Jakob Nielsen have since 1996 published the article "Top 10 Mistakes in Web Design", recently updated 2011. At the top of the list we find the mistake Bad Search. It has not always been that way. In 2005 it was listed as number 5, and in 2003 it was not listed at all.

From experience I know it is often much better to search using an external search engine with keywords and the extension site:theWebSite.com. This technique works on Google, Bing and Yahoo and most likely some more search engines. Problem is that it's probably not known and used as much as it could be. It leaves those users with either navigating the structure on the site or using the web sites internal search engine. If it is bad, the user will probably not find what she wants and move away from the site.

Now, Jakob Nielsen is not the only one expressing the problem with search. Peter Morville, author of Information Architecture for the World Wide Web, says that search is the "worst usability problem on the Web". Still "search" is one of the most used technique to find content on the web. I often see misspelled words that does not render a search result, nor a suggested word list (see example image below). BBC do suggest when I type though.

BBC search result on misspelled word

At the back end professionals struggle to improve search in different ways. Real-time display of possible keywords based on the characters entered in the search textbox is one. Improving ontology and taxonomy is another. Display suggested words when there is no search result, could be yet another way to improve search. But are there more to do to improve search?

  • 1
    +1 for the great question, yes their is a need to break the shackles in search modules!!.. hope to get good answers
    – sree
    Commented Mar 28, 2012 at 14:41
  • 1
    I suggest a tag for Web, if this is about search results that are web pages. The issues differ between searching within a site, an OS or even a program.
    – JOG
    Commented Mar 28, 2012 at 15:05
  • Search has improved by leaps and bounds in the last 8 years. Remember "search for keyword" vs "search for phrase"? Ugh. Perhaps how search has already changed would be a good question...
    – Zelda
    Commented Mar 28, 2012 at 15:20
  • 1
    There're literally books written on this subject. This question can't be answered. See Lou Rosenfeld's Search Analytics for Your Site and Greg Nudelman's Designing Search.
    – dnbrv
    Commented Mar 28, 2012 at 16:17
  • 1
    Agreed, "search" is a very complicated thing. It's how many people experience the web, it's on millions of pages. It has many problems and many solutions, I think this needs more focus. The question as outlied hardly even indicates what specific problems there are with search.
    – Zelda
    Commented Mar 28, 2012 at 18:20

2 Answers 2


If anyone had an answer that wasn't 'in a book' or 'copy Google' they'd probably keep their lips sealed for exactly the reason your questions alludes to: search is big business and still often a big problem, if you can make it better you can make money.

In the interim here's a few things that most websites are missing from their search functionality that Google currently offers:

  • Type aheads. They reduce spelling errors and help to offer search queries that you already have answers for.
  • 'Did you mean' text that offers alternative search queries. Often fixing spelling problems and, again, leading to queries you can recognise.
  • Multi media, multi-resource results i.e. images, videos, articles rather than just links to the articles on your site.

Working at a search-based company I know that it's painfully slow to update your search capabilities and introduce new features. It's not cost effective for smaller businesses or those that aren't primarily search platforms e.g. BBC. Having said that, lots of companies just use embedded google or bing searches whereby Google actually handles the search for them but only delivers results that relate to their domain name.

I'd say at present the best, low-cost way for people to improve their onsite search would be to use google's custom search https://cse.google.co.uk/cse/ and at very least offer results from their whole spectrum of content i.e. most business currently have their site plus social media accounts that have accessibly apis etc.


I agree that we can certainly improve search and a lot can be done to make both the technology and experience efficient and user friendly.

With current technologies

Given the current state of technology, I think the following should be included in search engines:

  • Auto-suggest
  • Better categorisation. Now, this can go bad if not implemented properly as discussed on this article at NN/g. This is part of scoped search and it can lead to bad UX if the search engine forces categories by default. It's useful when you use it limit the scope in case of too many results rather than forcing it on the user.

We can take a hint from previous version of spotlight.

enter image description here

The search query was categorised based on metadata. We replicated this behavior on our search engine and observed better findability from users. enter image description here


I believe future search engines should be based on Natural Language Processing (NLP) and Machine Learning (ML). Instead of looking for an item and then filtering it according to your needs, the search engine should do it in the background with the help of NLP and ML. Let's say, we're looking for a cheap thumb drive made by Transcend. In current search engines, we have to type thumb drive in search and then filter results based on our needs. Instead, we should be able to find desired results by typing - "I am looking for a cheap thumb drive made by Transcend". The search engine should then filter results in the background with the help of NLP and ML.

I think this should be the future of search. More human than machine.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.