Enero 2019
 << < > >>


¿Quién está en línea?

Miembro: 0
Visitante: 1

rss Sindicación

Computational Queries: The New Vista in Search

Computational queries have emerged as a new category of search behavior. The most common type of computational query I use is how far from HERE to THERE. Sometimes I do this when driving around town. Sometimes I do this when helping someone plan a road trip. Some times I do this when I read an interesting article about astronomy.

Another type of computational query is how long does it take to bake cookies? There are differences in baking times. Unlike the distance-based computational queries baking-based computational queries require additional explicit criteria from the user (unless they are willing to settle for random information).

Both types of computational queries rely upon what I shall call Responsive Contexts: the additional information returned or required for resolving the query correctly changes based on circumstance.

In a distance-based query the additional information may include road closings and openings, traffic accidents, time of day (allowing for rush hour), and congestion due to construction projects and other factors (such as politicalmotorcades). The search engine or app may or may not have this information when you run the query.

In a baking-based query the additional information may include size of oven, type of oven, ingredients used in the recipe, and desired texture of the food. There is less information volatility in a baking query unless you have to pay attention to altitude, outside temperature, humidity, and things like that (and, yes, there are cooking articles to go into these kinds of details).

The simple query how long does it take to boil a pot of water also has Responsive Contexts. A few years ago it was deemed sufficient to give a single, simple example. Now people are modifying their queries to look for answers based on their elevations above sea-level, and maybe also based on their source of heat. General search and app search may not see enough volume in these types of queries to anticipate this kind of computation.

Youll know the search volume is sufficient when you start to see answers in the search results or apps that take your location and elevation into consideration when you type in how long to boil water. If you just searched on different types of pots, maybe a smart query resolver will take the different materials into consideration. If it knows you are in the middle of a national forest a smart app should anticipate that you are using sterno and a camping stove.

As Web marketers we dont yet have enough data about everything to speak in broad terms of Responsive Context but we are moving in that direction. I see this happening in site search more and more often. Site search drives more than 70% of all Web search (far more than Google). I suspect the percentage is even higher in app-based site search. But we dont bake Responsive Contextualization into our site search tools very often.

One way to do this is to allow the searcher to configure persistent filters rather than force them to re-enter those filters on every search. Another way to do this is to take whatever is in the users shopping cart into consideration. If I just puta camping tent in my shopping cart maybe when I search for chair, bench, table, or stool I am looking for more camping gear. Unfortunately the major retail sites dont yet give us this kind of smart search, but they will. They already suggest additional products based on what we have carted or purchased.

It happens in all kinds of ecommerce. If you register a domain name you are inundated with additional deals for HTTPS certificates, privacy services, Web hosting, etc. Merchants even offer you discounts on additional purchases. But they have not yet contextualized their site search tools to take these recent transactions into consideration.

If you are just blogging you can create Responsive Contexts by subdividing your articles to cover the different contexts. Its a bit like creating three images of different sizes to be used in a Responsive Ad unit. The search engine, seeing the very different contextual information, pulls the best matching information from your article to satisfy the query.

I have created Responsive Contexts in articles but its not easy to do. You can use tabular data layouts, structured markup, and sometimes just well-highlighted paragraphs of text (use bold and italics to call out important facts, not keywords). A good example of where youre likely to find Responsive Contexts in structured markup is customer reviews on sites like Yelp and TripAdvisor. By slightly changing the query you can get to different reviews or comments on the same business listings; ideally, we want the search engines to take notice of where we are and what we are doing to help us drill down to the correct comments or reviews, although that wont always be computationally possible.

To feed the right answers to the Computational Queries you have to define the problem in such a way that the search engine or app sees the relevance to the query. You must then provide contexts for the answers you include. And you cannot possibly anticipate all the possible contexts that people will need to qualify their answers.

But assuming we can incorporate Responsive Contextualization into our editorial policies, how should we measure placement, visibility, off-site conversion, and insight conversion? We dont have proper tools for doing this.

Well probably have to invent new types of meta markup that can be tracked. I hate the idea of pumping more analytics code into Web pages but marketers want to know what the ROI is. If you are contemplating developing Responsive Contexts for 100,000 items in your query base you had better know what that will cost and have a projection for what kind of revenue will come back from the effort. Hence, you need to run experiments and without good analytics markup you wont have good experimental data.

Click-tracking is our best opportunity for measuring Responsive Contextualization. This may lead to problems down the road. Well be designing things like intrapage doorway widgets that are used to measure performance based on context.

But standards for measuring Ad Viewability may help us solve that sticky problem. If the context is viewed and an encoded action is performed then maybe we wont have to resort to in-page illusions (that search engines will almost certainly object to) in order to capture user activity data. The data will become vital to next generation on-page optimization because today we dont measure intrapage performance even though the search engines have been displaying random sections of pages in the SERPs for years.

Marketers have yet to let go of their outdated ideas about pages and (queries, keywords, or topics). It is not that the page is relevant to some query, it is that only some portionof the page is relevant to the query. This is an important concept that people need to understand. A few years ago in an article titled These Arent the Meta Descriptions Youre OptimizingFor I wrote that the whole page is the meta description. I warned readers then that I would say very little about this for a few years. That is because by ignoring meta descriptions and focusing on putting indexable text on the page that targets many queries you gain a competitive advantage over other people who are still fussing with meta descriptions.

You should assume that the meta description tag is as dead, useless, and irrelevant to search engine optimization as the keywords meta tag. Your natural instinct will be to argue the point, which just holds you back. Stop and think about which helps you more:

A search engine displays a text snippet from a meta descriptionA search engine displays the same text snippet from on the page

The second situation gives you a relevance boost for the query. The first situation provides nothing toward computing the IR score that helps your page rank for the query in the first place.

Worse, most marketers think about meta descriptions the wrong way anyway. They choose the best possible meta description for a single query and limit the amount of text (usually to about 25 words); instead, you should be including meta text that addresses multiple queries, up to 250 words.

But you can and should be putting those 250 words on the page so they help with your optimization. And since the search engines will grab those words anyway, you dont need meta descriptions. You have not needed meta descriptions for at least five years. Worse, you have been DE-optimizing your pages by trying to craft ideal meta descriptions that favored one query over many others.

The age of Computational Queries and Responsive Contexts may finally help marketers abandon meta descriptions and other outdated practices. The idea of optimizing for topics wont fit into this model, although many people will try to say its the same thing. A topic has no context and therefore cannot be responsive to a computational query.

But not everything will become a computational query, either. Can I paint my car is a very different query from how much paint do I need for a 2001 Toyota Corolla? You may see the same results for these two queries, especially if you have been searching on topics related to 2001 Toyota Corolla. But the answers you need are very different. For example, some communities may forbid people from painting their own cars. That is a context you wont include in your question although you may search for it later (or before you get to can I paint my car?).

Related queries reveal more about the context to the search engine but are hidden from the marketer. The searcher may want to know things like:

How to paint a carCan I paint my own car in [LOCATION]Cost to paint a carHow much paint for a [CAR MODEL]Tools for painting your own car

Where in that sequence do you fit? The search engine may help you appear in queries for people who want to legally paint their own Ford Probe in their home town assuming you are relevant to that context, and you may not appear anywhere in a query related to a community across the country. You wont know the search context for the query that brought the visitor to your site. And that same visitor may run other queries after visiting your site to fill out their knowledge about how to do the project.

This is not a traditional local search query, but its a localized computational query. Apple, Bing, and Google are laying the groundwork for resolving these kinds of queries but their solutions are, I think, still years in the future.

What the search engines want to do is see that the searcher is trying to paint their car. They might develop algorithms that assemble the whole plan for the searcher. Such results would look very different from what we see in the SERPs today. For example, the searcher might see:

A local government site with regulations about painting your carA local paint store with the type of paint you needA local hardware store with the tools you needA how to guide with pictures and/or video showing you how to do itExamples of people who painted their own cars (maybe blog posts)

And then you would see normal (lacking context) search results for whatever the last query was. We could call this a Search Results Plan. Does it sound too far-fetched? We already see ads for related services based on these kinds of queries. The organic results are far less organized. When organic catches up to paid search in this area marketers will scramble to create proper contextual clues in their content to help resolve these kinds of queries.

Stop obsessing over data science and artificial intelligence. Its not all about how the search engines implement data analysis and AI decision-making algorithms. They have to have good data to work with. The marketers who get on top of the contextualization issue will be the ones who succeed in the next generation of search optimization strategies.

As search engines become better at interpreting searcher intentions and move away from their current (and very horrible) one solution fits all methods for generating search results they should create new marketing opportunities for Website owners who just want to participate in queries. Although search results exclusion will increase the diversification in searcher contexts should lead to more smaller sites earning traffic because they will better fit the needs of individual searchers than the larger, by-the-data Websites that continue to chase keywords.

Youll see yet more calls for brand-focused optimization in this new universe of search. Brand-based optimization will never go away but it is a terribly crude strategy for meeting tomorrows search needs. SEO sophistication has to look down the road far enough to have the pieces in place so that the search engines can evaluate the new data structures (not structured markup) and informational components so that they can develop the algorithms to make better use of the information they are indexing and sharing with searchers.

This is one area where publishers must be the innovators. In five years you can look back at this article and say with all the smugness you can muster, Well, he got THIS PART wrong but you probably wont even be involved in Web marketing if you dont start evolving now.

Share on LinkedInShare on reddit
06:12:34 . 21 Avr 2016
Admin · 25 vistas · Escribir un comentario

Vínculo permanente hacia el artículo completo


Este artículo no tiene Comentario por el momento .....

Escribir un comentario

Estatuto de los nuevos comentarios: Publicado

Se mostrará tu URL

Por favor introduce el código contenido en las imágenes.

Texto del comentario

   (Salvar el nombre, el mail y la URL en los cookies)