This is literally an advertisement for Algolia masquerading as an article. They literally went through each feature their product provides and listed it out. Going to their page, the animation is practically a point by point repeat of the "article."

Maybe it's okay because it's a YC company that it's also posted here, but it really just leaves a bad taste in my mouth.

Did the article do a bad job talking about the UX of search boxes? Did the author try to hide their interest in the topic?

I really don't see the problem here.

It's the fact that the blog:

  1. Lists only features of their product.

  2. Doesn't lists any features that their product doesn't support.

  3. Aims mostly at shaping expectations so that one would find their product.

  4. Uses the guest-blog trick to boost rating / cross-linking.

It's is all fine and dandy in the culture where somewhat things are newspeaked into hacking growth or whichever new euphemism is in vogue.

If they knew about some other feature search-boxes should have, but which their product doesn't, don't you think they'd add it to their offering? #1 and #2 being true is the equilibrium-state of building a product to address every problem you're currently aware of.

> Why not skip this step and offer real results as they type?

This is one of the most annoying thing for me for general search(if it's something very precise, OK). How annoying it is seems to be counterintuitive, since a lost of search components do it.

There can be half a dozen of UI elements popping under a text field. Autocorrection is one, autocomplete and input history are another two, autosuggestion can happen depending on the device (Os5~7 Blackberry phones heavily did it). IME is also right on this spot and individual keystrokes don't have much meaning in this case anyway.

It's usually sad to have two dropdown like components auto updating at each keystrokes, especially when there is a network request for each update on the one hidden below the other. It can wait for an ENTER key, really.

In pretty much every case where I'm searching for something with a search bar, I prefer the list of results getting updated as I type. Why? It's simply faster.

On Spotify for example, click "search" start typing until I see the band I want in the "top result", press the down arrow, press enter.

If I had to press enter and wait for the whole list of results to load, that would take much longer.

For me the experience fall apart at the "press the down arrow" part.

I took a screenshot of a typical google search session [1] with the search suggestion appearing below the system's native dropdown. For any field where I use the IME, it's the IME that wins, and the dropdown in the background is just distraction [2].

This is the easiest case to reproduce on a blank site, but it's the same for search box where I have autocomplete values (which is a more common problem I think ?); the autocomplete popup will take precedence on any element displayed in reaction to the keyboard, and a "press down arrow" will go to the autocomplete, and not the site's dynamic dropdown.

[1] http://imgur.com/YikLtdi

[2] The other "best" case is when a result is actually relevant, but I can't see it because it's hidden, or it changes/disappears when I dismiss the native popover because the action counted as a focus change or a keyboard event.

Ah, I guess my experience is different. I don't get that "system" autocomplete on Google or anything else. So, i just use the one that the website gives me.

If you had to prioritize these, which would be your top three?

Mine are "Search across multiple attributes", "Ranking" & "provide real results from the first keystroke", in that order.

Would be great to know what everyone else thinks, so we can improve our service accordingly (https://starthq.com - Greplin-like cloud search).

I think the highest priority needs to be "Put the results the user is looking for at the top of the list". Second might be "Don't put anything on the list unless it's likely to be what the user is searching for". I don't know if there needs to be a third, but if there was it would be something like "Do it fast and unobtrusively".

Applying this to the article, let's look at the examples in #6. Correcting "Gmae of Th" to "Game of Thrones" is great, but what are the chances that the user would have typed "Gmae of Th" if they were searching for "The Hunger Games", "Enders' Game", or "The Game"? I'd argue that it's close enough to zero that the last 3 choices shouldn't be on the list.

I would rather it search for "Gmae of Th" because that's what I typed into the box.

I made this one last year https://bitly.com/18y2kPE Does a few of those things albeit not all that well. Also due to Google charging for its API's it only can do 100 searches a day from them but Bing and Blekko keep working.

Careful, haven't you guys heard that Matt Cutts has banned guest blogging? [1]

Those links to Algolia should be no-followed under decree from Google.

[1] - http://www.mattcutts.com/blog/guest-blogging/

I find it weird that this is not a blog post on Algolia's website, since it is so closely related to what they do...

This is a pretty tried and true marketing tactic for brand exposure. "Cross-blogging" or "guest blogging" lets them illustrate their market experience to a new audience, vs. the audience currently visiting their site, who presumably already believe that they're experts.

Good points. Case study: Verizon FiOS DVR fails miserably at #3, meaning you have to type out more letters using a d-pad.

I'd never heard of Algolia before, but their demos seem very impressive.

Fun trivia, the search engine for HN have been switched out to a Algolia one. Try searching in the bottom.

It's a pretty good one as well.

Frankly I was blown away a little while ago when I first used the new search, it's lightning fast and really finds whatever you're searching for. I'd really love to know more about the infrastructure behind.

We have done an interview with Leanstack.io where we discuss our infrastructure: http://blog.leanstack.io/how-algolia-built-their-realtime-se...

Thanks ;)

Some interesting info here: https://news.ycombinator.com/item?id=7456827

What about the patents on search results.

> It’s hard to imagine a product without search

I don't want to nitpick, but when an article begins with such an evidently silly statement, it discourages me from reading the rest.

I don't whats silly with this. I agree that most products where you put data, there should be a search. Not always, but in most cases.