Search UI Question

13 Feb 2012 - 9:47am
2 years ago
8 replies
1298 reads



Users can search a full-text collection that includes free and fee-based documents. All works have a document surrogate ("content detail page") that's visible to all users.


In search results, we don't want to show the snippet (keyword in context) for fee-based documents, because persistent users may be able to piece together what they need without paying. The current plan is to replace the snippet with "Matched on Full Text."


* Is there a better way?

* Are there any good/bad examples of how to offer a peek without giving away the store?


Peter Morville


13 Feb 2012 - 9:55am

Might each article have a short synopsis that could be used instead of keyword-highlighted snippet? Think of Google results, where it chooses whether to use text extracted from the page body or the meta description. In this case, you could use a short, curated meta description of each document for the fee-based documents.

"Matched on Full Text" tells the user why the document was retrieved, but doesn't aid the user in understanding what the document is. My hunch is the latter is more important.

13 Feb 2012 - 10:14am

That makes a lot of sense. Thanks Nick!

13 Feb 2012 - 10:53am
greg merkle

We are looking at 3 options in our search in Factiva.

1. Keyword in Context.  

2. Snippet which is a lead paragraph up to 140 characters.  

3. Metadata about the document and the noting that the search terms are "in there"   Our metadata contains Companies, People, Subjects, industry and Region.  In addition, we have author and source data as well.

The latter is important to the non-professional searcher as in many cases, the search is part of the "learn" rather than the "lookup"

13 Feb 2012 - 2:30pm

I did something similar to what Nick describes, with one minor difference: whether or not the excerpt is a synopsis, the link to the full text details from each search result takes you to a content "overview" page if you are not sufficiently permissioned to view the content detail page.

The service I was working on contained rather extensive and deep bodies of text, however. Revealing snippets based on slightly different queries was never likely to reveal enough content to fully answer a user's question. But "summary" access provided some level of satisfaction (so that a user could determine if the answer was in the text) while promoting the value of the content enough to prompt purchase among non-subscribers.

I suppose the other model would be similar to what and other paywalled sites provide: A limited pool of "complete" access, after which access is basically cut off completely.

14 Feb 2012 - 7:25am
Alan James Salmoni

This takes me back...

I did my Ph.D. around search engine usability and information scent. One thing I found that was counter-intuitive was that people made more accurate judgements about the relevance of a document to an information need with just the title - more so than the title and snippet! This was a significant difference and unexpected.

Someone else found the same thing after we did ( - this is a WWW 2010 conference paper from some people at Google). We also tested our finding repeatedly and replicated it even with subtly different study designs so I feel quite confident that it's valid. My guess is that titles (assuming it's author generated - this was back in the late 1990s when computer generated titles were less common) are tailored very high-level descriptions of a document's content. The snippet, however, identifies and shows search keywords which seems to communicate a false sense of relevance, hence less accurate judgements.

So maybe just the title is a good way to go?

The problem we also found was that users didn't enjoy using just the title. They preferred to be spoon-fed and travel through irrelevant content rather than stop and really think what a title meant. Lower user satisfaction probably isn't going to be a good thing even if people end up working smarter.

But since then, I've found the expert users (people knowledgeable in a particular domain) are so affected: they didn't seem so dissatisfied at having to think about a document's relevance.

Hope this helps!

14 Feb 2012 - 8:24am

That's fascinating (and helpful). Thanks Alan!

14 Feb 2012 - 11:18am
Christine Donahue

I am working on something similar, for a research journal. It doesn't seem as though our users would be able to piece together what they need from reading the abstracts of the articles though.

What I'd want to think about in your situation is how your users would be able to piece together the snippets? And if so, is there anyway to try and eventually convert those types of users to purchase eventually? Is that called a loss leader?

In my case, I've opted to include title, authors, DOI, basic publication information. I also offer a link to expand a "Quick Abstract", which would fold out only part of the abstract below the search result listing. I use a key icon to show that content is locked to non-members, and a link to purchase. There is also an executive summary, it's a pdf file, and it's locked to non-members. I don't know that the exec summary is very important to users.

A site that I used for reference was That's where I got the idea to use the Quick Abstract. Now based on Alan's post above, I will be really curious to watch if people actually use the abstract!

The ieee digital library is a new, and it's definitely worth looking at.




17 Feb 2012 - 8:23am

Thanks Christine! That's very helpful. And I expect I will push back on the snippet paranoia :-)

Syndicate content Get the feed