SEO + AI Search 2026: the complete checklist for staying visible on Google and in AI answers

Someone looking for a supplier today can stop at an answer. They type the question into Google and read the AI Overview summary, or ask ChatGPT or Perplexity, compare two or three names and only open a website at the very end. If your name is not among them, the lead never even shows up in your analytics.

This guide is for working on both fronts with a single project: staying visible in Google’s results and becoming a source that answer engines cite. It is written for people who have to decide what to do over the coming months, so every chapter ends with checks and actions. The appendices at the end (30-point checklist, content brief template, KPI table) can be used straight away with your own team or with a supplier.

Why search has changed: from the results page to AI answers

Search has changed because a growing share of questions get a written answer before any click. The website is still the destination, but the decision about who to consider is increasingly made inside the answer, where only a few sources appear.

What has happened in Italy over the last eighteen months

In Italy, the market we work in, the change came in three closely spaced steps, all documented by Google.

Date	What changed	Why it matters for a B2B company
26 March 2025	AI Overview arrives in Italy, including in Italian, for adult users signed in to their account	A summary with a few links appears above the results: the top organic position is no longer the first thing people read
8 October 2025	AI Mode available in Italy, as a search tab and in the Google app	Conversational questions, with follow-ups. Google says the questions are two to three times longer than a classic search
3 June 2026	Google introduces generative AI performance reports in Search Console	For the first time you can see impressions in AI Overview and AI Mode, page by page
31 August 2026	The report becomes available to all sites	Every company can measure its presence in Google’s AI features without external tools

Alongside Google there are the assistants many people also use for work: ChatGPT with its search feature, Gemini, Perplexity, Copilot. Each draws on a different index and cites sources in its own way, and this has technical consequences we look at in the chapter on website basics.

What happens to clicks

With a summary on the page, people click less. A Pew Research Center study of 900 US adults found that, when an AI summary appears, users click on a traditional result in 8% of visits, against 15% without a summary, and only click a link inside the summary in 1% of cases. These are 2025 US data on AI Overview, so treat them as an order of magnitude.

For a B2B company the practical point is this: fewer informational visits, but the people who arrive from an answer have often already compared options and suppliers. Visits go down, while the average quality of those who stay can go up. If you only look at traffic, you conclude that SEO “doesn’t work any more”; if you also look at impressions, brand searches and leads, the picture is different.

What it means if you sell to other businesses

In B2B the typical question is long and full of constraints: industry, size, regulations, timescales, integration with existing systems. These are exactly the questions where AI Mode and the assistants do their best work, because they break them into sub-questions and look for answers to each one (Google calls this query fan-out). Every sub-question is a chance to be cited, even if your page does not rank for the full question.

Self-assessment question. Take the three questions your sales team hears most often on a first call with a prospect. If you put them into AI Mode and ChatGPT today, who gets cited? Are you among the sources, or is it a competitor or a trade publication?

Classic SEO, AEO, GEO and AI visibility: what changes

These are four names for parts of the same job. Classic SEO gets pages into the results, AEO (Answer Engine Optimization) makes them suitable for giving direct answers, GEO (Generative Engine Optimization) works to get generative engines to pick them as a source, and AI visibility measures how much the brand appears in answers and how it is described.

The definitions we use on projects

Acronyms vary from one agency to another, so it is worth settling the vocabulary before talking budget and priorities with the team.

	Classic SEO	AEO	GEO	AI visibility
Goal	Rank pages in the organic results	Answer precise questions directly	Be chosen as a source in generated answers	Measure the brand’s presence and description in answers
Unit of work	Page and keyword group	Text passage, definition, FAQ	Citable passage plus brand authority	List of prompts and market questions
Where it shows	Google and Bing results pages	Featured snippets, voice answers, AI Overview	AI Overview, AI Mode, ChatGPT, Gemini, Perplexity, Copilot	Search Console and Bing reports, manual checks, monitoring tools
Typical metric	Rankings, clicks, CTR	Appearing in the snippet	Citations and links among the sources	Share of voice on prompts, accuracy of the description
What it takes	Technical, content, links	Question-and-answer structure	All of the above plus off-site mentions	A repeatable checking method

To go deeper into the generative side, see our guide to Generative Engine Optimization, and the glossary entry on AI Overview for anyone who needs to explain it inside the company in two lines.

What stays the same

The most useful thing to know comes from Google itself. In its guide to optimising for generative AI features, updated on 10 July 2026, it writes that optimising for generative search is still SEO. In its documentation on AI features it adds that, to appear as a link in AI Overview and AI Mode, a page must be indexed and eligible to show a snippet, with no additional requirements.

In plain terms: anyone with a technically healthy site and useful content starts with an advantage. Anyone who has neglected the basics will not catch up with a file or “AI” markup, because Google states that none is needed.

What changes

Three things change, and they are where the difference is made.

1. You compete for a passage. The engine extracts a paragraph, a table row, a definition. A long, muddled page can rank and never be cited.
2. Sources are few. A results page has ten organic links; an answer usually shows a handful. The margin is narrower, and being recognisable as a source counts for more.
3. The brand matters off the site too. Assistants also build their answers from what others write about you: publications, associations, communities, reviews.

That is why, in our GEO work, we keep technical work, content and relationships with publishers together: separately, each of the three delivers less.

Watch out. Be wary of anyone selling GEO as a service separate from SEO, with promises of “getting into ChatGPT” within a few weeks. The levers are the ones described in this guide, and they take time. Always ask what documentation a proposal is based on.

The 10 technical basics to check on your site

Before rewriting a single line, it is worth checking that the engines can read the site. A blocked crawler wipes out the chances of being cited on that engine, and often nobody notices for months.

Who reads the site on whose behalf

Each engine uses its own crawlers. Knowing which bot works for which service is the first check, because rules copied from the internet often lump them all together.

Engine	Where it gets its sources	Bot to let through	Practical note
Google AI Overview and AI Mode	Google’s index	Googlebot	Snippet rules (nosnippet, max-snippet) also apply
Gemini app	Model plus grounding on Google’s index	Googlebot; Google-Extended controls use for Gemini	Blocking Google-Extended does not remove you from AI Overview and AI Mode
ChatGPT search	OpenAI crawler and third-party search providers	OAI-SearchBot	GPTBot is about training and can be blocked separately
Copilot and Bing AI answers	Bing’s index	Bingbot	Verify the site in Bing Webmaster Tools
Perplexity	Its own index	PerplexityBot	Perplexity-User fetches pages at the user’s request

The sources are Google’s crawler documentation and OpenAI’s page on its bots. OpenAI says it takes about 24 hours for a robots.txt change to be picked up, and distinguishes OAI-SearchBot (search), GPTBot (training) and ChatGPT-User (actions started by a user). If you want to avoid training while staying citable on ChatGPT, you can block GPTBot and leave OAI-SearchBot open.

The ten-point checklist

These are the ten checks we open every audit with. The first four are the quickest and the ones where a mistake costs most.

• Robots.txt. Googlebot, Bingbot, OAI-SearchBot and PerplexityBot can access the sections you want cited.
• Decision on Google-Extended and GPTBot. Blocking them or not is a legitimate choice, as long as it is made knowing what they control and is put in writing.
• Firewall and CDN. Anti-bot protection does not turn away search crawlers. Robots.txt can be correct and the firewall can still refuse the requests: you only see it in the server logs.
• Indexing on Google and Bing. Site verified in Search Console and Bing Webmaster Tools, XML sitemap submitted to both, key pages present in the index.
• Snippets allowed. On the pages you want cited there is no nosnippet, overly low max-snippet or data-nosnippet on important blocks. Google applies these rules to AI features too.
• Content in HTML. The main text is in the page code, not locked in images, scanned PDFs or scripts the crawlers do not run. Check it with the Search Console URL Inspection tool.
• Heading structure. A single H1, H2s and H3s in logical order, each section with a heading that says what it is about.
• Internal links. Important pages can be reached in a few clicks, with descriptive anchors (“bag filter maintenance”, not “click here”).
• Speed and stability on mobile. Core Web Vitals within the thresholds on the pages that bring in leads.
• Consistent structured data. Valid JSON-LD for Organization, Person, Article, BreadcrumbList and, where relevant, Product or Service, always matching the visible text.

The extended version, engine by engine, is in our SEO checklist for AI search.

To do. Open your site’s robots.txt (your-site-address/robots.txt) and look for the lines mentioning OAI-SearchBot, PerplexityBot, GPTBot and Google-Extended. Then ask whoever manages your hosting and CDN whether any anti-bot rules are active. It takes ten minutes and can explain a complete absence from ChatGPT or Perplexity.

llms.txt and special markup

llms.txt is a proposed Markdown file that summarises a site for language models. Google’s guide states that no machine-readable files, AI text files or dedicated markup are needed to appear in Search, and none of the major answer engines has publicly stated that it uses llms.txt to choose sources. If your site generates it automatically (we do on digitalrunner.it), keeping it costs little. It belongs at the bottom of the priority list.

The same goes for schema markup: Google writes that there is no schema.org markup specific to AI features. Structured data is still useful for describing who wrote what and what type of page it is.

Watch out. The most common block we find in audits is a rule copied from an article that shuts out GPTBot, OAI-SearchBot, PerplexityBot and Google-Extended in one go. The intention was to avoid model training; the result is disappearing from ChatGPT and Perplexity search.

How to create content that answers questions, not just keywords

Content gets cited when it contains a passage that answers a precise question completely and makes sense even when read on its own. The keyword says what the page is about; the question says what it needs to contain.

From keyword to question

Keywords are still useful for understanding volumes and language. But people typing into AI Mode or ChatGPT do not enter “MES software SME”: they write a request with context and constraints. The editorial work starts there.

Keyword	The customer’s real question	What the page needs to contain
MES software SME	How do I choose an MES for an engineering company with three plants and an ERP already in place?	Selection criteria, integration timescales, what to ask the supplier about the ERP
joinery dust extraction system	What airflow do I need for a line with three CNC machines?	Airflow table by type of machining, ATEX rules on dust, installation times
GDPR consultancy for businesses	What do I need to do if a non-EU supplier processes my customers’ data?	Steps in order, documents needed, cases where legal advice is required
(write your own)
(write your own)

The last two rows are for you: try filling them in with two keywords you are working on now.

Where to find the real questions

The best questions do not come from SEO tools, useful as they are. They come from the people who talk to customers.

• Emails and quote requests from the last six months: the recurring questions, in the customers’ own words.
• The objections your sales team hears on the first and second call.
• Long queries in Search Console (filter for queries of more than six words).
• “People also ask” and related searches on Google.
• Questions customers ask technical support after purchase.

For how to classify these questions by intent (informational, comparison, choosing a supplier), see the article on search intent.

Example. Imagine a company that makes industrial filtration systems. In a month, the salesperson collects 40 questions from customers. Grouped together, they become eight themes: filter types for dust, maintenance costs, replacement frequency, emissions regulations, downtime, compatibility with existing systems, energy use, warranties. Eight themes are already the basis of a year-long editorial plan.

How to write a citable passage

A citable passage has a recognisable structure. We apply it to service and product pages before the blog, because those are the pages that bring in leads.

1. A heading phrased as a question or a clear topic. “How much does bag filter maintenance cost?” works better than “Our support service”.
2. The answer in the first one or two sentences. The reader, and the engine doing the extracting, must find the substance straight away. The detail comes after.
3. One idea per paragraph. In the sentence carrying the answer, avoid vague references like “as mentioned above” or “this approach”.
4. The right format for the content. Comparisons in a table, procedures in a numbered list, definitions in one sentence.
5. Something that cannot be found elsewhere. Data you collected, a reasoned selection criterion, a project example, an original photo.
6. A source for every factual claim. A link to the standard, the body, the manufacturer.
7. A visible update date. And a review when prices, regulations or features change.

Google makes clear that there is no need to break content into tiny fragments: its systems understand detailed pages too. The rule is still to write for a person, with clear answers at the start of each section.

Mass-produced AI content

Many companies have tried filling their blog with automatically generated text. Google’s guide is explicit: it rewards original, experience-based content and discourages what anyone could produce in a few minutes with a generative model. AI can help with structuring, summarising and checking. The contribution that makes a page citable (data, experience, informed opinion) remains human.

Self-assessment question. Open the most important service page on your site. Do the first two sentences of each section answer a question a customer would ask? Is there at least one piece of data or one example a competitor could not copy?

Entities, sources, evidence and experience: the signals that increase citability

Between two pages that answer the same question well, engines tend to cite the one from a recognisable source: a clearly identified company, real authors, verifiable data. These signals are built on the site and confirmed off it.

Entities: making clear who you are

Google and the assistants think in entities: companies, people, products, places, concepts. The more consistent the information about you, the easier it is to associate you with the right topics and distinguish you from others with similar names.

• One sentence that says what you do and for whom, essentially the same on the home page, the about page and the service pages.
• Company name, address, industries served and products identical on the website, Google Business Profile, LinkedIn and industry directories.
• Products and services always called the same thing, without switching between acronyms and trade names.
• Organization structured data with sameAs pointing to the official profiles.
• Pages linked to one another with descriptive internal links, so the site draws a map of its topics.

Experience: who signs it matters

Google’s guidelines on helpful, reliable content use the E-E-A-T yardstick: experience, expertise, authoritativeness, trustworthiness. In practice it means real bylines.

1. Every article and every technical page has a real author, with a dedicated page: role, background, topics covered, LinkedIn profile.
2. The author page is marked up with Person and linked to the Organization.
3. The same experts also appear off the site: interviews, conference talks, articles in trade publications, listed on the author page.
4. The person’s name and the company’s name appear together, always written the same way.

Evidence: what makes a claim citable

An engine readily cites a precise, verifiable claim, and almost never a slogan. Here is the difference across a few hypothetical examples.

Weak claim	Citable version	What you need to write it
“We guarantee the highest quality”	“Across checks on 40 batches in 2025, the average reject rate was 1.2%”	Aggregated production data, stated method
“Fast delivery times”	“For standard configurations we deliver within 15 working days of the order”	A true, up-to-date operational figure
“We are regulatory experts”	“The standard applies to installations above a certain threshold: here is how to check, with a link to the official text”	A reading of the standard, a link to the primary source
“Customers all over Europe”	“We work with companies in 6 countries, with support in Italian, English and German”	Verifiable information about the service

The numbers in the table are made up for the example: yours must come from your own data, with a line explaining how you collected them.

To do. Choose a service page and underline every promotional adjective (“leading”, “innovative”, “quality”). For each one, ask yourself what data or fact would prove it, and replace the adjective with that fact. Where the fact does not exist, delete the sentence.

Sources: link to those who know

When you cite a standard, a market statistic or a technical specification, link to the primary source: the text of the law, the body, the original study. An article that points to another article summarising a study is worth less as a source, for people and for engines alike. This applies to AI search statistics too: be wary of figures without a link to the study.

Editorial architecture: pillar pages, clusters and supporting content

An isolated article rarely becomes a recurring source. A group of linked content works better: a main page covering the topic (pillar), a series of deep dives on each sub-question (cluster) and supporting content answering specific doubts, all tied together with internal links.

Why clusters work with AI answers

The reason is query fan-out. When a buyer asks AI Mode “what filter do I need for the dust from a paint line, how much does maintenance cost and what regulations apply”, the system searches separately for filter types, maintenance costs and regulations. A site with a pillar page and three well-made deep dives has three or four chances of being cited in the same answer; a site with a single product page has one.

The three levels

Level	What it is	Example (industrial filtration company, hypothetical)	Goal
Pillar page	A page covering the topic as a whole, often a service page or a guide	“Industrial dust filtration: a buyer’s guide”	Rank for broad searches and act as a hub
Cluster	A deep dive on a sub-question with commercial value	“Bag or cartridge filters: how to choose”, “Maintenance costs of a filtration system”	Be cited on each sub-question
Support	Short content answering a specific doubt	Glossary entry, technical FAQ, regulatory fact sheet	Cover specific questions and strengthen the cluster
Conversion	A page with the next step	Airflow calculator, site survey request	Turn the visit into a lead

How to build the map

1. Start from a topic you have a right to talk about. Two topics covered in depth beat ten skimmed over.
2. Collect the sub-questions using the method in the previous chapter and group them.
3. Give each sub-question a single page. Two pages on the same question compete with each other and confuse the engines.
4. Link both ways. The pillar links to each cluster page with descriptive anchors; each cluster page links back to the pillar and, where it makes sense, to neighbouring cluster pages.
5. Put a next step on every page that fits the question: a technical data sheet, a calculator, a short form.
6. Review every six months by comparing AI answers to your questions with the content you have: the sub-questions without a page of yours are your next editorial plan.

You can use this table to set up your first cluster. The blank rows are for filling in.

Page	Level	Main question it answers	Does it already exist?	Create or update by	Page it links to
	Pillar
	Cluster
	Cluster
	Cluster
	Support
	Conversion

Watch out. Clusters apply to the pages that sell, too. Many companies build a rich blog and leave their service and product pages as brochures. The citations land on the articles, which convert poorly, while the commercial pages stay invisible. In your plan, put the commercial pages at the centre of the cluster.

Brand mentions, digital PR and off-site authority

Answer engines trust what third-party sources say about a brand more than what the brand says about itself. To be cited, the name of the company and its experts needs to appear in publications, associations, communities and professional platforms, linked to the right topics.

What the available studies say

Three recent studies point in the same direction.

• Muck Rack, May 2026 report on more than 25 million links cited by ChatGPT, Claude and Gemini: 84% of citations come from earned media (journalism, research, institutional sources, encyclopaedias, third-party content), while paid content and advertorials account for 0.3%.
• Ahrefs, study of 75,000 brands: presence in AI Overviews correlates mainly with brand mentions across the web (0.664), far more than with the number of backlinks (0.218). It is a correlation, not proof of cause.
• Semrush, analysis of LinkedIn pages cited by AI in early 2026: LinkedIn appears in 14.3% of ChatGPT search answers and 13.5% of Google AI Mode answers, and on both engines personal profiles outweigh company pages.

The levers that work in B2B

The most useful mentions for a B2B company come from where the industry already gets its information. These are small channels, but they speak to the right audience and AI often uses them for niche questions.

Lever	What to do	What it produces	Typical time
Trade publications	Map the 10-15 magazines your customers read, offer exclusive data or an expert to interview	Editorial mentions and links on the right topic	2-3 months for the first pieces
Proprietary data	A small six-monthly survey with aggregated data and a stated method	A reason to be cited and picked up	4-6 weeks for the first report
Associations and trade fairs	Working groups, conference talks, up-to-date exhibitor listings	Stable, well-indexed sources	Tied to the industry calendar
Podcasts and video	Guest spots on specialist podcasts, technical videos on YouTube	Transcripts and show notes with name and link	Variable
Expert profiles	Bylined articles on LinkedIn, one idea per piece	Association between person, brand and topic	Ongoing
Reviews and communities	Systematically asking for reviews, answering in forums with name and role stated	Sources for comparison questions	Ongoing

The full 90-day plan, with activities week by week, is in the article on digital PR for AI search.

What to avoid

• Bought links and paid guest posts. Google’s spam policies treat exchanging money for links that pass value as spam. Sponsored content must be marked with rel="sponsored" or rel="nofollow".
• Blanket press releases. The same text sent to hundreds of addresses produces identical copies nobody reads.
• Wikipedia shortcuts. The conflict of interest guidelines discourage people who work for the company from editing its article. An article only makes sense when independent sources already talk about you.
• Improperly solicited reviews. They get removed and leave a mark on your reputation.

Example. Imagine a maker of automation components. Every six months it publishes a short report on average machine downtime due to faults, with data aggregated from its own service call-outs. It offers the report to two trade publications, the technical director talks about it at a conference and turns it into three LinkedIn articles. Within a year the company’s name appears on several independent sources, always next to the same topic.

Self-assessment question. In the last twelve months, how many times has your company’s name, or one of your experts’, appeared on a site you do not control? How many of those mentions are about the topic you want to be cited for?

How to measure visibility, traffic and leads without chasing useless metrics

You measure on four levels: visibility (impressions in the results and in AI features), presence in AI answers (citations and brand description), qualified traffic (visits from search and assistants that take useful actions) and leads (contacts and opportunities by source). No level is enough on its own, and raw traffic is the least useful of the four.

What you can measure today

• Search Console, generative AI features report. Since 31 August 2026 for all sites: impressions in AI Overview and AI Mode by page, country, device and date. Clicks and queries are missing, so you know which pages appear but not for which questions.
• Bing Webmaster Tools, AI Performance report. In public preview since 10 February 2026: how many times pages are cited in Copilot and Bing’s AI answers, with the grounding queries.
• GA4, “AI assistants” channel group. A custom channel collecting referrals from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com, placed above the referral channel. Clicks from AI Overview and AI Mode stay within Google organic traffic.
• Manual checks. A fixed list of 20-30 questions from your market, repeated every month on the same engines, noting who is cited and how the brand is described.
• Monitoring tools. Ahrefs Brand Radar, Semrush and similar tools estimate share of citations across a set of prompts against competitors. Useful for the trend, bearing in mind that AI answers vary from one request to the next.

Useful metrics and metrics to drop

Metric	Useful?	Why
Total organic sessions	Not much, on its own	Mixes informational and commercial traffic; with fewer clicks per search it can fall while leads rise
Average site position	Not much	An average across thousands of queries says nothing about the pages that sell
AI feature impressions per page	Yes	Shows which pages Google uses as a source; cross-check with conversions
Citations on a fixed list of questions	Yes	Measures presence where decisions are made, against competitors
Brand searches	Yes	People who read about you in an answer often search for you by name later
Leads and opportunities by landing page and source	Yes, the most important	Links visibility to revenue
Number of backlinks	Not much, on its own	The quality and topical fit of the source matter more than the total
“AI visibility scores” with no stated method	No	Without knowing which prompts and which engines, the number cannot be compared or verified

Linking visibility to leads

For a B2B company a citation is worth something if it brings qualified enquiries, and those enquiries need to be recognised in the CRM. Three measures we apply on projects:

1. Source through to the opportunity. Source, medium and landing page reach the CRM as properties of the contact and the opportunity. That way you can compare close rates and timescales by source. We cover this in the article on B2B lead management metrics.
2. A field in the form. “How did you hear about us?” with options such as “Google search”, “ChatGPT or another AI assistant”, “article or magazine”, “event”. It recovers what analytics misses, because people who discover you in ChatGPT often come back by searching for the brand and show up as direct traffic.
3. A single monthly report. AI impressions, citations on the sample questions, brand searches and leads by page in the same document, read together.

To do. In Search Console, export the 30 pages with the most impressions from the AI features report. In GA4 or the CRM, check how many of them generated at least one lead in the last three months. Pages with lots of views and no leads are the first to review: they are almost always missing a clear next step.

Further reading. On how to turn a citation in AI Mode into a quote request, with pages that convert even with few clicks, we have written a dedicated article: how to get cited by Google AI Mode without losing leads.

Final self-assessment checklist

This self-assessment helps you see in ten minutes where you stand and which area to start with. Answer honestly: an unticked box is as useful a piece of information as a ticked one.

Technical

• I know which AI crawlers robots.txt allows and have checked that the firewall does not block them.
• The site is verified in Search Console and Bing Webmaster Tools, with a sitemap submitted.
• The pages that bring in leads are indexed and have no snippet restrictions.
• The main content of key pages is HTML text, not PDFs or images.

Content

• I have a list of at least 20 real customer questions, collected from sales and support.
• Service and product pages answer those questions in the first lines of each section.
• Every important page contains at least one original piece of data, example or criterion.
• I have at least one topic covered by a pillar page and linked deep dives.

Authority

• Every piece of content is signed by a real person with an author page.
• Company details are consistent across the website, Google Business Profile, LinkedIn and directories.
• In the last six months the brand has been mentioned on at least three independent industry sources.

Measurement

• I read the Search Console AI features report every month.
• GA4 has a channel collecting traffic from AI assistants.
• Every month I check a fixed list of questions on AI Overview, AI Mode, ChatGPT and Perplexity.
• In the CRM I can see the source of every lead and every opportunity.

How to read the result

Boxes ticked	Where you are	Where to start
0-5	The basics are not in place yet	Technical and measurement: without these, any content work stays invisible or unmeasurable
6-10	Solid base, citability still to build	Content on real questions and a first cluster on the commercial pages
11-13	Good set-up	Off-site authority and linking citations to opportunities in the CRM
14-15	Mature set-up	Broaden the topics, increase proprietary data, benchmark against competitors on more engines

Self-assessment question. Look at which of the four areas has the most unticked boxes. Is that where outside help makes most sense, or is it an area your team can cover on its own over the next three months?

Appendix A: 30-point SEO + AI Search checklist

Thirty checks in four areas. We suggest printing it, filling it in with the team and noting next to each item who is responsible and the date.

Technical (10)

• 1. Robots.txt: Googlebot, Bingbot, OAI-SearchBot and PerplexityBot allowed on the sections to be cited.
• 2. Written decision on GPTBot and Google-Extended, with the reason.
• 3. Server or firewall logs checked: no blocks on search crawlers.
• 4. Site verified in Search Console, XML sitemap submitted and error-free.
• 5. Site verified in Bing Webmaster Tools, sitemap submitted.
• 6. Key pages indexed on Google and Bing (URL inspection page by page).
• 7. No nosnippet, restrictive max-snippet or data-nosnippet on the blocks to be cited.
• 8. Main content in HTML, checked with Search Console URL Inspection.
• 9. Core Web Vitals within thresholds on mobile for the pages that bring in leads.
• 10. Valid, consistent structured data: Organization, Person, Article, BreadcrumbList, Product or Service.

Content (8)

• 11. List of at least 20 real customer questions, updated every quarter.
• 12. Each question has a single reference page on the site.
• 13. Each section opens with one or two sentences answering the question.
• 14. Comparisons in tables, procedures in numbered lists, definitions in one sentence.
• 15. At least one original element per important page (data, example, criterion).
• 16. Primary sources linked for every factual claim.
• 17. Visible update date and scheduled review of key pages.
• 18. At least one complete cluster (pillar, deep dives, support, conversion) with internal links both ways.

Authority (6)

• 19. A real author on every piece of content, with an author page and a linked LinkedIn profile.
• 20. “What we do and for whom” sentence consistent across home, about and service pages.
• 21. Company details identical on the website, Google Business Profile, LinkedIn and directories.
• 22. Map of the 10-15 industry publications and associations, with a contact for each.
• 23. At least one piece of proprietary content a year (report, survey, technical guide) offered to the media.
• 24. Systematic review requests to customers and a reply to every review, including critical ones.

Measurement (6)

• 25. Search Console AI features report exported every month.
• 26. Bing Webmaster Tools AI Performance report checked every month.
• 27. “AI assistants” channel in GA4, above the referral channel.
• 28. Fixed list of 20-30 questions checked every month on AI Overview, AI Mode, ChatGPT, Gemini and Perplexity.
• 29. “How did you hear about us?” field in the form, with an option for AI assistants.
• 30. Source and landing page stored in the CRM on contacts and opportunities.

Appendix B: Content brief template for citable articles

This is the template we fill in before writing an article or rewriting a page. It tells the writer, in-house or external, what the text needs to contain before they start. Copy it into a document and fill it in field by field.

Part 1: goal and audience

Field	What to write	Your brief
Working title	The main question, in the customer’s words
Page type	Pillar, cluster, support, service or product page
Reader	Role, industry, decision stage (discovery, comparison, choice)
Main question	Just one, phrased as they would ask Google or an assistant
Two-sentence answer	The summary that must appear at the start
Action after reading	What we want the reader to do (download, calculate, request a site survey)

Part 2: structure

Section (H2 or H3)	Sub-question it answers	Format (text, table, list, definition)	Original element (data, example, photo)

Part 3: evidence and sources

Claim to support	Primary source (link)	Internal data available?	Who provides it

Part 4: links and final check

• Pillar page the article links to:
• Other cluster pages to link:
• Existing pages that should link to this article:
• Author signing it, with author page:
• Expert to interview for the original contribution:
• Planned review date:

Before publishing, check:

• Each section opens with the answer to its sub-question.
• There is at least one element a competitor cannot copy.
• Every figure has its source linked or a description of how it was collected.
• The main text is in HTML, with any PDFs only as an extra download.
• The page has a clear next step that fits the question.
• Author, publication date and update date are visible.

Appendix C: KPI table: organic visibility, AI mentions, qualified traffic, leads

The indicators to track every month, split by level. Fill in the “Starting point” column with the first month’s figure and the “6-month target” column with a realistic value agreed with the team.

Organic visibility

KPI	Source	Frequency	Starting point	6-month target	Notes
Organic impressions on commercial pages	Search Console, filtered by page	Monthly			Exclude the blog to see the pages that sell
Impressions in Google’s AI features	Search Console, generative AI report	Monthly			List the top 10 pages
Citations in Copilot and Bing	Bing Webmaster Tools, AI Performance	Monthly			Note the most frequent grounding queries
Brand searches	Search Console, queries containing the company name	Monthly			Grows when mentions are working

AI mentions

KPI	Source	Frequency	Starting point	6-month target	Notes
Questions where the brand is cited (out of 20-30)	Manual check or monitoring tool	Monthly			Same questions, same engines, same method
Share of citations against 3 competitors	Monitoring tool or manual check	Monthly			Read the trend, not the single month
Accuracy of the brand description	Manual check	Quarterly			Industry, products and location attributed correctly?
New mentions on industry sources	Manual search, alerts, monitoring tools	Monthly			Separate publications, associations and communities

Qualified traffic

KPI	Source	Frequency	Starting point	6-month target	Notes
Sessions from the “AI assistants” channel	GA4, custom channel group	Monthly			ChatGPT, Perplexity, Gemini, Copilot
Organic sessions on commercial pages	GA4	Monthly			More useful than total organic
Key actions per session (downloads, calculators, contact clicks)	GA4, events	Monthly			Measures visit quality

Leads

KPI	Source	Frequency	Starting point	6-month target	Notes
Leads from organic search	CRM	Monthly			With landing page
Leads from AI assistants (referrals plus form field)	CRM	Monthly			Add the two sources without double counting
Opportunities opened from these leads	CRM	Quarterly			The figure management cares about
Close rate by source	CRM	Quarterly			Compare organic, AI, campaigns and word of mouth