A Search Is Not Just A Search
For nearly fifteen years, search has been stuck in a pretty familiar presentation mode: A blinking cursor inviting your query, and after a split second, a refreshed page featuring ten blue links, more or less.
In the past few years, a significant new feature has crept into the results portion of this otherwise predictable interface. Called “universal search,” the idea is to incorporate more than simple HTML pages into the results. A search for “London restaurants”, for example, might bring up maps and local results, as well as videos, images, organized reviews, and of course web pages. Every major search engine, from Google to Ask, has incorporated some kind of universality into its search results.
But while universal search points the way toward a new approach to getting you the answers you seek, it’s a half step at best. The results change, somewhat, but the process is pretty much the same. You enter a query, you get a set of results. Not particularly new.
What I find interesting are entirely new approaches to the interface of search. What happens when search is no longer driven by the command line and the blinking cursor? What happens when, for example, your query is informed by where you happen to be, or who you happen to be, or what you happen to be doing at the time of the request? To explore these ideas, it’s best to step outside the current box of a web browser on your PC, and think about mobility.
To help folks think outside the PC box, I often use the example of buying wine at a supermarket. Imagine you’re in front of the wine aisle, staring at all those expensive bottles. You choose one at about eye level, knowing that the higher a bottle is, the more expensive it becomes. Are you getting ripped off, you wonder?
This is where search gets interesting. Instead of noting the vintage and vineyard, driving home, and Googling the wine, you whip out your phone. You point your camera at the label, click, and – voila – that’s a search. You turn your screen back to your gaze and there are your results: “2001 Stag’s Leap Cabernet. Avg price: $48. Avg. rating (friends): 8 of 10. Avg rating (Wine Spectator): 90 of 100.”
You turn the bottle over in your hand, toward the back, where the price tag is. $76. Yep, you’re getting ripped off.
But that’s not all. The results include these links: “Click here to find this product near you. Click here to buy. Click here for delivery options. Click here for detailed history of this product’s ecological impact.”
You click on “find the product near you” and see that it’s available, for $45, at a nearby liquor store. And for a few bucks, they’ll even deliver it to you. A few more clicks, and the bottle is on its way to your home. Satisfied, you place that $76 bottle back on the shelf and saunter out of the store.
Now that’s search!
Is it possible? It most certainly is. And it’s not far off. But the point of this story isn’t to show you how you might save a few bucks – it’s to demonstrate how flexible and powerful search can become when it is unhinged from the standard approach to which we’ve all become accustomed. When it comes to search, change is most definitely good.
May 8th, 2008 at 1:10 pm
“In the past few years [ ... ]”
Excite’s channelized search did this ten years ago.
http://searchenginewatch.com/showPage.html?page=2206141#hotbot
That aside, I think the web search metaphor has legs. A text box that gives ten blue links is like two wheels, handlebars and pedals: a useful thing that we don’t need to replace. New kinds of search will arrive, but like the skateboard and the segway, they won’t replace the bicycle.
May 8th, 2008 at 1:29 pm
Sounds fascinating John, but that would mean that you’d need either a text reader that could tell apart logos from actual letters. It would be sweet if the cell phone could scan the bar code on the product. Then again, what if in the future there aren’t any bar codes an everything is wireless, that way you’d just pick up the product and get it close to the cell phone (asi if it where a bluetooth device) and poof! You got your results. Pretty neat!I honestly think search will branch out but the classic box interface will remain as a backup when all else fails.
May 8th, 2008 at 1:36 pm
Sounds great John, but that would mean that Google’s servers should be equiped with character readers (or whatever they’re called) that could tell apart actual letters from logos and designs. Wouldn’t it be sweet if the cells could read bar codes? Or better yet I’ve heard that bar codes’ days are numbered and that products will hava a wireless type of coding. Now, that is a whole new ballpark because cells could interact with this like they do with bluetooth devices. Anyhow, I see where you’re going with this and agree with you. Although I believe the classic box interface will remain as a backup when all else fails.
May 8th, 2008 at 1:37 pm
If I may, I’ll plug our barcoder reader here:
http://code.google.com/p/zxing
It will read UPC barcodes. Other readers can do this too. So that piece of your dream is a reality. (Well, you need a decent auto-focus camera to read 1D barcodes, note.)
The reader sends you off to Google Product Search or upcdatabase.com. The real problem is lack of consistent data of the type you want available for all UPC codes. But that will get better over time.
May 8th, 2008 at 2:39 pm
Hey John - it’s not only possible… it’s here. As discussed in this post on our corporate blog, companies like Vodafone, SnapNow, and Mobot have made visual search a reality. Even Google’s getting in the act with its Neven Vision acquisition. Will be interesting to see how visual search unfolds. I think we can all agree that the scnerio you draw up here would be a big step up over current search functionality.
Cheers,
AG
May 8th, 2008 at 3:20 pm
[...] of search. They’ve given us no guidance, just asked us to ponder the topic. This is my first post, “A search is not just a search,” longtime readers will find it familiar, if updated. From it: In the past few years, a significant [...]
May 9th, 2008 at 11:43 am
For your dream example to work, every bottle of wine (brand, type, vintage) needs to be the heart of an information network. Network participation (from Robert Parker, Wine Spectator, local wine shops, individual consumers, etc.) needs to revolve around a specific wine - in this case the 2001 vintage Stag’s Leap Cabernet. They need to contribute data (reviews, ratings, price, local availability, etc.) using a common dictionary of contextual attributes - not just tags.
In the existing search metaphor, the unit of information is the Web page. When you search, you get 1,000 or 10,000 or 100,000 pages of results. Each page includes things that roughly match your keywords. There are plenty of choices and plenty of voices. But there is little coherence.
Each page is like a free agent, but without the structure needed to get multiple players and teams and leagues working together successfully.
In your example, you want to see all the local retailers for a specific bottle of wine. Instead, using Google you get an avalanche of Web pages. Search makes your decision more difficult, not easier and more convenient.
To get what you want, you need a new kind of data architecture and closer collaboration between market participants. This requires a language of brands, products, and attributes organized as what we call a collaborative search network. In contrast to the avalanche, you need a search process that’s targeted and efficient. A search process that gets you all the way down to a universe of choices (e.g. local availability, specials, wine tastings) around a single, specific product .
A common language of data is the key. Indeed, as Tim O’Reilly wrote years ago, “data is the new Intel inside.”
May 9th, 2008 at 7:30 pm
Finding something on the net should be a matter of:
1) Providing minimal typing by the user (preferably NONE)
2) No more than one page of results (preferably ONE RESULT…the correct one)
It seems to me that someday the above will be reality.
I wrote about this topic a bit more in a recent post here: Search Sucks.
May 25th, 2008 at 10:08 am
…….
The problem with Universal search is - how popular are these various inclusions compared to the traditional organic SERPs?
Do users really click on them or are they basically ignored?
Yahoo is incorporating social sites and bookmarking into the organics, whiles Google is incorporating their other verticals - but are they really that useful?
This is where personalization becomes helpful.
The future of search may become more personalized than universal with pushed intelligent suggestions gaining momentum
As far as the example you gave about purchasing wine, this could eventually result in stores employing technology that neutralized those comparison searches from being performed on their premises if they got out of hand or negatively affected their businesses.
A more intelligent option would have been for that searcher to use the wine review sites, then do a local search that would have offered details about store policies and push relevant PPCs that would be triggered by the wine brand and location queries