How well does a GWT webapp play with Google Web Crawler

[Draško]

I read the standard proposal how dynamic webapps should be made crawler friendly, but what if there is no server to generate snapshots?!

I built a simple multi-view webpage using MVP pattern and event handling. Unfortunately, the Google Webmaster Tool, as well as Google Analytics, do not see any structure and assumes it is just one page. Hence, the "visitors flow" view shows only one page (technically not wrong - but semantically). 

I tried to submit the sitemap but again, only one URL was recognized. As mentioned here, my assumption is that without <A> elements, it is not possible for a crawler to understand the structure. Unfortunately, the Anchor and Hyperlink widgets do not accept widget children (I need Image widget to dynamically reflect a state change), so I can't use them.

Please share your experiences and best practices in making simple (no-server) GWT webapps understood by Google Web Crawler, Analytics and Webmaster Tool.

[Raphael André Bauer]

It's simple:
- If there is no content on your plain html page there's no content
for GoogleBot (or any SearchBot of the big search engines to date)
- If you use Google's Ajax crawling approach you can turn a "blank"
ajax-only page into something the GoogleBot can index. Btw - Bing is
also supporting that spec. Downside: You need your own HtmlUnit that
turns ajax-js into html.

We are currently experimenting with this, but I am not yet sure if it
makes sense to use the Ajax Crawling approach for a site that heavily
needs good SEO.

A big blocker is also: If (and nobody knows really if they do ) - if
Google takes into account the speed when crawling sites via
_escaped_fragment_ it will almost certainly destroy any SEO ranking -
because crawling an Ajax site by your own HtmlUnit and returning the
result will be slow by definition.

Not exactly a GWT problem. But a problem.

Hope that helps,

Best,

Raphael

[Draško]

thanks for your reply raphael ... the problem i have with the "standard google's ajax crawling approach" is that I do not have a server logic, so no means to generate a servlet mapper that will spit out static versions of gwt webapp.

A friend of mine made a suggestion that the webapp should be done just in html, without any dynamics, and then enriched with the javascript (gwt). I am wondering what could be an elegant way to do such a design, avoiding any duplication of content and logic. Ideally, the content could be put in html and gwt does the voodoo ;-)