This is the second blog post about intent-based search. Before reading this post, we suggest you read the first post here. 

Introduction

In the previous post, we discussed how we went from traditional SEO work such as keyword research to Google updating its algorithms and now making intent-based search an important aspect of the SEO strategy. By combining the traditional SEO work of identifying keywords and clusters of keywords we are able to understand search behaviour within an industry. However, there are many cases where that research won’t provide you with enough information in terms of intent and context.

In this part, we will discuss how BERT (Bidirectional Encoder Representations from Transformers) and intent-based search affect the ones working with SEO and the users, and put forward a framework that creates a structure for working with this topic.

BERT for the user

As mentioned in Part 1, BERT looks at words before and after every word in a search string (bidirectional). By identifying and understanding the relationship between the words Google can determine the intent behind the search. This would, for example, mean that Google can attribute “more value” to a certain word in the sentence such as “buy” in order to determine that the word “buy” sets the intent for the query.

Basically, the idea is that users will be able to (and are encouraged to) use complete sentences and not keyword-based search queries when searching on Google as BERT will be able to understand it. This means that users can enter a full sentence into the search bar and BERT will be able to return an accurate result, most likely to a greater extent than when based on a number of stand-alone keywords.

But what about all those 15% of queries entered into Google every day that have never been seen before? Or, for stand-alone keywords that do not have a clear intent?

Understanding and identifying intent

There will undoubtedly be cases where BERT still won’t be able to understand the context or intent behind a search. Actually, Google says that BERT will improve the understanding of approximately 1 in 10 searches in English in the US (Google, 2019). The same will be the case for SEO-experts, so the question is how to tackle this issue?

In order to meet the intent behind a search that cannot be predefined, there is a need to understand the query at hand. Rather than looking at the keyword as a stand-alone word, we need to elevate and understand what content will serve the users in relation to the query.

Since this approach is fairly new there may be multiple ways of approaching the issue at hand, but here is one way to go about it. When coming across queries which are difficult to identify intent for, one approach is to collect data on all questions that relate to the query. Questions are key, as they provide more information regarding the intent than a specific keyword. If a user asks three questions in relation to a query we can with the help of data understand the funnel and/or flow that the users are looking for. Entering a page for Question A, we know that the user most likely will look for Question B and C as well – let’s provide them with content for those questions. This way we help users go through the funnel a lot faster, with highly relevant content, which in the end potentially could improve conversions.

How to work with intent-based search

As mentioned before, this area is rather new and there may be multiple approaches to this, but there are five main steps that are necessary to take in order to make sure that the intent is being taken into consideration when optimizing for organic search.

1. Keyword Research

Start out by conducting comprehensive keyword research, according to SEO-best practices. This step allows you to identify behavioural patterns within, and/or, in close proximity to the target company’s industry. Here, we are focusing on keywords and clusters of keywords in order to structure the data that we are provided with.

2. Keyword Mapping

The next step is keyword mapping, a traditional method within the field of SEO. In this step, we map a keyword (or keyword cluster) to a given landing page with the intention of that page ranking for this (or the cluster of) keyword(s). However, this is in this step the “intent” aspect comes into play. We can only map a keyword (or cluster) to a landing page if we are certain of the intent behind the query. If not, we need to move on to step 3.

3. Intent-Identification

For the queries that we are not sure of the intent of in Step 2, we can try to understand the intent based on related questions. Proper research in terms of how Google understands this topic or cluster of keywords is a quick and easy way to improve your understanding of the intent. Map out all questions for the query and try to identify patterns and behaviours in the data. Based on these insights, there will most likely be several questions and related queries to target in order to create a smooth customer journey on the website.

4. Optimization

Based on the Keyword Map we know which landing pages are targeting which keywords. Now we move on to optimize these pages based on SEO best practices. If a page does not exist, we need to create it!

5. Refinement

When the optimization is done we track the progress, evaluate the results and refine. Go back to step 1 or 2 (depending on how comprehensive your primary keyword research was) and go through the steps again to identify new keywords to target and/or re-optimize the existing content in order to be as accurate as possible in terms of how Google interprets the page.

This is a good starting point for making sure that the intent is taken into consideration when optimizing to increase organic traffic. Make sure that you don’t miss out on this, as it is evident that Google is focusing a lot on intent going forward, and we can most likely expect more algorithm improvements in this area in the near future.

Interested in more SEO? Check out this blog post about which SEO KPIs to monitor!