Does Sitemap show SERP ranking without "sandbox filtering"?

[gellydonut]

I recently restrutured a relatively new site so that major categories have their own subdomains. Almost immediately, these subdomains ranked surpisingly well in the SERPs (top 5 for major keywords) then dropped to the 50s after a few weeks. Classic "sandboxing" that I'm not particularily concerned about.

What I find interesting is that the statistics shown in my Google Sitemaps account still show these subdomains as ranked in the top 10 for the keywords averaged over the past 3 weeks despite the fact that they have actually appeared in the 50s for over a month. Is it possible that the the Google Sitemaps statistics show the position these subdomains would/will have without the sandboxing effect?

[tedster]

That's an interesting observation, gellydonut. Let us know if it changes anytime soon, please.

[ZoltanTheBold]

I don't think so. It does the same for me, and I'm definately not sandboxed. It seems to imply the position is some kind of average, although I'm not sure of the timescale.

I can't work out what is going on, so my deeply uninformed answer would be it's just crap.

[gellydonut]

Thanks Zoltan! I have to admit, I'm leaning towards your crap theory and, the way things have been going at Google lately, parsimony seems to agree.

Nevertheless, it's hard to give up hope of some insight into the esoteric workings of Google and I thought I'd post the devilishly dull details to see if anyone else has experienced something similar - or at least to record an SEO experiment.

Premise
The site mywidgets.com is left to brew online for one year with keyword rich text and minimal link gathering. Home page achieves PR5 (in all but 4 datacenters which reflect PR3) but no impressive search rankings. Directory structure is dynamically generated and sloppy using a mywidgets.com/keyword-x format.

Week 0:
Restructured the site using subdomains: keyword-one.mywidgets.com, keyword-two.mywidgets.com, etc, accepting the possible penalty for the (minimal) cross-linking of subdomains because I sincerely believe it makes for a more user-friendly experience. Actual code is an SEO wet dream of valid, best-practices standards-compliance (1400-1600 words per page, 5% keyword density concentrated in heading tags, title, and internal-linking anchor tags). While each subdomain contains content unique to its keywords, the structure of each is identical and makes for great empirical comparison.

Week 1:
For less competitive keywords (representing about 10 subdomains), each subdomain is immediately ranked in the top 10.

Week 3:
Subdomains drop to 30-50 for targeted keywords.

Weeks 3-7:
Intensive link gathering for "keyword-one." keyword-two.mywidgets.com, etc are left to languish as a control/because I am lazy.
Following the April 4-ish PR update, keyword-one.mywidgets.com reflects PR4 while other subdomains remain PR0.

Week 8:
In Google Sitemaps, all subdomains reflect accurate rankings (30-50) *except* keyword-one.mywidgets.com which shows a top 10 ranking for all major keywords. Google Sitemaps claims these ranking statistics are averaged over a three week period yet keyword-one has actually shown up in the 30-50 range for almost 5 weeks (although SERPs rankings have started climbing slowly in different datacenters, up 10 positions in the best cases).

I track 6 other sites using Google Sitemaps and all other statistics are accurate +/- 3 SERPs positions. I'll keep my eye on things and post any interesting future phenomena.

It probably is all crap and I definitely need a new hobby but thanks a million for your responses, guys!