This List is in no particular oder, as all are relevant.

• Check headers for meta information such as index and follow. It may surprise you how some people forget to check data which may be default, and particularly to c.m.s.'s.
• View source code either with a browser or in the "Back-end".
• Run the web page through Valid Html, not just to validate the code, but to help discover inadvertent mistakes. ( Also helpful thereafter, to quickly discover unwanted injections which may impact on security ).
• In todays search engine climate, ensure that the old meta description (which could be more or less, ignored in recent years), provides a useful description of that particular page, prior to a click through. Not garenteed, but you may notice over the years that when financial resources are either tight or directed elsewhere, then a Search Engine may decide to some "weeding". As with Title Tags, the description mustn't be similar to that of another page on the same site or domain. This will / could, be viewed as Duplicate Content.
• Run individual pages through the likes of http://www.sitereportcard.com , to see their analysis ( don't take everything too literally like misspellings and their take on what is correct with respect to keword density ). Use it more as a quick guide, though their image optimisation options can be very useful. Load time, is a big factor too, as broadband is fairly widespread in the Western World, and potential punters won't wait long for a page to load.
• Urls. Try to include the relevant on-page kewords into the url. With a C.M.S, try to find some url rewriting method. With the trusted plain old html, name the page "whatever-it-is-about.html" or with the .htm extension. There is nothing inappropriate with Plain old Html, and it is often simpler or more straighforward.
• Duplicate urls. With a cms; limit or exclude entirely, the likes of addresses dedicated to Printing, Email etc., unless your are satisfied that there will be no duplicity inferences. If in doubt; abandon it or seek help on the issue. Where competent, noindex or nofollow, may remove this possibility.
• Unique Content. There must be millions of webpages which are quite similar. Similar and the Same though, are two different things - the bread and butter of Solicitors, semanticists, lawyers and scientists. Try to isolate one topic (and its individual optimised phrases), to a particular page. That said, bits of all pages, can be lumped together on something like this page or a FAQ's page.
• Updating. Old news, is invaribly; just that. History, is some what a niche market, rather than the bastion of a well adjusted searh engine. The age and percieved suitability / or relevance of a document; may have a  time base.  Sometimes, this can be seen (and recorded) in the likes of a sitemap. Does this matter? It depends on the domain. Interestingly , a document may have headers which send incrrect timestamps.Nothing too weird on this,  as adjustments for innacuraies are made. Never the less, if money is tight; then a business (SE.) may try to meet its commitments to its shareholders.