1.4 Million Prompts Later The Answer Is Still It Depends And Nobody Will Tell You What It Depends On.

Somewhere between the millionth ChatGPT prompt asking "how do I rank number one" and the 1.4 millionth variation of the same question phrased with more desperation, the AI did what every SEO consultant has done since 2003: it said it depends.

The difference is the AI doesn't bill you $5,000 to say it.

And neither version tells you what the fuck it depends on.

We are living in the golden age of non-answers delivered at scale. AI has democratized the ability to sound authoritative while saying absolutely nothing. It took humans decades to perfect that skill. GPT-4 did it in a weekend.

ThePromptBros discovered they could ask an LLM "what's the best SEO strategy" and get a 2,000-word response that contains every possible answer and therefore zero useful ones. They screenshot it. They post it. They call it insights.

The AI is just regurgitating the same articles written by people who learned SEO from people who never ranked anything. It's a human centipede of search marketing advice, and you're eating at the back of the line.

The Tools That Check The Tools That Check The Mentions That Nobody Reads

There is now a tool to check whether AI mentions you, which means somewhere a SaaS founder saw the anxiety around AI search and thought "I can charge for that."

And they were right.

Because SEO has always been about selling people solutions to problems they don't understand, using metrics they can't verify, based on data that refreshes slower than their panic.

You want to know if ChatGPT recommends your brand? Cool. What are you going to do when it doesn't? Write 50,000 words of "helpful content" and hope the next training scrape picks you up? Optimize for the llms.txt file that nobody has proven actually works?

The answer, of course, is it depends.

On what? Shut up and buy the enterprise plan.

Everyone Is An Expert Now And None Of Them Agree

You can watch two SEO "thought leaders" with a combined Twitter following the size of a small nation argue about whether meta descriptions impact rankings. One will cite a study that analyzed 10 million pages. The other will cite a different study that analyzed 11 million pages and came to the opposite conclusion.

Both are wrong. Neither matters. Meta descriptions don't rank your page. They convince someone to click. Or they don't. The dependency isn't the algorithm. It's whether the person searching gives a shit about your 155 characters of self-congratulation.

But we can't sell a course on "write something that doesn't sound like a robot had a stroke." We can sell a course on "meta description optimization secrets the pros don't want you to know."

The pros don't care. The pros are busy actually ranking things.

Prompt Engineering Is Astrology For People Who Took One Python Course

The new grift is prompt engineering. Not the actual skill of coaxing useful output from an LLM—that's real work. The grift is the idea that there's a secret syntax that unlocks Google's algorithm if you just ask ChatGPT the right way.

"Act as an SEO expert with 15 years of experience. Analyze this keyword and tell me the exact steps to rank."

The AI doesn't have 15 years of experience. It has 15 years of blog posts written by people pretending to have experience. It's a confidence game played with matrix multiplication.

You're not unlocking hidden knowledge. You're getting a very expensive Magic 8-Ball that has read every Snake Oil Seller's LinkedIn carousel and learned to speak fluent bullshit.

The answer is still it depends.

On your industry. Your competition. Your site's authority. Your backlink profile. Your content quality. Whether Google decided your entire niche is spam this Tuesday. Whether OpenAI's crawler even bothered to visit your site during the last training run.

But nobody wants to hear that. They want the prompt that works. The hack. The shortcut. The thing that makes ranking easy.

Ranking was never easy. AI didn't change that. It just made it easier to sound like you know what you're doing while accomplishing nothing.

What The Data Actually Shows When You Stop Lying About It

Here's what 1.4 million prompts about SEO actually reveal: most people are asking the wrong questions.

"How do I rank for [keyword]?"

Wrong question. The right question is "Why would Google rank me for this keyword when 47 other sites have been targeting it for three years with better content, more links, and an actual brand?"

The AI won't tell you that. The AI will generate a content brief. It will suggest header tags. It will recommend word count. It will do everything except tell you that the page that converts is the one that answers the question they actually typed, and you haven't figured out what that question is yet.

Search Console shows you what people searched for. It doesn't show you what they wanted. That gap is where SEO actually happens. AI can't close it. Prompts can't close it. Only understanding your customer closes it.

But understanding your customer takes time. Asking ChatGPT takes 30 seconds. Guess which one wins.

The Tools Cost More Than The Results They Promise

You know what's hilarious? The amount of money spent on keyword tracking tools that refresh slower than Google changes the results.

You're paying $500 a month to watch numbers bounce around like a drunk toddler on a trampoline. Position 3 on Monday. Position 7 on Tuesday. Position 4 on Wednesday. Cool. What did you do with that information?

Nothing. You put it in a report with green arrows and sent it to someone who doesn't read it.

The tools exist to give you something to do. Something to track. Something to screenshot for the client call. They don't exist to make you rank better. They exist to make you feel like you're working on ranking better.

There's a difference.

Nobody Will Tell You What It Depends On Because Then You Wouldn't Need Them

Let's cut through it.

When someone says "it depends," what they mean is:

"The real answer is complicated, specific to your situation, requires actual analysis, and I either don't have time to explain it or I'm not sure myself."

Gurus say it because the real answer doesn't fit in a LinkedIn post. Agencies say it because the real answer is "we'd have to audit your entire site and we haven't done that yet." AI says it because it depends is statistically the most common phrase in SEO content and the model learned to parrot it.

The truth is ranking depends on:
Whether your content is better than what's already ranking.
Whether Google trusts your domain.
Whether your technical SEO is catastrophically broken or just mildly fucked.
Whether your backlink profile looks earned or bought.
Whether the keyword you're targeting is actually what people search for or what you think they search for.
Whether you're in a niche Google decided to nuke this quarter.
Whether you have the patience to wait six months or the delusion to expect results in six days.

That's what it depends on. But explaining all that doesn't sell courses. Saying "it depends" and moving on does.

The Reports Look Great And Mean Nothing

Your agency sent you a 47-page report. It has charts. Graphs. A beautiful dashboard nobody logs into. 200 keyword rankings you care about three of.

Somewhere on page 38 is the number that actually matters: revenue. And it didn't move. Or it went down. Or it went up but nobody can explain why because correlation is not causation but we're going to pretend it is for the sake of this retainer.

The executive summary exists so nobody has to explain what SEO actually did this month. It's a magic trick. Traffic goes up. Rankings stay flat. Conversions go sideways. And everyone nods because the arrows are green and who wants to be the person who admits they don't understand what "impressions increased 14% MoM" actually means for the business.

You spent four hours building a report nobody will read. You could have spent those four hours fixing the thing that's actually broken.

But fixing things doesn't generate reports. And without reports, how does anyone know you're working?

AI Is Just The Latest Thing We're Pretending To Understand

Before AI it was Core Web Vitals. Before that it was mobile-first indexing. Before that it was Panda, Penguin, Hummingbird, and whatever bird-themed algorithm update Google rolled out while lying about what it actually did.

The pattern is always the same:
Google announces something vague.
The SEO industry freaks out.
The gurus sell the solution before anyone knows the problem.
Six months later we find out the update did something completely different than what Google said it would do.
Nobody apologizes. Everyone moves on to the next panic.

AI search is just the newest thing. Everyone is scrambling to "optimize for AI" without knowing what that means. The tools are already here. The courses are already being sold. The LinkedIn carousels are already claiming "we analyzed 500,000 ChatGPT responses and here's what we learned."

What they learned is that people will pay for certainty in an uncertain landscape. And "it depends" doesn't sell as well as "do these five things and AI will recommend your brand."

Spoiler: AI doesn't give a shit about your five things. It regurgitates what it was trained on. You want to rank in AI? Be the thing it read during training. Be big. Be authoritative. Be cited. Be linked. Be the answer so obvious that even a probabilistic language model stumbles onto you by accident.

Or just keep asking better prompts. See where that gets you.

The Real Dependency Nobody Talks About

You want to know what SEO actually depends on?

Whether you're willing to do the work nobody wants to do. Fix the ugly page that converts instead of redesigning it into oblivion. Cut the form fields sales demanded. Speed up the site instead of adding another parallax scrolling effect your developer loves. Write content that answers the actual question instead of the keyword you want to rank for.

SEO depends on whether you're optimizing for the algorithm or for the human who has to use your site after they click. The algorithm gets you the click. The human decides whether you get the conversion.

Most people optimize backward. They chase rankings and wonder why nothing converts. They get the traffic and lose the sale. They win the impression war and lose the revenue battle because their retargeting budget is zero.

Your About page is a love letter to yourself and your customer skipped it to find the price. You mapped the user journey and the user just wanted to know if you deliver to their zip code. You added micro-interactions to every button and called it premium UX. The user called it slow and left.

That's what it depends on. And nobody will tell you that because it doesn't sound like expertise. It sounds like common sense. And you can't charge $2,000 for common sense.

So What Do You Actually Do?

Stop asking AI to solve problems you haven't diagnosed. Stop buying tools that track metrics you don't act on. Stop attending webinars where someone who has never ranked anything explains the latest Google update using a deck full of guesses.

Do the audit. The real one. Not the one that spits out a score. The one where you actually look at what's broken. Fix your site speed. Unfuck your internal linking. Rewrite the content that sounds like it was extruded from a content factory and tastes like it too.

Figure out what question your customer is actually asking. Then answer it. Clearly. Quickly. Without fourteen paragraphs of throat-clearing about your brand values and commitment to excellence.

And when someone tells you "it depends," ask them what it depends on. If they can't answer, they're guessing. If they won't answer, they're selling.

You don't need 1.4 million prompts. You need one good answer. And the only way to get it is to stop outsourcing your thinking to tools that don't think and experts who don't know.

The answer is still it depends. But now you know what to depend on: your ability to figure it out yourself.

Nobody's coming to save you. The gurus are busy selling the next course. The AI is busy hallucinating best practices. And Google is busy pretending the algorithm makes sense.

What happens when someone who has been ranking things for fifteen years gets tired of watching people who never ranked anything sell courses about it? This. NeverIndexed. Where the only thing we're selling is the truth nobody else will say out loud.

Welcome to the fight.

Frequently Asked Questions

Why do SEO tools give different answers to the same question?
Because they're measuring different things, using different data sets, refreshing at different intervals, and optimizing for different outcomes. One tool tracks rankings daily but samples a limited set of locations. Another tracks weekly with broader coverage but slower updates. They're both "right" within their methodology and both useless if you don't understand what they're actually measuring. Most people don't. They just pick the tool with the prettier dashboard and the answer they wanted to hear.
What does 'it depends' actually mean in SEO and why won't anyone explain it?
It means the answer is specific to your site, your industry, your competition, your technical setup, your content quality, your backlink profile, and about fifteen other variables that require actual analysis to diagnose. Nobody explains it because the real explanation takes time, requires expertise, and doesn't fit in a tweet or a LinkedIn carousel. It's easier to say "it depends" and move on than to do the work of figuring out what it depends on for your specific situation. Explaining dependencies doesn't scale. Selling courses does.
Are AI-powered SEO tools just guessing or do they actually know what works?
They're pattern-matching based on training data that includes every SEO blog post, case study, and forum thread ever written—most of which contradict each other. The AI doesn't "know" what works. It predicts what sounds plausible based on what it's read. If the training data is full of outdated advice, guru nonsense, and correlation-not-causation case studies, that's what you get back. AI tools are confident. That doesn't make them correct. They're very good at sounding authoritative while saying nothing you couldn't have Googled yourself in 2019.
How can I tell if SEO advice is real or just repackaged guru nonsense?
Ask if the person giving the advice has actually ranked anything recently. Not "built an agency." Not "spoke at a conference." Not "has 50,000 LinkedIn followers." Ranked something. Real SEO advice includes specifics: the industry, the competition level, the timeline, the challenges, the things that didn't work. Guru nonsense is generic: "create great content," "build quality links," "optimize for user intent." If the advice sounds like it could apply to any site in any niche at any time, it's useless. If it includes dependencies and caveats, it's probably real.
Why do SEO experts contradict each other on basic ranking factors?
Because Google's algorithm is a black box that changes constantly, ranking factors vary by query type and industry, and most "experts" are extrapolating from their own limited experience or citing studies that measured correlation instead of causation. One expert works in local SEO where Google Business Profile dominates. Another works in SaaS where backlinks and content depth matter more. They're both right within their context and both wrong if you generalize their advice to every situation. The contradiction isn't the problem. The problem is nobody admits their expertise has boundaries.
Is prompt engineering for SEO just another way to avoid doing real work?
Yes. Real prompt engineering is a skill—getting useful output from an LLM requires understanding its limitations and how to structure queries. But the SEO version is mostly people asking ChatGPT to do their keyword research, write their content briefs, and generate their meta descriptions because it's faster than thinking. The AI isn't doing strategy. It's doing busy work at scale. If your entire SEO process can be replaced by a well-crafted prompt, you weren't doing SEO. You were filling in templates and calling it optimization.
What should I actually depend on when everyone says it depends?
Depend on your ability to diagnose what's actually broken on your site. Run the audit. Fix the technical issues. Improve the content that isn't answering the user's question. Build links that make sense for your industry. Test what works for your audience instead of copying what worked for someone else's. Depend on data you can verify and actions you can measure. And when someone says "it depends," ask them what it depends on for your specific situation. If they can't answer, they're guessing. If they won't answer, find someone who will.