I’d add a few more that I’ve found to be important and inspired by some of the nuances of Google’s UI:

1) I’d modify “Allow sorting and refinement tools if appropriate for your users and content” to reveal some hierarchy within the results themselves rather than cluttering things up with more options, when possible. Do this by building a bit of default sorting with sublists and grouping related items together.

This sounds more complicated than it is and is visually simply to understand. For example, if results two and three are subsections of the first result, then indent them and move closer to their “parent”. A good example is an ego search for my name – the first two results from my site are grouped together, then the next two from flickr, then two more from some w3 list I posted to ages ago, etc.

2) give sneak peeks of other types of searches for the same term by giving snippets above the “real” results and a link to more, or even modifying the look of a single result to get across that it’s a different media. For example, searching for Obama will have the results you’d expect and also some snippets of recent news, a thumbnail and rating next to the youtube clip, etc.

This last one is actually more relevant than you might think, this isn’t just a problem for search companies with various flavors of engines. Depending on what kind of site you’re building or working on, it’s worth questioning when a page of mainly text with a long list of links is the best fit for the type of thing being searched and what to do when it isn’t, whether to make modifications to a single search result UI to highlight the differences or have a separate type of search for a particular type of data rather than trying to have only one search result UI to rule them all. Context over consistency – sounds simple, often isn’t

That’s a good list, and it certainly matches our experience. We’ve also written a little on this: What to include in search results. That says pretty much the same thing…

Cheers, James

We have the following list for our results, I’ve generalized it to match a general SERP instead of an specific one (docs, products, etc).

1. Navigation Artifacts
BreadCrumb: Searched Terms + Categorie or paths followed (via facets or traditional breadcrumbs usage depending on the artifacts placed to filter and navigate the SERPS)
Search form with searched terms
Number of items/pages in SERP

2. Specification Artifacts
Clusters and/or Filters to narrow down SERP

3. Related searches
Semantic or grammatical corrections suggestions
Releated search terms (to narrow via specification SERPs)

4. Manipulation artifacts
Sorting
Comparison of results

5. Expanded navigation artifacts
Visited results
Related results
Related goals
Advertising spaces

6. Results artifact
Depends on what you are implementing. But in every problem we have faced on our different portals we decide to select data that can help user understand what she would find out on the result once it is clicked and help her decide on which results might prove interesting goals to follow and which don’t. For example on a retail site we’ve decided to place a brief item description, photograph, geographical location and price, since we have been able to discern through ethnographic studies this are the variables that our users use to inform their purchase decision.

7. Alternate revenue artifacts
Result highlighting or special treatment
Advertising spaces

We also take into account the default sorting of the results, and it varies on what kind of implementation you are working on, always treating only the results that correctly matches user expectations. On some sites we’ve decided to give better placements to the results that have payed for special treatment, results through which we know we can have higher take rates, ranking based on proximity against search expectation, etc.

Thanks for the list,

Implementing changes on my current design now!

…prefilled with the query just run so the user can edit, rather than starting from scratch.