Conversational search and query variants

Jul 18th, 2018

We’ve said it before and we’ll say it again, voice search is growing and it is now a cornerstone of forward-thinking search marketing strategies

As brands and businesses raise their game in terms of search performance, it is vital that they understand some of the nuances that really boost this type of search query. Over the coming blog we’ll look at conversational search, compound queries and implicit queries, and see how these considerations can be the difference between high organic search performance and disappearing in the search engine results pages (SERPs).

What is conversational search?

Conversational search is the bridge between human and computer interaction. It is the way in which a user searches for the answer to a query on a search engine using phrases more adept to the spoken word. This type of search can be in a written format or delivered orally to an assistant or other platform that allows voice search.

The main principle is that a user can ask a question in sentence format, and that device can respond with a full sentence. One element of conversational search is that the technology can analyse all of the words in a conversational search query, rather than picking out specific keywords.

An example of this would be a user typing in a search similar to this: ‘Volkswagen Golf top speed’, and having the following answer returned; ‘the top speed of a Volkswagen Golf is 120mph.’

What are compound queries?

A compound query is the process of submitting a search query and receiving an answer from Google before submitting another query which will correlate with the first and allow Google to learn from it to provide a better answer.

Google will use cues and prompts from the initial search in a machine learning format to make sure that it is offering the best possible answer. The reason that this happens is that the latest generation of searchers are using both newer technologies and a longer, more conversational form of search.

Conversational search as we mentioned earlier is the ability to use natural language to find a solution. With the invention of Google assistants and those on smartphones such as Siri, this is fast becoming the search norm.

To help understand compound queries a little more and to ensure that they are being used to best effect, it is vital to break the topic into two distinct areas – intent revision queries and chained queries.

Intent revision queries – These types of query reflect everything that we have mentioned in the past about machine learning. These types of search mean that in order to get the most comprehensive answer a user does not have to start a search again but can edit or revise their initial search. An example that is widely used when it comes to this type of search is as follows: A user starts by asking ‘show me cookbooks’, then revised the response with a second query of ‘show me the vegetarian ones’ which inherently refers back to the previous query thus preventing a second, entirely separate search.

Chained queries – These searches work in a similar way, but instead of revising the initial query the user can instead run additional queries around the same topic. An example of this would be if a user first asked ‘on which continent does France sit?’, they could then follow that with a chained query, such as ‘what is the capital?’. This is not revising the query, but it is dependent upon having asked the first query.

What are implicit queries?

Initially, when a user typed a query into Google they expected that the search engine would match the words directly from one to the other. These matches represented and displayed the search results best placed to deliver an answer. This is very basic, early-format, search and as Google and others have gotten more refined implicit queries have become more prominent.

A good example of this would be if a user was to conduct a basic or ‘explicit’ search and type in something like ‘restaurants’ they would have originally been shown the webpages and documents that ranked best for that term.

Now however, it is possible to get an ‘implicit’ query that looks at other data and allows for a better result.

The implicit aspects of the ‘restaurants’ query therefore can include:

  • Which device is being used?
  • The user’s location
  • The user’s language (of search)
  • The user’s search history
  • Which browser they are using

As we move forward hyperlocal search and other tools will also play a part offering the search engine to see:

  • The users method of transport
  • Which side of the road they are on
  • How fast they are moving
  • Other appointments – syncing to calendars etc

Understanding where Google pulls the answers from and the decisions that it makes in terms of what to display based on data and its wider knowledge, is vital. Over time we can expect the number of implicit searches to become the social norm as displayed below.

Unlike text search, voice search is not just direct – but often colloquial. If businesses and brands can adapt for this and understand how people speak and search then they are without question heading in the right direction.

Don’t miss a trick – sign up to our blog to ensure you’re up to date with the latest news, views and best practice in the search marketing industry. Alternatively, you can download one of our many free resources to help take your efforts to the next level.

Facebook Twitter Instagram Linkedin Youtube