3 Reasons Why Your Site Search Isn't Working

Brought to you by Dr. Scott Brave

Coming from consumer search experiences on the web with the likes of Google and Yahoo, it's no wonder online marketers question why they can't provide better search results to people browsing on their own websites. It's a recipe for a common complaint: "The search on our site sucks!"

Fair enough. In order to fix search, you first have to understand what is going wrong.

Site search isn't working for many companies, and there are three key principles to really understand what is going wrong. Once we understand what's wrong, we can arrive at some fundamental conclusions on how to make site search better and, therefore, how you can make your site search a cut above the rest.

1. The critical information is not in the document
All full-text search technologies basically work the same way: They look for a match between the words in a user's query and the words in the text of the documents searched. Whether those documents are web pages, PDFs, or Word docs, the fundamental assumption is that the engine can figure out what content best meets a user's needs.

Unfortunately, this is where the very concepts of full-text searching miss the mark, and here's why:

While processing documents is a good start, the words within a document do not necessarily match the way a user understands the topic and phrases the question.

For instance, a user might come to an online appliance retailer looking for a "stove," but the search yields only "stove-top safe" kettles and pots, not stoves. Why? Because instead of referring to a generic term like "stove" on the website, the retailer instead uses a manufacturer's terminology of "cooktops" and "ranges."

It sounds like a simple fix, where all you need to do is set up a synonym so that "stove" is the same thing as a "range," right? But what about all the long-tail terms and content associated with niche products that could number in the thousands?

Let's go out on a limb and say the words do exist in the document. There may be thousands of documents that contain the search terms, but which documents are the best? A traditional search engine will assume that the one with the most occurrences of the keywords is the most valuable, but this is very often not the case.

For example, if we look at a work by Shakespeare -- or any great work of literature -- the meaning cannot be identified simply by looking at the words within it. It's synthesized in the reader's mind, and different readers may derive different meanings based on their own unique makeup and experiences.

The common work-around for this principle is to have experts manually tune and tweak search, but that can lead to major time delays in getting problems fixed. Not to mention this approach lacks scalability and is incredibly labor-intensive. Companies focus on the most popular content, and miss out on highly profitable long-tail content and products.

If the information to improve search isn't in the document and it's simply not practical to manually tune search, what's a marketer to do? We've hinted at the solution. Your visitors know the answers.

2. Actions speak louder than words
If visitors are the answer, it would seem logical that the answer to fixing search -- in our world of ratings, stars, and thumbs ups and downs -- is just to ask them. Turns out there are a few key challenges with explicit means of collecting information stemming from who participates, when and why:

Low coverage. A small subset of the population rates content, and when they do, the ratings only tend to cover the most popular content. Where does that leave the majority of our content? In the long tail -- unranked and, therefore, undiscovered.

Lacking context. Assuming that there is enough coverage, is it meaningful to your visitor's context? Let's say we have a bunch of ratings for a particular camera. Someone looking for a "lightweight camera" might think it stinks, while someone looking for a "cheap camera" might love it. You can't ignore the context of what a person is looking for.

Biased responses. In general, the people who do participate in explicitly rating something online represent a very small subset of the population. These opinions almost always represent fringe opinions that are either extremely positive or negative because those are the people motivated to be heard.

Inaccurate/incomplete feedback. Even when an individual decides to provide some form of explicit feedback, it often is not fully representative of even his or her own experience. Take Yelp reviews, for instance. A person might have 50 great dining experiences at a restaurant, but may base the review on the last visit -- when something went horribly awry. Whether people are rating a restaurant or ranking what's useful to them on your site, that feedback cannot always be trusted.

So, if asking people to tell us which documents are valuable and why they're of value doesn't work, what is the right way? The answer is in observing what people do, not what they say.

3. Observing search behavior alone is not enough
In order to improve search, we need to observe more than just search behavior. Search and navigation have traditionally been seen as two separate paradigms, with separate interfaces and separate systems driving them. But in reality what's happening? Visitors come to your site and express interest or intent through their actions.

They might have first expressed that intent through a Yahoo or Google search that brought them to your site. They might then express it in the pages they visit and engage with, the links they click, and maybe the site searches they perform. How long does their mouse linger over a piece of content? Do they spend time comparison shopping, hopping back and forth from one document to the other? This expression of interest may span multiple searches and actions and finding content that holds true value for that interest and intent may also take multiple steps.

If we only looked at search, we'd never see any of this. We have to look at the entire experience.

Learning from your site search mistakes
What's really remarkable is that once we take a step back and think of the entire online experience as a single unified expression of intent and value, we can do a lot more than fix search. We can start to make recommendations and optimize the visitor experience with every interaction they make within a site -- from the moment visitors arrive, to every step they take through the site, as well as with every search they perform. The true goal is to understand the user's intent and then automatically surface the exact products and content they are truly looking for.

Your visitors hold the answer to improving search on your site, but the question is: Are you listening?

Dr. Scott Brave is co-founder and CTO of Baynote.