Traditional Search + AI — Practical B2B Guide

How to use Google Search Console, Query Fan-out, Answer Architecture and AI data to prioritize topics, develop products and generate B2B demand

SalesBot — practical guide | current as of August 13, 2026

This guide is based on our case study comparing two real, anonymized B2B niche websites: a new topical site and a long-established industry website. We combine standard Google Search Console data with the new Generative AI performance report. The goal is not to “hack AI.” The goal is to build one information architecture that can perform across traditional Search, AI Overviews, AI Mode, social search and the B2B buying process.

Why this guide matters now

Until recently, a marketing team could manage SEO with a relatively simple model:

keyword,
page,
ranking,
click,
conversion.

Generative search adds another layer:

user problem,
query,
query fan-out,
retrieval and grounding,
answer synthesis,
source selection,
click or further interaction.

Google officially describes AI Overviews and AI Mode as part of Google Search. They rely on core ranking systems while also using mechanisms such as RAG/grounding and query fan-out. In practice, companies should not build two disconnected strategies called “SEO” and “AI SEO.” They should build a better information system.

The key shift: we are no longer managing URL rankings alone. We are managing the company’s ability to provide the best answer to a customer problem across multiple Search surfaces.

1. Start with the correct measurement model

This is the first and most important step.

AI Overview and AI Mode data are already included in the overall Search Console Performance report. The separate Generative AI report isolates that part of visibility; it is not an additional traffic channel that should be added to standard Search impressions.

Example:

10,000 total Search impressions,
2,500 Generative AI impressions.

This does not mean 12,500 impressions. It means roughly 25% of total Search visibility came from generative Search surfaces.

The current Generative AI report primarily provides impressions and dimensions such as Pages, Countries, Devices and Dates. Do not invent metrics the report does not provide, such as a separate “AI CTR” when you do not have a separate AI click count.

2. Build two parallel data sets

For every site, export two reports.

A. Overall Search Performance

Collect:

Clicks,
Impressions,
CTR,
Average position,
Pages,
Queries,
Countries,
Devices,
Dates.

B. Generative AI Performance

Collect:

AI impressions,
Pages,
Countries,
Devices,
Dates.

Then join the data at URL level.

3. Align the analysis period

Do not compare metrics from mismatched date windows.

If standard Search covers July 14–August 10 and the AI report covers July 13–August 9, use the common period — July 14–August 9 in this example — before calculating AI share.

This small methodological rule determines whether the analysis is credible.

4. Calculate AI Share of Search Visibility

SalesBot working metric:

AI Share of Search Visibility = Generative AI impressions / Overall Search impressions × 100

In our case study, using the exact same 27-day period:

Site	Overall Search impressions	AI impressions	AI Share
New niche site	9,763	2,834	29.0%
Established niche site	24,613	1,003	4.1%

The new project therefore had roughly seven times the share of generative AI within total Search visibility.

This is not a Google ranking factor. It is a diagnostic metric.

How to interpret AI Share

Avoid rigid cross-industry benchmarks. Treat AI Share as a characteristic of the category and the content architecture.

Lower AI share may be more common for queries that are:

product-specific,
navigational,
branded,
straightforwardly transactional.

Higher AI share may occur more often for problems that are:

complex,
interpretive,
comparative,
regulatory,
dependent on synthesizing multiple sources.

These are hypotheses to test on your own data, not universal rules.

5. Join Search and AI data URL by URL

Your most important working sheet should contain one row per URL.

URL	Search impressions	Clicks	CTR	Position	AI impressions
Page A	2,500	144	5.76%	…	522
Page B	737	58	7.87%	…	296
Page C	…	…	…	…	…

Only this level of analysis shows whether Google uses the same page in traditional Search and generative Search.

6. Classify pages into four Dual Search groups

Type A — Dual Winners

Strong in both standard Search and generative AI.

These are your highest-value assets.

Action: PROTECT + EXPAND.

avoid unnecessary URL changes,
keep facts current,
add evidence,
strengthen internal links,
add video, case studies and tools,
improve the path to conversion.

Type B — Search Winner / AI Weak

The page performs in traditional Search but has little generative visibility.

Do not assume it is broken. It may be serving a simple transactional intent perfectly.

Still, check whether it lacks:

use-case context,
selection criteria,
comparison,
limitations,
FAQ,
documentation,
“when to choose” guidance.

Action: DIAGNOSE.

Type C — AI Winner / Search Weak

This is one of the most interesting groups.

It may include:

definitions,
instructions,
documentation,
case studies,
narrow use cases,
service pages.

A page can generate modest traditional traffic while still playing an important source role in AI answers.

Action: INVESTIGATE + CONNECT.

identify the problem it solves,
connect it to the correct hub,
update it,
add an appropriate CTA,
assess whether it can become a Dual Winner.

Type D — Invisible

Weak in both standard Search and AI.

Action: KEEP / UPDATE / MERGE / REDIRECT / RETIRE.

Do not delete automatically. First assess its business function, seasonality, documentation value and role in the customer journey.

7. Calculate Cross-Surface Alignment

Cross-Surface Alignment = the share of top URLs that appear in the leading group for both standard Search and AI.

In our case study:

new site: 9 of the top 10 URLs, or 90%,
established site: 3 of the top 10, or 30%.

High alignment means one content architecture works efficiently across both environments.

Low alignment can be equally useful diagnostically: generative Search may be surfacing different assets from your archive than standard Search.

8. Calculate AI URL Coverage

AI URL Coverage = AI-visible URLs / Search-visible URLs × 100

In our comparison:

new site: approximately 69%,
established site: approximately 51%.

This metric shows how much of the active Search architecture also participates in the generative layer.

9. Calculate AI Visibility Density

AI Visibility Density = AI impressions / number of AI-visible URLs

Approximately:

new site: 50+ AI impressions per URL,
established site: around 4.

The difference was roughly twelvefold.

This is not a Google metric. It is a measure of generative content productivity.

10. Marketing: stop starting with “what should we publish?”

After a Dual Search Audit, the order changes.

Ask first:

What already wins in Search?
What already wins in AI?
What wins in both?
Where is demand high but the answer weak?
Where is there a product or tool gap?

Only then build the content backlog.

11. Build a Demand Map

Combine four demand sources.

Search demand

Search Console,
Keyword Planner,
Google Trends,
SEO tools.

Prompt demand

real user questions,
AI monitoring,
a recurring prompt panel,
prompt research.

Commercial demand

email,
forms,
CRM,
sales calls,
service questions.

Social demand

YouTube,
LinkedIn,
TikTok,
Instagram,
comments,
Search Console platform properties when available.

The key point: classic keyword volume does not describe the whole market.

12. Build a Customer Problem Map

For every cluster, record:

who is asking,
what outcome they want,
what triggered the need,
what they fear,
what constraints they face,
what decision comes next.

This is the Jobs to Be Done layer.

A query for a single product can represent at least five different jobs:

understand what it is,
check the price,
compare it with an alternative,
assess fit for a process,
buy or submit an RFQ.

13. Use the Query Fan-out Map as the bridge between marketing and product

For every important problem, map the likely subproblems.

Example main question:

How should our company prepare for a new requirement?

Fan-out:

What is it?
When does it apply?
Who is affected?
What are the exemptions?
What data must be collected?
What documents are required?
What should the supplier provide?
Which products are in scope?
How do we assess compliance?
What should we do step by step?

Do not present this as a hidden Google query log. It is a problem model and an information coverage plan.

14. Marketing sees a question; Product Development should see a product signal

If users repeatedly ask:

How do I check whether my solution meets condition X?

Marketing may create a guide.

Product Development should ask:

do we need a validator?
a calculator?
a configurator?
a document generator?
a new service bundle?

This is how Search becomes an input to Product Discovery.

15. Use Search as a Product Discovery system

Search Console can reveal signals such as:

Problem Frequency — how often a problem appears,
Feature Demand — which capabilities users look for,
Documentation Gap — which documents are missing,
Comparison Demand — what users compare,
Service Gap — recurring faults and service needs,
Product Naming — the language the market actually uses,
New Category Signal — emerging category language or a new way to frame the problem.

Once a month, marketing should give Product Development more than a traffic report. It should deliver a list of new customer problems and decisions.

16. Study differences between Search Winners and AI Winners

Example:

standard Search wins on “product X,”
AI wins on “when to choose X vs Y.”

That suggests the market does not only need the product. It needs decision qualification.

The answer may be:

a comparison page,
a selector,
a consultation,
a bundle,
a new service.

17. Build a Canonical Answer Architecture

After Demand Mapping and Query Fan-out, do not create one page for every possible question.

Build a hierarchy.

Level 1 — Canonical Guide

The best high-level answer.

Level 2 — Decision Pages

comparisons,
selection guides,
variants for different customer segments.

Level 3 — Application Pages

specific use cases,
industries,
operating conditions.

Level 4 — Products

specific solutions,
parameters,
configurations.

Level 5 — Evidence

documentation,
research,
tests,
case studies,
proprietary data.

Level 6 — Action

calculator,
selector,
form,
Direct RFQ.

18. Add an Evidence Map

For every important claim, record:

source,
date,
author or data owner,
evidence type,
confidence level,
next review date.

Prefer:

primary sources,
official documentation,
manufacturer data,
proprietary measurements,
first-party case studies,
documented expert knowledge.

Content becomes more valuable when it is not simply another summary of information that already exists everywhere else.

19. Build an Actionable Content layer

Every important cluster should contain at least one executable asset.

Instead of only:

“how to calculate X” → build a calculator,
“what to ask a supplier” → build a request generator,
“how to choose a product” → build a selector,
“how to run an audit” → build a checklist or wizard.

Marketing stops being only a publisher. It becomes a provider of tools that help the user complete a task.

20. Then build a Social Topical Map

Only after you know:

Dual Winners,
Search Winners,
AI Winners,
query fan-out,

decide which topics deserve expansion beyond the website.

YouTube

demonstrations,
tutorials,
comparisons.

business interpretation,
ROI,
case studies,
executive implications.

Shorts / Reels

one question,
one myth,
one mistake,
one parameter.

Website

canonical source,
structured information,
current commercial data,
conversion.

A Social Topical Map should not be a separate social calendar. It should expand problems whose value has already been validated or strategically justified.

21. Add a Conversion Map

For every important URL, define:

Micro action

watch a video,
download a checklist,
use a calculator.

Qualification action

select a variant,
provide parameters,
check compatibility.

Conversion

consultation,
test,
RFQ,
purchase.

Do not stop reporting at impressions and clicks.

22. In B2B, add a Direct RFQ layer

If the user or agent already knows what is needed, simplify the next step.

A structured RFQ should capture:

product or solution type,
use case,
parameters,
quantity,
location,
target date,
additional requirements,
contact details.

The goal is to shorten the path:

Search → Answer → Qualification → RFQ.

23. Create a shared Marketing + Product backlog

Signal	Marketing response	Product Development response
New definition/problem	guide	category naming
High comparison demand	comparison page	selector
Cost questions	ROI article	calculator
Documentation questions	knowledge page	document pack
Compatibility questions	FAQ	configurator
Repeated errors	troubleshooting	product/UX improvement
New use case	case study	new variant/service
AI visibility gap	answer enhancement	richer product data

SEO becomes more than a marketing function. It becomes part of Product Discovery.

24. Use a 30-day experimental cycle

Do not change 100 pages at once.

Select:

5 Dual Winners,
5 Search Winners / AI Weak,
5 AI Winners / Search Weak.

Record a baseline for each.

Apply one coherent class of change, for example:

content refresh,
Evidence Layer,
internal linking,
microtool,
CTA improvement,
video asset.

After 28–30 days compare:

Search impressions,
Clicks,
CTR,
AI impressions,
Conversions.

25. Introduce a 90-day strategic rhythm

Month 1 — Diagnose

Dual Search Audit,
Demand Map,
Query Fan-out,
URL classification.

Month 2 — Build

Canonical Pages,
Evidence,
Microtools,
architecture improvements.

Month 3 — Expand

Social Topical Map,
Product Layer,
RFQ,
measurement and conversion.

Then repeat the cycle.

26. Your marketing dashboard should now have two layers

Search Layer

impressions,
clicks,
CTR,
conversions.

Generative Layer

AI impressions,
AI Share of Search Visibility,
AI-visible URLs,
AI URL Coverage,
AI Visibility Density,
Cross-Surface Alignment.

Do not manufacture an “AI ROI” if the available data cannot support it.

27. Product Development dashboard

The product team should see:

most frequent problems,
rising clusters,
qualification questions,
documentation gaps,
compared solutions,
microtool opportunities,
new category opportunities,
signals from Search and sales.

The monthly marketing report should evolve from:

“we had 10,000 impressions”

to:

“here are 10 customer problems growing in Search and AI, and 3 of them look like product opportunities.”

28. Five questions for the monthly Marketing + Product meeting

What is growing?
Which answers are still missing?
What does AI use differently from standard Search?
Which problem requires a product or tool rather than another article?
Which insight can become revenue, differentiation or a new category?

29. Do not build a separate “AI SEO department”

In our case study, the new site had 90% overlap between its top Search and AI pages.

That is a strong operational argument for a combined process:

SEO + Content + Product + Analytics

rather than isolated silos:

SEO vs GEO vs AEO vs AI SEO.

30. What should a strong content asset contain in 2026+

Not special “AI chunks.”

Not artificial fragmentation.

Not a separate page for every prompt.

A strong asset should contain:

a clearly defined problem,
a direct answer,
explanation,
decision criteria,
data,
evidence,
limitations,
freshness,
examples,
a next step.

31. How to score a new topic before publication

Score each factor from 0 to 5.

Demand

Are people searching for it?

AI relevance

Is the problem complex enough to require synthesis?

Commercial value

Does it lead to a product or service?

Evidence advantage

Do we have proprietary data or experience?

Product opportunity

Can we create a tool, service or new product variant?

Competitive gap

Are existing answers incomplete?

Actionability

Can the user take a next step?

Prioritize topics with the highest combined value.

32. Master sheet — recommended columns

Problem.
Persona.
Job to Be Done.
Search query cluster.
Search impressions.
Clicks.
CTR.
Average position.
AI impressions.
AI Share.
Search tier.
AI tier.
Dual Search classification.
Query Fan-out branch.
Canonical page.
Evidence.
Content gap.
Product gap.
Microtool opportunity.
Social format.
CTA.
Conversion.
Owner.
Review date.
Status.

33. How to recognize the highest-value opportunity

The most interesting pattern looks like this:

the problem is growing,
AI begins to use your content,
standard Search is also growing,
sales teams hear the same question,
existing market solutions are incomplete.

This may no longer be only a content opportunity.

It may be a category opportunity.

Marketing should then work with Product Development and leadership.

34. What this means for new product development

Search and AI can signal demand for:

a new bundle,
a new service,
a new documentation package,
a calculator,
a new support tier,
a configurator,
a new product category,
a consultation service,
a new commercial model.

SEO research can become an input to Product Discovery.

35. First 30 days checklist

Week 1

export both Search Console reports,
align date ranges,
join URLs,
calculate AI Share,
calculate AI URL Coverage.

Week 2

identify Dual Winners,
Search Winners,
AI Winners,
Invisible URLs,
calculate Cross-Surface Alignment.

Week 3

build Query Fan-out maps for 3 strategic clusters,
prepare an Evidence Map,
identify content gaps and product gaps.

Week 4

Launch:

1 canonical-page update,
1 new decision page,
1 microtool,
1 video/social asset,
1 improved CTA or RFQ flow.

Then measure the next full period.

36. How to make a case study and guide genuinely valuable

Publish the methodology

Do not publish only the result. Show the formula, date range, scope and limitations.

Show anonymized source data

Search Console screenshots or tables significantly increase credibility.

Share a worksheet

Readers should be able to reproduce the analysis on their own site.

Publish a 90-day update

A snapshot becomes a longitudinal case study.

Add business outcomes

Ultimately the most important metrics are:

leads,
RFQs,
pipeline,
revenue.

State limitations clearly

Methodological caution makes the material stronger, not weaker.

37. Final conclusion

We do not need two independent strategies called SEO and AI SEO.

We need one better information architecture.

The target model is:

Demand Map → Customer Problem Map → Keyword + Prompt Research → Query Fan-out Map → Canonical Answer Architecture → Evidence Map → Dual Search Measurement → Social Topical Map → Product/Action Layer → Direct RFQ → Conversion → Governance.

In this model:

traditional SEO provides findability,
AI Search expands retrieval and synthesis,
social media expand discovery surfaces,
tools help complete tasks,
Product Development uses demand signals,
Direct RFQ converts interest into qualified commercial intent.

The goal is not to create “more content for AI.” The goal is to increase the number of assets that are the best answer for a specific customer decision — regardless of whether they are discovered through classic Search, AI Mode, an AI Overview, video or an agent.

38. Case study methodology note

This guide is based on two anonymized B2B niche websites.

Standard Search data covered July 14–August 10, 2026. For AI Share calculations, we used the exact common period July 14–August 9, 2026.

Common-period data:

new site: 9,763 overall Search impressions and 2,834 AI impressions,
established site: 24,613 overall Search impressions and 1,003 AI impressions.

AI impressions are part of the overall Search Performance report, not an additional pool to be added on top.

Metrics such as AI Share of Search Visibility, AI URL Coverage, AI Visibility Density and Cross-Surface Alignment are SalesBot working diagnostic metrics. They are not official Google metrics or ranking factors.

39. Sources

Google Search Central — Generative AI performance reports in Search Console: https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports
Google Search Central — Guide to optimizing for generative AI features in Google Search: https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
Google Search Central — AI features and your website: https://developers.google.com/search/docs/appearance/ai-features
Google Search Console Help — Performance report: https://support.google.com/webmasters/answer/7576553
Google Search Central — Search Console and Google Analytics: https://developers.google.com/search/docs/monitor-debug/google-analytics-search-console
Google Search Central — Creating helpful, reliable, people-first content: https://developers.google.com/search/docs/fundamentals/creating-helpful-content

Zobaczmy, gdzie w Twoim ekosystemie tracą się wartościowe zapytania

Napisz, co sprzedajesz, do jakich firm chcesz docierać i jak obecnie powstają zapytania. Sprawdzimy, czy największy potencjał znajduje się w stronie produktowej, nowym klastrze tematycznym, widoczności w Google i AI, konwersji, analityce czy w połączeniu kilku elementów.

Sprawdź potencjał swojej firmy

Kontakt bezpośredni: kontakt@salesbot.pl

Hello world!