Search behaviour analysis

Before a company can optimise its website, it has to understand how its audiences search. That means knowing which words they use, which questions they ask and where they are in the decision-making process.

Search behaviour analysis, traditionally called keyword research, is the foundation of SEO work. Without that knowledge, you are guessing. With that knowledge, you can build a data-driven strategy that brings in more relevant traffic.

At Carnaby, we call this work building a phrase cloud. The phrase cloud is a list of keywords that also shows how your audiences search and make decisions.

The search term still matters

Search intent and search term go together

Modern SEO revolves largely around search intent (“user intent”), and rightly so: it is crucial to understand why someone is searching. But the search term itself has not become irrelevant.

Someone searching for “accountant Stockholm” wants an accountant in Stockholm, not in Gothenburg, and not general information about accounting. The phrase “accountant Stockholm”, or a natural variant of it, should appear on the page.

Search intent explains why someone searches. The search term affects whether the searcher finds the website.

The three searches have different search intents. Each search term needs its own optimisation, both for the intent and for the actual words in the phrase.

In the AI era, this becomes even clearer

Google AI Overviews mainly appear for question-based searches with words such as “how”, “what”, “why” and “which”. But they are becoming more common for commercial and transactional searches too.

AI Overviews (AIO) can give you brand exposure. But rankings in the organic results drive clicks. And to rank, you have to match both the search term and the intent behind it.

How we build a phrase cloud

We move from competitor analysis to data-driven content

Building a phrase cloud is one of the cornerstones of our SEO work. It gives us both a list of search terms and an understanding of the whole search landscape your site will compete in.

Among other things, we start with your current traffic (it is an important data source). But we also look at what works elsewhere in the industry by analysing what competitors rank for.

Step 1: Identify the “Hollywood phrases”

During the competitor analysis, we identify the big, broad search terms. We call them “Hollywood phrases”. They are often searched thousands of times a month and are often single-word searches or brand names.

The name alludes to how hard it is to make it in Hollywood: the phrases are sought after and fiercely contested, and they are often dominated by big brands with years of authority behind them.

Step 2: Branch out into long-tail variants

From the Hollywood phrases, we work our way out towards variants, synonyms and niches.

Each of these long-tail phrases has a lower search volume than “web agency”. But together, long-tail phrases often make up the majority of the search terms that drive traffic to a site, and also the majority of the search terms identified in a nearly complete phrase cloud.

Step 3: Map the search intent for each phrase

For each search term, we analyse the intent:

Each search intent needs its own content. People who want information need guidance. People who are ready to buy need a clear call to action and contact details.

How customers search along the customer journey

Searches differ between the stages

Audiences search differently depending on where they are in their decision-making process.

Each stage has its own search terms that you need to be visible for. If you only optimise for the late stage (“hire a web agency”), you miss everyone who still wants to learn more. If you only optimise for the early stage (“why do companies need SEO”), you get visitors who are not ready to buy.

A complete phrase cloud covers the entire customer journey.

How AI Overviews affect different types of searches

Google AI Overviews have changed what the results page looks like for some searches.

AIO are becoming more common for other types of searches. Three types of searches made up a growing share of the keywords that triggered an AI Overview between October 2024 and October 2025:

• Commercial searches: from 8.15 to 18.57 per cent
• Transactional searches: from 1.98 to 13.94 per cent
• Navigational searches, for example for brands: from 0.84 to 10.33 per cent

AIO can give you brand exposure when your expertise is cited and your brand is mentioned, which builds trust. But for visits to your website, rankings in the organic results still matter a great deal.

How the process works

The analysis results in an action plan

When we carry out a search behaviour analysis, we follow a systematic process:

1. Competitor analysis: we identify the main competitors in search engines (not necessarily the same as your business competitors) and analyse which search terms they rank for.
2. Identify Hollywood phrases: we take the big, broad, high-volume searches as the starting point for the rest of the analysis.
3. Expand into the long tail: using tools and manual analysis, we build out the phrase cloud with variants, synonyms and niche phrases.
4. Map search volume and competition: for each phrase, we analyse:

• Monthly search volume
• Level of competition
• Business value (does the phrase lead to conversions?)

1. Classify by search intent: each phrase is categorised by search intent: informational, investigative, transactional or local.
2. Map against the customer journey: we place the phrases along the customer’s decision-making process to ensure coverage at every stage.
3. Create a content strategy: the phrase cloud becomes the basis of a content strategy in which every important page on the site has:

• Its own primary search term
• Its own role in strengthening the site’s main pages
• A clear link to search intent

The result is a data-driven plan for which search terms you should optimise for, in what order and with what content.

How we maintain the phrase cloud over time

Phrase clouds are living documents

Regular updates keep your optimisation aimed at the search terms people use now, not at yesterday’s searches.

Do you want to understand how your audiences search?

We help you map the search landscape your site will compete in and build a data-driven content strategy that drives traffic.

Contact us for a search behaviour analysis →