried not to set geo targeting. They were .com sites. A UK hosted site that sets geo targeting to US, may already have many of the advantages of a US hosted site.
Come to think of it, I have a counter case. I worked with a .co.uk site that was hosted in the US. They couldn't get any UK market share with that setup (and weren't trying for US market share). In January 2009 Google made some algorithm adjustment and the site took off like a rocket in the UK. It started gaining UK market share just like a UK hosted site.
2) I wouldn't expect traffic to increase right away. I would expect the rate of growth after a year or two to be greater if the hosting matches the target country.
There is also the matter of response time and user experience. The US hosted .co.uk site I mentioned has always been at a disadvantage compared to its competition in terms of response time and usability because the traffic has to traverse the Atlantic. I believe that Google is including user experience metrics more and more into the rankings and hosting close to your audience is the type of thing that could make a big difference.
Reducing latency at both the DNS query and server request/response should result in a better user experience.
As Chrispcritters said, the latency for US accesses will be much lower, resulting in faster access for US visitors and most probably gaining better response and conversion rates.
This itself should result in improved rankings as the time-on-site will increase.
This is a very interesting question and one we've been discussing this week since it is very noticeable that post-Panda in the Google.com results we are only being beaten by US hosted sites whatever their quality.
Coincidence? Maybe, maybe not, but it is also noticeable how many US hosted sites are now appearing in Google.co.uk that do not target the UK for business.
Is anyone else seeing this?