How AI Overviews Are Changing Keyword Research and Search Intent Analysis for Modern SEO

Google’s AI Overviews are changing what it means to rank for a keyword. A page can now appear inside an AI-generated answer, beneath it as a traditional organic result, alongside a featured snippet, or nowhere visible at all if the search is resolved directly on the results page.

That shift has serious implications for keyword research, search intent analysis, organic traffic forecasting and keyword cannibalisation. Ranking position alone is no longer enough. You need to understand which parts of a query can be answered by an AI Overview, which sources are likely to be cited, and whether your own pages are competing for the same underlying intent.

This is where a structured publishing system becomes valuable. SEO Letters helps you move from keyword discovery and topical clustering to fully structured articles, internal links, schema, images and publishing workflows, so your SEO operation is not relying on isolated content briefs and manual copy-paste work.

What Are Google AI Overviews?

AI Overviews are generated summaries that appear at the top of some Google search results. They use information from multiple sources to provide a direct response, with links to pages that Google considers relevant or supportive.

The format can vary according to:

• Query type
• User location
• Device
• Search history
• Topic sensitivity
• Availability of reliable source material
• The current state of Google’s search systems

An AI Overview may summarise a definition, compare products, explain a process or combine several related questions into one answer. It can also encourage follow-up searches, which means the initial search is becoming more of a starting point than a final destination.

This whole thing creates a visibility problem. Your content might be cited without receiving the same level of traffic it would have gained from a conventional top-three ranking. In another case, a page may attract fewer clicks but more qualified visitors because the AI Overview has already filtered the search towards a specific need.

The practical conclusion is simple: SEO teams need to optimise for visibility, citation potential and post-click value at the same time.

Why AI Overviews Matter for Organic Traffic

Traditional SEO reporting usually focuses on metrics such as:

• Average ranking position
• Impressions
• Click-through rate
• Organic sessions
• Conversions
• Backlinks
• Share of voice

These remain useful. They no longer tell the full story.

An AI Overview can reduce clicks for straightforward informational queries while increasing the importance of being included as a cited source. It may also change the layout of the search page, pushing conventional results lower and making visual elements, follow-up prompts and product modules more prominent.

In practical terms, you may see four different outcomes.

AI Overview outcome	Likely effect on traffic	SEO response
Your page is cited and appears in organic results	Visibility may increase	Strengthen authority, clarity and conversion paths
Your page is cited but not ranked prominently	Brand exposure may rise without equivalent clicks	Improve the page’s depth and reasons to visit
A competitor is cited instead of your page	Your organic opportunity may shrink	Analyse source quality, coverage and intent alignment
No page is cited from your market	The query may be answered from broad sources	Build differentiated, expert-led content

The important point is that AI Overview visibility is not a replacement for organic rankings. It is another search surface that needs its own measurement framework.

A sensible reporting model could track:

• AI Overview presence by target query
• Citation frequency
• Citation position or source prominence where observable
• Traditional ranking position
• Organic click-through rate
• Branded search growth
• Assisted conversions
• Engagement from users arriving through informational pages
• Changes in impressions and clicks after AI Overview deployment

Google does not provide a universal, stable report for every AI Overview appearance. Third-party platforms may estimate visibility, but their methodologies differ. Treat the data as directional rather than absolute.

The Connection Between AI Overviews and Search Intent

Search intent analysis has always been central to SEO. The problem is that many keyword strategies still classify intent too broadly.

A keyword may be labelled as informational, commercial or transactional, but that simple label does not explain what the searcher expects to do next. AI Overviews are exposing the limits of broad intent categories because they often answer several sub-intents within one result.

For example, consider the query:

best project management software for remote teams

This appears to be commercial investigation. Yet the searcher may also want:

• A shortlist of suitable tools
• Pricing information
• Collaboration features
• Integrations
• Security details
• A comparison with general project management software
• Recommendations for a specific team size
• Evidence from reviews or practical use cases

An AI Overview can bring those sub-intents together. A generic article that repeats a list of software products may be less useful than a page that explains the selection criteria, compares products carefully and provides evidence for its recommendations.

Search Intent Is Becoming More Granular

Modern intent analysis should consider at least five layers:

1. Primary intent: The main task the user wants to complete.
2. Supporting intent: Questions that help the user make a decision.
3. Contextual intent: Conditions such as industry, budget, location or experience level.
4. Sequential intent: The likely next search after the initial question.
5. Conversion intent: The commercial action the user may take after finding the answer.

This framework helps you build content that does not merely match a keyword. It matches the journey around the keyword.

Example: One Keyword, Several Intent Layers

Take the keyword how to reduce SaaS costs.

Intent layer	What the searcher may need	Suitable content element
Primary	A practical cost reduction process	Step-by-step framework
Supporting	Which costs are often overlooked	Audit checklist
Contextual	Guidance for a growing business	Scenario-based examples
Sequential	Software spend benchmarks	Original data or comparison table
Conversion	Help implementing the process	Service, tool or downloadable template

If you publish only a short definition, you may match the surface query while missing the commercial and practical needs underneath it. AI systems are increasingly able to assemble a response from pages that cover those layers clearly.

AI Overviews and Keyword Cannibalisation

Keyword cannibalisation occurs when multiple pages on the same website compete for substantially similar search intent. It is often described as two URLs ranking for the same keyword, but that definition is too narrow.

The deeper issue is intent overlap.

Two pages can target different keywords and still cannibalise each other if they answer the same user need. Equally, two pages can mention the same keyword without causing a genuine problem if each page serves a distinct purpose.

AI Overviews make this more complicated because Google may assess your entire site as a collection of related answers. If several pages make similar claims, repeat the same explanations or target the same entity without clear differentiation, Google may select only one as a source.

Common Forms of AI-Era Cannibalisation

1. Exact-keyword cannibalisation

Two pages target the same primary phrase, such as:

• Best accounting software for freelancers
• Best accounting tools for freelancers

The wording changes slightly, but the intent is virtually identical.

2. Subtopic cannibalisation

A broad guide and a more focused article cover the same subtopic with no clear hierarchy.

For example:

• How to conduct a technical SEO audit
• Technical SEO audit checklist
• Technical SEO audit process

These could be separate assets, but only if each has a different job. If all three repeat the same checklist, they may dilute relevance.

3. Funnel-stage cannibalisation

A product page, comparison page and blog article all attempt to rank for a commercial query. Each page might contain similar benefit statements and calls to action, so Google has no obvious reason to prioritise one.

4. Entity cannibalisation

Multiple pages discuss the same company, product, service or topic but present no distinct angle. This can create ambiguity around which page should be treated as the authoritative source.

5. AI answer cannibalisation

Several pages provide fragments of the same answer, but none is sufficiently complete or clearly structured to become the preferred citation source.

This last category is increasingly relevant. You may have strong coverage across ten articles, yet Google may cite a competitor with one well-organised page because it can extract a clearer answer from that source.

How AI Overviews Change Keyword Research

Keyword research can no longer stop at search volume, difficulty and cost per click. Those metrics help estimate demand, but they do not tell you whether a query is likely to generate clicks or whether your page can become a useful source within an AI-generated answer.

A stronger keyword research process considers the following dimensions:

Research dimension	Core question
Demand	How often is this topic searched?
Competition	Which sites currently dominate it?
Intent	What does the searcher actually want to accomplish?
AI susceptibility	Can the query be answered directly in a summary?
Source potential	Would Google need external pages to support the answer?
Differentiation	What can your site explain better or prove?
Business value	Does the topic support a meaningful commercial outcome?
Cannibalisation risk	Do you already have pages serving this need?

AI Susceptibility Scoring

Some queries are more vulnerable to click reduction because they can be answered quickly.

Examples include:

• What is domain authority?
• How many grams are in a kilogram?
• What is a canonical tag?
• When is the next bank holiday?
• How long should a meta description be?

This does not mean these topics are worthless. They may support topical authority, brand recognition and internal links. It means the content strategy should be more selective, especially if production resources are limited.

A simple scoring model might look like this:

Score	AI susceptibility	Typical query type	Content approach
1	Low	Complex, personal or high-stakes decision	Invest in detailed expert content
2	Moderate-low	Multi-step commercial research	Build comparisons and evidence
3	Moderate	Broad explanatory query	Add original examples and practical depth
4	High	Simple process or definition	Keep concise, support with internal links
5	Very high	Factual answer with little nuance	Publish only if strategically useful

The scoring is not a prediction of exact traffic. It is a prioritisation tool.

Search Volume Needs Context

A high-volume keyword may generate less traffic than expected if Google answers it directly. A lower-volume query may produce more valuable visits if it signals a specific problem and leads to a commercial action.

For example, compare:

• SEO audit
• SEO audit for Shopify stores with international subfolders

The first phrase has broader demand. The second may attract fewer searches but reveal stronger relevance, clearer intent and higher conversion potential.

When AI Overviews compress generic answers, specificity becomes a defensive advantage. Pages that address an identifiable audience, platform, constraint or decision often have more reasons to be visited.

A Modern Keyword Research Framework

If you are updating your SEO process, use this repeatable framework.

Step 1: Collect keywords from multiple sources

Do not rely on one keyword tool. Combine:

• Search Console queries
• Competitor page titles
• Customer support conversations
• Sales call notes
• Reddit and specialist forums
• People Also Ask questions
• Related searches
• Product reviews
• Internal site search
• Keyword research platforms
• Existing content performance

This gives you language from real users rather than a list based purely on estimated volume.

SEO Letters can support this wider planning process by combining keyword research, difficulty ratings, topical authority clusters and content-gap analysis into a publishing workflow. That matters when you are trying to build a complete topic map rather than produce one article at a time.

Step 2: Group keywords by problem, not just wording

Create clusters around the problem being solved.

A possible cluster for content refresh could include:

• How to update old blog posts
• Content refresh checklist
• How often should you update website content
• Signs a blog post needs updating
• Content decay in SEO
• How to improve declining organic traffic

These phrases are related, but they may not all deserve separate pages. Grouping allows you to identify the central topic and decide which questions belong in supporting sections.

Step 3: Analyse the current search results

Review the first page and record:

• Content type
• Search intent
• Page depth
• Heading structure
• Publication and update dates
• Use of original research
• Brand authority
• Product or service positioning
• Internal linking patterns
• Whether the result appears suitable for AI citation

Do not copy competitor structures mechanically. Look for weaknesses, gaps and assumptions.

Step 4: Identify the dominant and secondary intents

A page should normally have one dominant intent. It can address secondary questions, but those supporting sections should reinforce the main purpose.

For instance:

• Primary intent: compare SEO writing tools
• Secondary intent: understand pricing, workflows and integrations
• Conversion intent: start using a writing platform

If the article also tries to be a complete guide to keyword research, it may become unfocused. That is where cannibalisation often starts.

Step 5: Assign one primary URL to each intent

Create an intent map that gives every important topic a home.

Topic	Primary URL	Page type	Supporting URLs
AI Overview keyword research	/ai-overviews-keyword-research/	Pillar guide	/search-intent-analysis/, /keyword-cannibalisation/
Search intent analysis	/search-intent-analysis/	Supporting guide	/keyword-research/, /content-brief-template/
Keyword cannibalisation	/keyword-cannibalisation/	Diagnostic guide	/content-audit/, /internal-linking/
AI writing workflow	/ai-seo-writing-tool/	Product page	Related guides and use cases

This is not just an organisational exercise. It sends clearer relevance signals to search engines and gives your internal links a logical purpose.

Step 6: Score business value and AI risk

A practical prioritisation score can include:

• Search demand: 1 to 5
• Commercial relevance: 1 to 5
• Ranking difficulty: 1 to 5, reversed if necessary
• AI susceptibility: 1 to 5, reversed if click protection is important
• Content differentiation potential: 1 to 5
• Cannibalisation risk: 1 to 5, reversed

You can weight commercial relevance and differentiation more heavily than volume. This prevents your team from building a large library of generic pages that attract impressions but little revenue.

How to Optimise Content for AI Overview Citations

There is no guaranteed formula for appearing in an AI Overview. Google’s systems are dynamic, and citations can change by query, market and time.

You can still improve the usefulness and extractability of your content.

Answer the central question early

Do not bury the main answer beneath a long introduction. Start with a concise explanation, then add the reasoning, examples and qualifications.

A strong structure often looks like this:

1. Direct answer
2. Definition or context
3. Detailed explanation
4. Practical process
5. Examples
6. Risks and limitations
7. Related questions
8. Next action

This works for human readers as well. People usually want orientation before detail.

Use explicit headings

AI systems need to interpret relationships between sections. Descriptive headings are more helpful than vague ones such as “What You Need to Know”.

Use headings like:

• How AI Overviews affect keyword research
• What keyword cannibalisation looks like in an AI search result
• How to map one primary URL to each search intent
• Metrics for measuring citation and organic visibility

Clear headings also improve accessibility and scanning.

Include evidence and first-hand insight

Content that makes unsupported claims is easier to overlook. Strengthen important points with:

• Original data
• Screenshots where appropriate
• Methodology notes
• Expert commentary
• Tested workflows
• Customer examples
• Clear dates
• Links to authoritative sources
• Limitations and caveats

E-E-A-T is not a box-ticking exercise. It is a way of demonstrating why your page deserves attention.

Make claims specific

Compare these two statements:

AI Overviews are changing SEO.

For informational queries with a simple factual answer, AI Overviews may reduce the need for a click, so content teams should track query-level changes in impressions, click-through rate and citation visibility rather than relying on average rankings alone.

The second statement provides a condition, an implication and a measurement framework. That is more useful.

Add original value beyond the summary

If the AI Overview can provide the same information from your first two paragraphs, the page needs a stronger reason to be visited.

Useful differentiators include:

• A detailed decision framework
• A proprietary scoring model
• Real implementation examples
• Industry-specific guidance
• Interactive templates
• Product comparisons
• Original research
• Expert commentary
• Data that changes over time
• A tool or workflow the reader can use

This is particularly important for commercial and high-consideration topics.

Preventing Keyword Cannibalisation in AI Search

Keyword cannibalisation should be managed during planning, not only after traffic declines.

Build an intent ownership model

For each cluster, define:

• The main user problem
• The primary URL
• The target audience
• The conversion objective
• The supporting questions
• The pages that should link to it
• The pages that should not compete with it

A useful rule is to ask:

If a search engine had to select one page from this cluster as the best answer, which URL should it choose?

If your team cannot answer that, the content architecture probably needs work.

Consolidate overlapping pages

Consolidation may involve:

• Merging two weak articles
• Redirecting an outdated page
• Rewriting one page for a narrower intent
• Changing a blog article into a product-led guide
• Removing duplicated sections
• Strengthening the preferred URL
• Updating internal links
• Reviewing canonical tags

Do not merge pages solely because they share a keyword. First compare their impressions, links, conversions, ranking history and actual intent.

Use content differentiation deliberately

Two related pages can coexist when they have different jobs.

Page	Primary purpose	Differentiation
What is technical SEO?	Explain the concept	Beginner definition and terminology
Technical SEO audit checklist	Help users perform an audit	Practical checklist and workflow
Technical SEO services	Convert commercial visitors	Service scope, process and proof
Technical SEO for ecommerce	Address a specialist use case	Facets, templates and platform issues

The headings, examples, internal links and calls to action should support those differences. Merely changing the title is not enough.

Strengthen internal linking

Internal links can clarify your site’s topic hierarchy. Link from supporting pages to the primary resource using natural, descriptive anchors.

Good examples might include:

• AI Overview search strategy
• keyword cannibalisation audit
• technical SEO content workflow
• topical authority planning

Avoid linking every page to every other page. That creates noise and makes the hierarchy less obvious.

How SEO Letters Supports AI-Era Content Operations

A modern SEO team has to research topics, assess competition, plan clusters, produce content, add internal links, publish across platforms and review performance. Doing each task manually creates delays, inconsistency and a growing backlog.

SEO Letters is designed as an AI writing and publishing engine for that complete workflow. You can research keywords, identify topical gaps, create structured articles and publish to WordPress, Shopify or webhooks without moving content through a chain of disconnected tools.

Its workflow can support:

• Keyword research with difficulty ratings
• Topical authority clusters
• Competitor site-gap analysis
• Structured long-form article generation
• Internal linking recommendations
• Schema and image support
• Brand voice configuration
• Product-aware affiliate and ecommerce content
• Multi-language generation across 21 languages
• Direct publishing destinations
• Performance monitoring
• Scheduled autonomous campaigns
• Content refresh campaigns

The autonomous campaign scheduler is especially relevant when AI Overviews are causing search visibility to shift. You can define a topic, cadence and publishing destination, then allow the system to research and prepare content repeatedly while your team focuses on strategy, review and commercial priorities.

Automation still needs governance. A published page should have a human review process for facts, claims, brand suitability, legal issues and subject-matter accuracy.

A Practical Content Workflow for AI Overview Visibility

Use this process when creating a new article.

1. Define the query and user problem

Write the target query at the top of the brief. Under it, describe what the reader is trying to decide, understand or do.

Do not accept a vague description such as “learn about SEO”. Specify the task.

2. Review the SERP and AI Overview

Record:

• The presence of an AI Overview
• The questions it addresses
• Cited sources
• Search features
• Dominant page formats
• Content gaps
• Commercial pages appearing in the results

Repeat the search at different times where the query is important. Results can fluctuate.

3. Create a source-worthy outline

Your outline should include:

• A direct answer near the start
• Clear definitions
• The main process
• Exceptions and risks
• Examples
• Supporting data
• A measurement section
• Relevant next actions

Avoid adding sections solely to increase word count. Long content with weak information architecture is not a serious competitive advantage.

4. Assign one intent to the page

Write a one-sentence intent statement, such as:

This page helps SEO managers identify and resolve keyword cannibalisation caused by overlapping informational content.

If a section does not support that sentence, move it to another page or remove it.

5. Add proof and practical detail

Include:

• Worked examples
• Decision criteria
• Tables
• Templates
• Technical instructions
• Relevant limitations
• References to reliable sources
• First-hand observations from testing or implementation

This gives the content a reason to exist beyond a summarised definition.

6. Connect the page to the site architecture

Add links to:

• The relevant pillar page
• Supporting topic pages
• Product or service pages
• Related tools
• Case studies
• Contact or rightbar pathways where appropriate

The rightbar can act as the contact path for teams that need help building content campaigns, reviewing their SEO architecture or setting up a publishing workflow.

7. Publish and monitor query-level results

Track performance before and after publication. Allow enough time for indexing and ranking changes, but do not wait indefinitely if the page shows clear signs of overlap or weak intent alignment.

Hypothetical Example: A SaaS Website Losing Organic Clicks

Imagine a SaaS company that publishes three pages:

1. How to reduce software costs
2. SaaS cost reduction strategies
3. SaaS spend management guide

All three pages target similar informational queries. They contain overlapping advice about removing unused licences, reviewing renewals and negotiating contracts.

The site ranks between positions six and eighteen across the cluster. Its impressions are stable, but clicks decline after AI Overviews begin appearing for several queries.

Diagnosis

The problem is not only the AI Overview. The site has not clearly assigned ownership of the topic. Google may understand that the pages are related, but it has limited reason to select any one of them as the definitive source.

Recommended action

• Keep SaaS spend management guide as the broad pillar page.
• Redirect or consolidate the two overlapping articles.
• Create a separate page specifically for SaaS renewal negotiation, if that query has distinct commercial intent.
• Add original cost benchmarks and a downloadable audit framework.
• Link supporting pages to the pillar using descriptive anchors.
• Add a product-led section only where the company’s software genuinely solves the problem.

The objective is not to make every page rank. It is to make the site’s topical structure easier to understand and more useful to the reader.

Measuring Success When Clicks Are Not the Whole Story

AI Overviews require a broader measurement model. Traffic remains important, but it should sit within a visibility and business framework.

Recommended KPI categories

Visibility KPIs

• Traditional ranking position
• AI Overview inclusion
• Citation frequency
• Branded search impressions
• Share of voice
• Featured snippet ownership
• Search feature presence

Engagement KPIs

• Organic click-through rate
• Engaged sessions
• Scroll depth
• Return visits
• Time on key sections
• Internal link clicks
• Downloads or tool interactions

Business KPIs

• Leads
• Assisted conversions
• Product trials
• Revenue from organic sessions
• Conversion rate by landing page
• Pipeline influenced by organic content
• Cost per qualified organic lead

Content quality KPIs

• Pages updated on schedule
• Cannibalisation incidents
• Content decay
• Average time to publish
• Fact-checking completion
• Internal link coverage
• Percentage of pages with a defined owner

A page that receives fewer clicks but generates more qualified leads may be performing better. A cited page that increases branded searches may also be creating value that standard last-click attribution misses.

Use a consistent reporting window and annotate major Google changes, page consolidations and content refreshes. Otherwise, your conclusions may be based on timing rather than causation.

Common Mistakes SEO Teams Make

Treating AI Overviews as a separate content type

AI Overviews are a search presentation layer. They do not remove the need for strong pages, clear site architecture, credible information and useful experiences.

Chasing every question with a separate article

This creates thin topical coverage and increases cannibalisation risk. Some questions belong in one comprehensive guide.

Assuming a citation guarantees traffic

A citation can improve visibility, but it may not produce a click if the answer is complete. Give users a strong reason to continue to your page.

Measuring only average position

Average position can hide query-level changes, SERP layout shifts and brand exposure. Segment your analysis by intent and search feature.

Using AI-generated content without editorial control

Automation can increase production capacity, but it does not replace subject expertise. Review claims, statistics, product statements and regulated topics before publication.

Ignoring existing pages

A new article may look strategically sound until it competes with a page that already has backlinks, impressions or conversions. Always audit the site before creating another URL.

A Simple Cannibalisation Audit Template

Use this template for each topic cluster:

Audit question	Finding	Action
Which URL currently receives the most relevant impressions?
Which URL has the strongest backlinks?
Do multiple pages serve the same primary intent?
Are titles and headings too similar?
Is one page clearly the preferred authority?
Do internal links point consistently to that page?
Should pages be merged, redirected or differentiated?
Does the cluster have a commercial next step?

Complete the audit using Search Console data, analytics, crawling software and manual content review. A spreadsheet is sufficient for a small site. Larger websites may need automated URL clustering and regular content inventories.

The Strategic Shift: From Ranking for Keywords to Owning Answers

AI Overviews are encouraging a broader change in SEO strategy. The strongest sites will not simply publish pages for a list of phrases. They will build authoritative answer systems around important topics.

That means:

• One clear owner for each major intent
• Supporting pages for distinct subtopics
• Internal links that reflect the hierarchy
• Original evidence and expert interpretation
• Content designed for humans first
• Structured information that search systems can interpret
• Regular updates as search behaviour changes
• Commercial pathways that feel relevant rather than forced

Keyword research remains essential. Its role is expanding.

You are no longer asking only, “How many people search this phrase?” You are also asking:

• What decision sits behind the query?
• Can an AI Overview answer it without a click?
• What evidence would make our page worth citing?
• Which URL should own this intent?
• Does our existing content already answer it?
• What will the reader need next?
• How can this topic contribute to measurable business growth?

Key Takeaways

AI Overviews are changing the relationship between rankings and traffic. They may reduce clicks for simple informational searches, while increasing the value of credible sources, differentiated analysis and strong brand visibility.

Keyword research now needs to include:

• Search intent depth
• AI susceptibility
• Citation potential
• Business value
• Content differentiation
• Cannibalisation risk
• Topic and URL ownership

Keyword cannibalisation is no longer just a problem of two pages targeting the same phrase. It is a problem of unclear intent, duplicated answers and weak information architecture.

If you are building content at scale, SEO Letters is the blog writing and publishing tool that can help manage the work between strategy and the live page. Use it to research topics, map clusters, generate structured articles, add internal links, support brand-aware writing and run scheduled content or refresh campaigns.

The winning approach is not to publish more pages blindly. It is to create a disciplined publishing operation where every page has a defined purpose, a clear audience, a measurable business role and a place within the wider topical authority model. That is the standard modern SEO teams should be working towards.