I've spent 8 years in content marketing, working with some of the best-known agencies in B2B including Grow and Convert, Foundation, and Siege Media. For the last two years, I've worked directly with Belkins as a content specialist, collaborating on strategy and producing content that consistently ranks on page 1 and brings in 5 and 6-figure contracts.

In that time, I've watched the marketing world panic about AI.

They're scrambling to add llms.txt files to their websites. Rewriting headings as questions. Publishing 300-word blog posts. Wondering how to leverage Reddit. Hiring AEO, GEO, LLMO, and AI SEO consultants who promise they know how to get ChatGPT to mention their brand.

Most of it's noise.

Here's the question to ask yourself. When someone asks ChatGPT a question about their specific situation — their industry, their problem, their budget — is your company going to come up? Or is it going to recommend one of your competitors?

The answer is not what most of LinkedIn is telling you.

The content that's winning the AI game today is exactly the same content that was winning before AI. If your content wasn't ranking on page 1 of Google and converting readers into buyers before ChatGPT, your company is not going to surface in LLM conversations now.

In this article, I go into how AI platforms and Google currently work together, how AI has changed content marketing, and how to build a winning strategy.

How AI answers questions

The assumption is that getting mentioned by ChatGPT is a completely new technical problem to solve. Something to do with prompts, structured data, or some secret handshake with OpenAI's algorithm.

It isn't.

When a user asks ChatGPT for a product recommendation, a vendor comparison, or a solution to a specific business problem, the model doesn't rely on what it already knows. It searches the web in real-time. Just like a user would on Google.

OpenAI has confirmed that ChatGPT searches the web automatically whenever it determines that a user would benefit from current information.

Perplexity lists its sources openly for every response, and those sources almost always correlate with what's ranking on page one of Google.

Google's AI Overviews are essentially a summarizer sitting on top of its own search results.

This is the mechanism. AI visibility requires Google visibility.

Google rankings determine your company's visibility in AI

If you rank on Google's first page for a given keyword, then you become visible to AI when users ask question related to that keyword.

Grow and Convert analyzed over 400 bottom-of-funnel keywords across 16 clients and found that when a client ranks on Google's first page, they appear in ChatGPT and Perplexity responses for that same keyword 77% of the time. When they rank in the top 3 positions, that number jumps to 82%.

Traditional search by "googling" questions isn't going anywhere, either. According to Datos' State of Search Q4 2025 report — based on clickstream data from tens of millions of desktop users — traditional search held steady at roughly 10% of all desktop activity in both the US and EU throughout 2025, with only minor quarter-to-quarter fluctuation. Google accounts for 93–95% of all desktop searches in the US. AI tools, while growing, represented less than 1% of total desktop events as of December 2025.

The race to "optimize for AI" is really just the same race it has always been. Rank on Google. The difference is now you have to rank on Google to rank on AI.

86% of all citations in AI responses to product-related queries come from industry-specific domains — not Reddit, not Forbes, not Wikipedia — the companies and publications that rank for those keywords.

Grow and Convert analyzed 2,440 domains cited across five different clients and found that 59% came from direct industry vendors and 27% from industry-adjacent publications. Only 14% came from general sites. For Wrike, 97% of citations came from within the project management industry. For Toro TMS, 95% were direct vendors in the trucking and logistics space.

If you want AI to recommend your company, you need to rank on Google for the keywords your buyers are searching. That part hasn't changed. What's changed is how to craft the content and how to track it.

What was winning before AI is still winning now

Before I get into what's changed, I want to make the case for what hasn't because so many companies are abandoning top-of-funnel strategies and declaring that "SEO is dead."

Bottom-of-funnel content converts a higher percentage of readers into buyers

Top-of-funnel content was never generating much in the way of revenue. It was generating traffic. Vanity metrics. Impressions. Shares. It made marketing teams look busy and gave executives something to put in a report.

It stopped being effective at growing B2B businesses before 2020.

Top-of-funnel content used to at least generate traffic. Now it doesn't even do that. Both ChatGPT and Google AI Overviews answer informational questions directly, without sending the user anywhere.

When someone asks ChatGPT "what is content marketing," the model answers the question completely, cites nothing, and recommends no one. No brand appears. No link gets clicked. The user gets their answer and moves on.

Bottom-of-funnel content means articles targeting keywords that signal buying intent, like "best [product category] software," "[competitor] alternatives," or "[use case] tools for [industry]." They convert at a fundamentally different rate. Grow and Convert has documented through multiple case studies that bottom-of-funnel keywords convert 10x to 25x better than top-of-funnel keywords in traditional SEO.

When someone asks "what's the best project management software for remote teams," AI searches the web, reads what's ranking, and makes a specific recommendation. That's the scenario where your brand can appear.

If your content strategy is built around high-volume informational topics, you're investing in content that produces zero brand visibility in AI and shrinking traffic in traditional search.

Satisfying search intent is how you rank (and get mentioned by AI)

Most SEO content is written for Google's algorithm, not for the human doing the searching.

People keyword-stuff. They chase word counts. They try to reverse-engineer what Google wants to see and produce the mechanical output of their findings. The result is content that technically checks boxes but doesn't help or interest searchers. Because it gets no engagement, it doesn't rank, and even when it does, it doesn't convert.

Google's algorithm is, at its core, a measure of user behavior. Click-through rate. Time on page. Bounce rate. Pogo-sticking back to the search results. The signal that matters most is whether real people found what they were looking for. Writing for Google and writing for humans is largely the same thing.

But understanding search intent isn't as simple as knowing whether someone wants information or wants to buy something. There's broad search intent — what the average person searching a given keyword is looking for — and then there's narrow search intent, which is specific to your target buyer.

Narrow search intent asks, who specifically is my customer, what's their knowledge gap, what pain point drove them to search, and what event likely triggered that search today?

For one of the Belkins articles I wrote on appointment setting pricing, the narrow intent wasn't "I want to understand pricing models." It was, "I need a lead generation solution because my current methods aren't working, I don't understand what I'd be paying for, and someone just told me in a meeting that we're not on track to hit quarterly targets."

That level of specificity is what produces a first-page ranking. And it's exactly the kind of content that gives AI enough context to recommend you in the right moment. That's when someone describes a situation that precisely matches what your published content articulates you solve.

Subject matter expert interviews and proprietary information are what make content unique

Now, the internet is being flooded with AI-generated content. Most of it says the same things in slightly different words. It's generic, it's forgettable, and it doesn't give a reader, or an LLM, any reason to choose you over the next result.

The answer is to write content that can't be replicated, because it contains information that exists nowhere else.

That means your content needs two things:

1. Subject matter expert (SME) interviews

2. Proprietary information

SME interviews are the informational backbone of any article worth reading. No amount of research matches what an engineer who built a product can tell you about how it works, or what a decade-long industry practitioner thinks is actually true versus commonly accepted wisdom.

If an expert pushes back on a widely-held belief during an interview, the content should lean into it. That unique perspective is the nucleus of a compelling argument that other articles aren't making.

Proprietary information means:

• Client examples

• Internal data

• Frameworks

• Specific case study figures

• Tools built in-house

It gives your content something concrete and unique. When the proprietary data in Constitution Lending's articles demonstrated that they close faster than competitors, require lower down payments, and don't reject applications at the last minute, ChatGPT began echoing that content as a sales pitch.

The more specific and differentiated your content is, the more precisely AI can match you to a buyer's specific situation.

How to adapt your content program for AI

So far, the argument is that what worked best before AI is still working now. Rank on Google, satisfy search intent, write interview-based content with proprietary information. That hasn't changed.

Here's what has.

You need more content, more specificity, and more context

Let's say your target customer start a conversation with an LLM about a product that he needs. AI then searches Google for a product-related query, such as "project management software for remote teams."

The LLM finds your company in the Google search results alongside three of your competitors. It reads all of the available content, then it recommends the one whose content best matches the specific context of the user's situation.

That means the company with the most specific, detailed, and differentiated content wins the recommendation. Not necessarily the one with the highest domain authority or the most backlinks.

And users are giving AI more context about their business needs than ever. According to Datos' Q4 2025 data, query lengths are shifting. In the US, the fastest growth is in 6–9 word queries, and 15+ word queries are becoming increasingly common and volatile — people are getting more comfortable expressing complex, specific needs directly in search. The same behavior is happening in AI tools. Users aren't asking "what's the best CRM." They're describing their company size, their industry, their current tech stack, their budget, and their timeline.

Your content needs to be a natural answer for the conversations users are having with AI. That means going deeper on use cases, industries, company types, and scenarios. It means being explicit about who you are best for, who you're not the right fit for, what situations you handle better than competitors, and what proprietary information you have that it's true.

Attribution has changed

A lot of marketing teams have been looking at their Google Search Console and Google Analytics and concluding that organic traffic is collapsing. Content must not be working anymore. Some are drawing that conclusion, cutting content, and pivoting to other channels.

But that conclusion is wrong, and it's costing them.

What's actually happening is that AI is compressing the research phase of the buying journey. Someone asks ChatGPT for a recommendation, gets your brand name, opens a new tab, and Googles you directly. That conversion lands in your analytics as branded direct traffic, not as content-driven organic revenue. The content did the work. The attribution just doesn't reflect it like it used to.

Datos' Q4 2025 data shows that searches per Google user dropped nearly 20% in the US year over year. But organic and paid clicks actually increased, peaking in Q3 and Q4. Less searching. More clicking. AI is doing the exploratory research and sending higher-intent users straight to brands. The funnel is getting shorter, but conversions from content are not disappearing.

If you're only measuring performance by traffic to blog posts, then you're not measuring the return on investment your content is producing.

The solution: start by measuring revenue from content and work backwards

Impressions and clicks are not the metrics that matter. They never were. The only number that tells you whether your content program is working is revenue. You can't see revenue from content without connecting your search data to your CRM.

The full pipeline view looks like this:

Impressions → Clicks → CTR → MQLs → SQLs → Opportunities → Deals → Revenue

The basic tech stack to track and combine all of that into a single dashboard looks like this:

Google Search Console → Google Analytics 4 → HubSpot → Looker Studio

Impressions and CTR come from Google Search Console. Conversions from organic traffic come from GA4. Opportunities and deals come from your CRM — HubSpot, Salesforce, or equivalent. Looker Studio or another data visualization tool connects to give you a real-time picture.

In the Belkins engagement, seven articles produced 30 sales-qualified opportunities within nine months of the first article being published. That number only became clear because the data was tracked end-to-end from URL clicks through to HubSpot opportunities.

Without that pipeline view, those results looked like modest traffic numbers and a handful of leads. With it, they looked like what they were: a content program generating 6-figures in revenue.

Build the dashboard in Looker Studio. Add a "how did you hear about us" field to your demo request and contact forms. Ask your sales team to ask the same question on discovery calls. More and more prospects are explicitly mentioning ChatGPT and Perplexity as the place they first heard about a brand. Sometimes, that attribution is only visible if you ask for it directly.

The brands that will win in AI search over the next few years are not the ones adding llms.txt files or trying to scale content on Reddit. They're the ones building a content program around bottom-of-funnel keywords, written articles that actually differentiate their product with specificity and proof, and set up the measurement infrastructure to see the revenue those articles produce.

That was the winning strategy before ChatGPT. It's still winning.