Written by: Arjun Karnik, Growth Marketing Specialist
Key Takeaways From Arjun Karnik’s Test Lab
- Topical authority comes from expert coverage of a subject, not from domain authority scores or backlink counts. No public topical authority score exists.
- Results depend on mapping and covering the full fan-out question space instead of chasing keyword volume or raw publishing volume.
- Regular content audits using Google Search Console reveal cannibalization, gaps, and decaying pages that need consolidation, rewrites, or refresh.
- Freshness is critical: pages can lose 78–99% of their value in two months without maintenance, while consistent updates drive AI citations and impressions.
- Arjun Karnik runs a public test lab on his own site and publishes both wins and misses so the method can be inspected in the open.
Talk Through Your Own Question Map
Why Topical Authority Is A Coverage-And-Freshness Game
The problem most founders face is structural: they are optimizing the wrong variable. They publish content, follow the “publish and interlink” advice, and watch it fail to compound. The retrieval layer does not reward volume alone.
A single buyer prompt triggers dozens of hidden fan-out queries underneath it. The answer a retrieval layer assembles comes from what those sub-queries return, so coverage of the mapped question space is what compounds. Pages that rank for both the main query and fan-out sub-queries account for 51% of AI Overview citations, while pages ranking only for the main query account for under 20%. The same coverage now determines whether ChatGPT, Google AI Overviews, Perplexity, and Gemini cite a business, not just whether Google ranks it.
Pew Research Center tracked 900 US adults across 68,879 Google searches in March 2025 and found an 8% click rate when an AI summary appeared, against 15% without one. G2’s March 2026 survey of 1,076 B2B software buyers found that 71% use AI chatbots for software research, and 69% switched their intended vendor based on what the assistant told them. Being in the answer functions as a vendor-selection event, not just a visibility metric.
How To Build A Topical Map From Fan-Out Queries
The core differentiator in this build sequence is where the query list comes from. The target is the machine’s questions, the sub-queries a retrieval layer runs when assembling an answer. Keyword tool volume data reflects what buyers type, but not what the system actually retrieves against. Extract fan-out queries directly from ChatGPT instead.
The extraction method runs in four steps:
- Prompt the assistant with the buyer’s real question.
- Capture the sub-questions it retrieves against.
- Group those sub-questions into themes.
- Deduplicate the themes against existing coverage.
The output of that process becomes the production queue. It produces a map of the questions the retrieval layer is trying to answer, instead of a keyword list sorted by volume. Query fan-out is the mechanism most marketers miss: a single buyer prompt triggers dozens of hidden fan-out retrieval queries, and the answer is assembled from what comes back.
The table below shows how that map turns into a pillar-and-cluster structure, with each cluster page tied to a specific buyer-language question.
| Pillar | Cluster Page | Buyer-Language Question Answered |
|---|---|---|
| How To Build Topical Authority | How To Build A Topical Map From Fan-Out Queries | “What questions does AI actually retrieve against?” |
| How To Build Topical Authority | How To Audit Existing Content For Topical Gaps | “Which of my pages are cannibalizing each other?” |
| How To Build Topical Authority | How To Keep Topical Authority From Decaying | “Why did my rankings drop without any changes?” |
Internal linking rules are concrete: use descriptive anchor text, keep every page reachable within a few clicks, and link sideways between cluster pages as well as up to the pillar. Every supporting page links back to the pillar, the pillar links out to every cluster page, and related supporting pages cross-link to each other.
On Arjun’s own site, pages rewritten to match extracted ChatGPT fan-out queries earned citations while control pages did not. A page titled “What is GEO” was relabelled “How To Get Your Business Recommended By AI Search,” with the slug, title, H1, and H2s all realigned to buyer questions. Citations followed within weeks. These are his findings from his own test lab, not general laws about how the web behaves.
Buyer-language alignment is not cosmetic. AI Overviews match at the passage level rather than the page level, so a page “about” a topic loses to a page with a paragraph that directly answers the exact question. Jargon in the title becomes a barrier at exactly the moment the machine is matching a question to an answer.
How To Audit Existing Content For Topical Gaps And Cannibalization
No competitor gives a concrete method for this audit. Here is one. Open Google Search Console and look for four specific signals:
- Pages with impressions but near-zero clicks. This is the “impressions up, clicks down” scissors. The content is being consumed to construct answers and not sending traffic back. Flag every page where impressions are climbing while clicks are flat or falling.
- Queries where two or more of your own URLs alternate in position. A strong signal of active keyword cannibalization is when the URL Google chooses to rank for a query keeps switching between two pages on the same site on different days. Filter the Performance report by a target query, switch to the Pages tab, and look for multiple URLs pulling impressions simultaneously.
- Pages with impressions but no coverage of the mapped fan-out question. Cross-reference your Search Console query data against the fan-out map. Any mapped question with no ranking URL becomes a gap page to queue for production.
- Pages whose impression curve is declining month over month. These are decay candidates. Catch them at 20% decline. Catching a page at a 20% loss gives far more to work with than catching it at 70%.
Each finding maps to a specific action. Cannibalized pages should be consolidated into one stronger URL with a 301 redirect from the weaker one, because two competing pages split the signals that a single page would concentrate. Thin pages that already have impressions should be rewritten to match the mapped question language, since they have proven demand but not yet proven coverage. Gap pages, mapped questions with no ranking URL, should be queued for production in dependency order so that foundational questions are answered before the pages that build on them. This is a this-quarter task, not a theory.
Auditing fixes what is already broken, but it does not stop pages from decaying in the first place. That requires a different mechanism.
How To Keep Topical Authority From Decaying
Freshness is part of the build, not an annual chore. In Arjun’s own tests, pages dropped 78% to 99% in two months without maintenance. That decay stays invisible unless the site is instrumented for it, and by the time it shows up in a monthly report the position is already gone.
76.4% of pages cited by ChatGPT were updated within the prior 30 days. Seer Interactive analyzed 7,683 pages and 47,097 citations across ChatGPT, Gemini, and Perplexity from March to June 2026. The analysis found that 75% of cited pages had been updated within the last year, and consistently cited pages averaged under six months since their last update. Approximately 50% of sources cited for a given prompt will change within 13 weeks.
The refresh trigger is concrete. Impression-decay tripwires connect to Search Console signals and auto-queue an update when a page starts falling, instead of waiting for a quarterly audit. AI Growth Agent (Arjun discloses a partnership) runs this loop on autopilot, executing 5 to 8 autonomous actions a day. Those actions mix new articles with updates, which removes founder time from the equation rather than adding to it.
Freshness is the challenger’s lever. Competitors struggle to sustain it, while incumbents tend to neglect it. A competitor with a decade of authority and a stale library loses to a challenger publishing and refreshing at cadence, because the game resets weekly.
See The Refresh Loop Applied To Your Site
How To Measure Topical Authority Without A Score
No public topical authority score exists. Any tool offering one is reporting a proxy. The real proxies are:
- Number of pages receiving organic impressions across the mapped question space
- Percentage coverage of the mapped fan-out question space
- Citations and mentions across ChatGPT, Google AI Overviews, Perplexity, and Gemini
- Share of answer, meaning how often the business is named in the synthesized response
- AI referrers such as chatgpt.com appearing in analytics, which convert like referrals rather than cold search traffic
AI Growth Agent clients average more than 12,000 additional AI citations and mentions, over 100,000 additional bot visits, and a 20% or greater lift in impressions. These results are measured across the first twelve weeks and are cited here as AI Growth Agent’s results, not Arjun’s.
One honest caveat applies: buyers frequently copy an answer and paste a name into a browser, which lands as direct or branded traffic and never gets attributed to the AI answer that caused it. Whatever you measure is a floor, not a ceiling.
With those proxies in place, the next question is how quickly they move.
How Long It Takes To Build Topical Authority
Coverage and impressions arrive in weeks. Citations follow in one to three months. Compounding begins after month three. New articles on Arjun’s own site reached thousands of monthly Google impressions within weeks, and the GEO subfolder went from zero to the only source of new impressions on the domain in 60 days.
The market benchmarks for comparison: roughly $5,000 a month for a content engine against roughly $10,000 a month for 7 to 10 human-written articles with no refresh loop. The second number buys better prose. The first buys volume, structure, and freshness, the three things the channel actually rewards. These are category benchmarks, not Arjun’s rates.
Extending The Map To AI Surfaces And YouTube
Once the map is built and measured, the same question space can extend to other surfaces. The same mapped question space feeds every surface. YouTube accounts for approximately 23.3% of all AI Overview citations and is the most-cited content format in nearly every vertical. The questions buyers ask are the same questions they type into YouTube. A fan-out map built for AI citation also becomes a video content brief. The surfaces differ, but the question space stays the same.
Where Arjun Karnik And AI Growth Agent Fit
Arjun Karnik is a twenty-year tech marketer and former B2B software CMO who runs a public test lab under his own name. He is a person, not the lab. What he runs is a public test lab, a working demonstration of the method on his own site. The reason to trust the method is that it is self-verifying: ask an AI assistant about these topics and see who gets cited.
The reason the build survives contact with reality is cadence. AI Growth Agent executes the same 5 to 8 daily actions on autopilot. That cadence is what makes the build sustainable at scale. Arjun discloses the partnership with AI Growth Agent wherever it is named. His numbers are his; their published case studies are theirs, cited as such.
Review The Method On A Live Call
Frequently Asked Questions
Is Topical Authority The Same As Domain Authority?
No. Domain authority is a third-party metric that estimates the strength of a site’s backlink profile across an entire domain. It is not a Google ranking factor. Topical authority is earned through expert coverage of a specific subject and can be built by a small team without a decade of link equity. A site with high domain authority but thin coverage on a topic loses to a focused site that has answered the full question space around that topic.
How Many Pages Do I Need To Build Topical Authority?
You need enough pages to cover the mapped fan-out question space, refreshed continuously. That is a cadence question, not a fixed total. A narrow subject may be genuinely complete at a dozen articles. A broad subject can run past 150 articles and still have gaps. The reference cadence is 5 to 8 autonomous actions a day via AI Growth Agent, mixing new articles with updates. As noted earlier, pages can lose most of their value in two months without maintenance.
Do I Stop Doing SEO?
No. Technical fundamentals, structure, and quality content serve both traditional search and AI retrieval. What changes is the target and the metric. Content built for citation still performs in Google, and as the timeline section shows, the same content also performed in traditional Google search. The build sequence described here compounds on both surfaces.
Can AI-Generated Content Build Topical Authority?
Google penalizes low-quality content, not production method. Relevant, structured, fresh, and specific content wins regardless of how it was produced. The variables being judged are quality, structure, and freshness. Scaled low-value production can violate Google’s spam policies regardless of whether AI or humans produced it. The test is whether the content adds genuine value to the reader, not how it was written.
How Do I Know Topical Authority Is Working?
Track citations and share of answer across ChatGPT, Google AI Overviews, Perplexity, and Gemini. Track AI referrers such as chatgpt.com in analytics. Monitor impressions and decay curves in Google Search Console. No public score exists, so these proxies form the measurement system. Attach the honest caveat: buyers frequently copy an answer and paste a name into a browser, so whatever you measure is a floor. The correct response is to instrument for citations and share of answer rather than to keep grading the channel on the metric it no longer produces cleanly.
Conclusion: The Topical Authority Build Sequence
Topical authority is a coverage-and-freshness game. Every top-ranking result stops at “publish and interlink,” leaving the two things that actually determine whether coverage compounds, fan-out query mapping and decay, completely unaddressed.
The execution sequence that compounds coverage is straightforward. Extract fan-out queries directly from ChatGPT, build the topical map from those questions, audit the existing library against Search Console signals for cannibalization and gaps, wire impression-decay tripwires as a refresh trigger, and measure citations and share of answer rather than rankings alone.
Arjun Karnik’s public test lab runs this exact method on his own site and publishes the numbers and the misses. AI Growth Agent executes the build at 5 to 8 autonomous actions a day via AI Growth Agent. The method is self-verifying: ask an AI assistant about these topics and see who gets cited.
Apply This Sequence To Your Own Domain
