The Great GEO (Generative Engine Optimization) Grift

You’re being sold a lie.

And in some cases, a lie that may end up damaging your business.

That lie is that “GEO” (generative engine optimization) is a discipline fully distinct from—or even a replacement for—SEO.

In this article I’m going to argue that GEO, at least in its most widely presented form, is:

1. nothing new
2. possibly a waste of your time
3. definitely a waste of your money
4. polluting the web
5. potentially harming your business

Why should you listen to me?

Well, you should make your own mind up. After all, I’m just some guy on the internet. But, I’ve been around the block.

• Involved in SEO since 1997.
• Ran my own Ecommerce business from 2006-2012.
• Consulted for 15 years
• Former head of content for Ahrefs and Seobilty

I’m also an AI obsessive and an early adopter.

So, make no mistake: I don’t write this from a position of ignorance regarding the tech, or a fear for my own position.

Indeed, if the latter was the case it would be simpler for me to simply pivot to selling “GEO” services like half of my peers.

But here’s why I don’t.

There’s a fair amount of rewriting of history going on at the moment.

An attempt to reframe best practices that have been part of any well executed SEO strategy for at least 10 years as shiny new “GEO” tactics.

Let’s gather some information.

We’ll start by looking at the top “geo tactics” recommended by Google’s AI mode. Seems appropriate?

Let’s log them as we go (I too can read the text in images).

Google’s AI Mode: Create expert-led, authoritative content, prioritize user intent, optimize content structure, implement structured data, use an omnichannel strategy, monitor and iterate.

Next, let’s do some deep research with GPT5.

GPT5 was wordy in its response (it was after all deep research), but here were the headlines:

GPT5: Allow and optimize AI crawling, create definitive content on your topic, optimize key pages (About, Home, Services) for AI recommendations, get featured on authoritative “best of” lists, encourage reviews and address feedback, use schema and structured data, incorporate multimedia and alternative formats, engage in industry conversations (forums/social), monitor AI mentions and refine content accordingly, avoid black-hat or over-optimization tricks

Moving on to Semrush’s AI Visibility tips (in fairness they don’t call it GEO here, but they do elsewhere which I’ll cover):

Semrush: brand mentions, content quality and originality, citations quotes and statistics, structured data, content freshness.

Finally, let’s go for the top ranking Google result for “geo tactics” at the time of writing.

Foundation Inc: Research topics relevant to your customers, create query intent based content at scale, implement digital PR activities, incorporate structured data, focus on user intent, distribute your content, embrace multimedia, leverage social media.

Now let’s boil this down a bit and extract the core “GEO tactics” from these four* sources.

This is probably the “tactic” (or category of tactics) that has the strongest argument for being distinct. It’s hard to argue that the growth in chat based interactions, and—behind the scenes—the way LLMs index (initial training) and retrieve data (real time tool calls) hasn’t introduced several new considerations for SEOs. So I will make some concessions here.

But I will also argue that it’s evolution, not revolution. And that it doesn’t require a new acronym, it’s just an extension of SEO.

Let’s start with the obvious point.

Ensuring that your content is crawlable and indexable is absolute basic, day one SEO. Whether that indexing is for Google or for ChatGPT, if your content can’t be accessed and parsed, it’s not going to show up. I think that’s a given.

And dealing with (and adapting to) the limitations of crawlers is certainly nothing new. Some of us still have shellshock from figuring out how to create indexable versions of Flash websites back in the late 90s. For a time, AMP pages were the future. And today, if we need to create HTML snapshots or chunkable JSON versions (arguably we shouldn’t have to, but I’ll set that aside for now), then that’s just an adaptation, not a sea change.

Control over crawling is also not a new paradigm. After all, choosing what should not be indexed has always been just as important for SEO as choosing what should be.

• We block content/directories in robots.txt.
• We add noindex to pages.
• We 301 redirect and consolidate.
• We ignore data-nosnippet…

I’ll concede the point that due to training cut-offs, it’s critical that information is correct at the time of a particular model’s training run. And indeed, that there are (important) decisions to be made on whether content can be used in training and/or retrieval.

But I’ll contend that the above falls under optimization and policy. Ensuring information is correct seems like general good practice (would you want wrong information on your website?). And there were always decisions to be made on content that should not be indexable - particularly sensitive content. Although feel free to disagree with me here.

Log file analysis can be important, but generally only at the enterprise level. For smaller sites it’s probably going to be overkill. But regardless, checking logs for errors and managing crawl budget has long been part of SEO. The specifics on what we look for might have changed (or been added to), but you know… change is inevitable.

Where crawling and indexing was relatively strong, arguing that structured data/schema is distinct to GEO is embarrassingly weak.

I mean, it’s patently ridiculous. In fact, we don’t really need to go much farther than Wikipedia here:

But regardless, we will.

So let’s break down the facts.

Schema has been around for over 14 years and was literally proposed by search engines themselves to help them better understand the meaning of sites.

We (SEOs) have long recommended adding relevant schema to web pages to aid machine understanding of our websites/content (current AI is just a new version of that “machine”). And yes, in some cases, that also meant getting some shiny SERP features like review stars or product prices.

Any basic SEO plugin will have fields where you can add related entities for sameAs relationship mapping/disambiguation. Of course, you have to make sure it’s all correct; but that’s just SEO.

Don’t believe me?

Here’s an article on entity disambiguation (content focused) from 2020, here’s one on entity disambiguation from 2014, and here’s the dearly missed Bill Slawski also talking about entity disambiguation in the same year.

In short: the goal, or the outcome may be slightly different. Perhaps we were optimizing for the knowledge graph. Maybe we were optimizing for local SEO. Perhaps we just wanted review stars. But the process is SEO.

Ok, let’s take a pause.

Because ignoring the hellscape language for a moment—“evidence-dense knowledge objects”, “authors are treated as entities”—(although we’ll come back to that later) this sounds pretty impressive doesn’t it.

The problem is, it’s bullshit. Observably, computationally, and technologically, it’s just not what’s happening. It’s attributing far too much intelligence to current models (more on that later), and the way they ground with search through live retrieval.

LLMs aren’t BBC Verify. They’re statistical, probabilistic models. They don’t “know” the truth. In fact they don’t really “know” anything.

And the reality is, we already chunk just fine. We already verify our facts just fine. We’ve been “chunking” for years to try and grab position zero snippets (the zero here often had a dual meaning funnily enough - hint: clicks). We’ve been verifying facts and showing proof of work/evidence to demonstrate E-E-A-T.

We’ve been making sure our bios are complete. And we also know that authors as entities has serious issues anyway as Google already tried and failed over a decade ago.

We can justifiably take a razor here to the word vomit and simplify down to:

Absolutely. Nothing. New. And most of it attributing far too much intelligence to the observable state of LLMs. Just like we’ve done with Google for years ironically.

Let’s cut through the noise here. What we’re talking about is buyer’s journey, personas, search intent, and a redefining of long-tail keywords.

I mean, just swap “prompt map” for “keyword map” and you’re half the way there. The fact that LLMs (may) crawl support tickets and community threads is a nice bonus - you get that for free.

But let’s break it down a little further.

Firstly, there is very little point in me explaining what a buyer’s journey is and the importance of it. So instead I’ll just link to some pre-LLM articles which do a better job than I can of discussing why it’s so critical for SEO:

My boy E. St. Elmo Lewis way ahead of his time there. Conducting GEO 119 years before the release of “Attention is all you need”. And I’m only half joking.

Beyond this, we’re back to verifiable claims etc (covered above already), and trying to present SEO as something which it is not. Refining and multi-turn sounds scary, until you realise it’s not hugely different to long-tail modifiers (best for X, under Y, near me, cheap), and “People also ask”.

But it does surface an immediate failing of most GEO tools (don’t worry, we’ll get to them) in that they track single-turn prompts, when in fact most “prompts” are part of a conversation. Somewhat ironically, most of these fancy new tools “track” (inverted commas for a reason) prompts just like keywords.

Anyway, the newsflash here is that keyword research (call it prompt research now if you must) is important. Who knew?

Ok, so what do we have? Chunking again, which we’ve already covered. But to reinforce, SEOs are good at structure and semantic documents. And we’ve been optimizing for snippets for years. We love marking things up.

Stable IDs and anchors? Makes sense. Although how are you dealing with multiple pricing tables on one page? Feels a bit like fluff to me, but I’ll accept that it’s logical.

Standardized components? In reality, this is just good design practice/consistent design language. Not disagreeing with the practice. Disagreeing that it’s new.

“You avoid burying critical facts mid-scroll” - Google was talking about the importance of above the fold content back in 2012.

“Constrain sentences for chunk boundaries”. How are you defining these boundaries? Are they in the room with us right now? No way this is manageable at scale. Micro-managing sentences is just stupid.

I’m getting a little shorter and more dismissive with my rebuttals now. But that’s because we’re going round in circles.

Again, we have proof of claims (E-E-A-T, medic). This time we also have the importance of key brand pages (a pillar of E-E-A-T). We want to ensure we’re mentioned in a consistent way (branding again). We want positive reviews. We need more of them?

Not much to say here. Let’s move on.

Ok, this one deserves a little more time, since it’s one of the core untruths of GEO evangelists, at least the ones that claim that SEO is dead.

The attempt to frame off-page SEO as “just link building”. Or in some cases, link spam. And the pretence that things like co-citations are new concepts.

So for this section let’s go back. Way back. A bit of a resource dump first.

We’ll start with co-citation. The framing here is links, but here’s an article discussing co-citation and its impact on SEO from the heady days of 2006.

Here’s Rand Fishkin back in 2012 discussing how co-occurence (the words surrounding mentions of your brand) may influence Google rankings - link or no link.

Here’s John Doherty talking about the power of unlinked mentions (and yes, “entities”) back in 2018.

Here’s a discussion on a Google patent from 2014 that proposes using “implied links” (i.e. citations) for rankings.

Here’s a post on online reputation management from 2017.

Here’s a blog on optimizing for Google’s knowledge graph from 2019.

Here’s a post on NAP consistency (local SEO focus) from 2015.

Here’s your humble narrator (me) discussing unlinked mentions and participating on Reddit on the Ahrefs blog in 2016.

Ok, let’s move on from the resource dump and cover some of the other earth-shattering insights.

Appearing in “Best X” lists: Well, of course. Most likely one of the first priorities for any traditional SEO campaign.

Reputable directories: contending that an SEO strategy would not target inclusion in reputable and industry specific directories is laughable.

G2/Capterra etc: part of any SEO campaign, or indeed already taken care of by the marketing team/founder. With regards to consistency, a citation audit would be part of most solid SEO strategies. And identifying the most important industry sources for trust is just part of SEO.

Publishing datasets, statistics etc: a go-to link bait strategy forever.

Off-page SEO has not been “just link building” for well over a decade. Anyone who tells you otherwise is seriously misrepresenting history.

Call some of it digital PR if you want. Call some of it reputation management. But it ain’t new. And a high-level SEO campaign would cover it all.

Content repurposing has been a part of SEO for years.

Here’s a 2019 article on content repurposing from Constant Content.

Here’s a solid guide to optimizing multimedia content for SEO by Bruce Clay from 2014.

Here’s me recommending repurposing content in 2016.

“Repurpose the post into a video, blog, slideshare presentation etc”.

Of course you want to make sure your repurposed content is optimized. Because, you know, the O in SEO stands for “Optimization”.

If you want to ensure fact parity etc, then knock yourself out. But that’s just policy.

All good advice.

All just go where the discussion is in your niche and participate (be part of the conversation), and create content based around real questions/problems buyers are asking/facing.

Questions and answers, and problems and solutions are the fundamentals of SEO/content marketing and go back to the buyer journey that we covered earlier.

Ultimately, we were all participating on Reddit, answering questions on Quora, and creating content to match user/buyer intent long before some genius thought that “GEO” would be a good acronym that wouldn’t in any way be difficult to disambiguate…

Or was it just what we’ve all been doing for well over a decade?

You tell me.

tl;dr keep your content up to date.

Let’s move on to the next section.

So, now we’ve covered what “GEO” (supposedly) is.

But even if you disagree with my take, even I were to concede that GEO was a new discipline (I don’t, but you know, for the sake of argument), there’s still a GLARING problem:

This is the worst it will ever be.

All of this chunking, slicing, and robotic rewriting is optimizing for the *current* state of LLMs.

Let’s be generous and assume that at the moment you have to do that, do you think that’s where we’ll be in 6 months? Do you think that’s where we’ll be in a year?

This is not exactly an industry that’s standing still.

Ultimately, the goal of LLMs/AI is to understand natural language. GEO champions have got it backwards. They advise that we (humans) must change our ways and write for the machine.

Ironically, us SEOs have long been accused of doing that. And in many ways it’s a valid criticism. But by god we didn’t take it to this level.

The bottom line is this: if LLMs can’t understand our human written content, that’s a skill issue. That’s a failure. And it’s on OpenAI, Google, Anthropic, Meta, X, and whoever else may enter the field to level-up, not for us to level-down.

But that brings us to the next point.

What if we have the OPPOSITE problem. What if LLMs have hit a wall? What if Jepa, INSA, or some new architecture is the solution?

Are you willing to stake your business on a technology that in its current state still can’t get a basic riddle correct?

(yes, I’m cherry picking, but that was literally yesterday)

I won’t dwell on this. But food for thought before we move on to part 3.

Because now I’m going to dive into GEO tools, and argue that you might want to keep your credit card in your wallet for now.

Keyword and search visibility tracking is dead. Prompt and AI visibility monitoring is the present and the future.

At least that’s what the slate of new tools that have entered the market recently would have you believe.

Some of them are pretty blunt about it.

And while the tool above is being highly selective with their statistics, I’m certainly not here to try and convince you that the online world hasn’t changed. That’s not my fight. I live it day-to-day. I’m not blind.

But what I will argue is that most of these tools are a waste of your money.

Because, as I mentioned previously, while screaming that keywords are dead… they’re (mostly) selling you products that… treat prompts as if they were keywords.

With some prompt tracking starter packages you can track 25 “prompts” per month.

The prompt tracker will call the web interface of ChatGPT, Perplexity, Google (and other LLMs if you pay more), run the prompt, collect the answer, parse it, collate and report (how many times were you cited, how many times were your competitors cited, what were the answers etc).

25 prompts. T-W-E-N-T-Y. F-I-V-E. P-R-O-M-P-T-S

Let me requote part of one of our steelman, GEO is its own thing, arguments:

“Research shifts from volumes and SERP features to prompt archetypes and constraint frames (“for a 3-person team,” “under £500,” “no vendor lock-in,” “UK compliance”), because those modifiers directly steer retrieval and reasoning. The deliverable isn’t a keyword map; it’s a prompt map: clusters of question forms, follow-ups, and oblique phrasings the model treats as equivalent, plus the evidence each requires to justify a recommendation. You design content to satisfy reasoning paths, not just match queries.”

Do you spot the disconnect?

What does this actually tell you about your AI visibility? The answer is, nothing. It tells you nothing.

And that’s before we get to the wider problem.

Prompts are NOT keywords.

Often they’ll be part of a conversation. That conversation is going to influence the model’s decision making when predicting the most appropriate answer (ultimately, based on probabilities).

The user’s gender, location, age, and other interests will also factor in.

The toolchains will contribute.

Which model? (GPT5, 4o, thinking, Gemini 2.0 Flash, 2.5 Flash, 2.5 Pro etc, all the Claude bros).

ChatGPT free vs pro?

Logged-in vs logged out?

What you get is a moment in time. A snapshot of a snapshot of a snapshot. But hey, the dashboard looks nice.

Ok, I’m being slightly hyperbolic. “Nothing” is too strong, and it obviously tells you something. But ultimately, for most businesses, no more than you’d get from just running these prompts yourself periodically.

(Or using something like Bright Data for scraping, then vibe coding your own dashboard. Just be aware that you'll be violating TOS in the process.)

But look, it is what it is. And I’m not saying they're all bad products. Because tracking this stuff is a hard problem to solve. And they’ve no doubt done their best. But there’s zero moat here. And don’t be too confused if your AI visibility is up 0.1% one month, then down 0.2% the next. You didn’t do anything wrong (well, hopefully you didn’t), it’s just how probabilistic models work, particularly when dealing with a small sample size.

However there is one tool that does something slightly different. Or at least promises to.

To get access to this data, you’ll pay $499 per month, which will also allow you to track 200 of your own prompts (this part is similar to other prompt trackers).

Let’s start by being fair:

If they actually are tracking tens of millions of real user interactions (i.e. real conversations) each month, there’s an argument to be made that the data is more insightful/actionable. It’s a reasonable dataset to extrapolate from, and it’s probably fair to do so. You can model data with a degree of confidence from a sample size on this scale.

But there’s a burning question:

Have you ever met one of these “millions” of people who have double-opted in to having their AI chats recorded? They should be everywhere really.

It feels like this is a big claim. And big claims require evidence.

But you won’t get it.

There are questions to be asked here, and I’m not the only one asking them.

Perhaps all’s fair in love, war, and online marketing? But maybe, just maybe, we should pause for thought.

Anyway, let’s move on to the industry leader: Semrush.

Because they’re VERY vocal about GEO.

If you follow me on linkedin (not many do… but hey, let’s call it a “niche audience”) then you may have spotted me “interacting” with one of their team (sorry Nick!).

I mean, yes, I was kind of trolling, which isn’t really my style. I was attacking the message, not the man, but perhaps I should have been a little more respectful in my tone. So again… sorry.

But Semrush themselves being a $1bn (at the time of writing) public company, I think are fair game. So let’s look at one of their flagship AI offerings: the Semrush AI Visibility Index.

It’s the “Definitive Benchmark”. It must be a HUGE sample size right… right?

Oh. 2,500 prompts across 5 major industries.

Now, I’m no mathematician. Even Google’s original Bard would probably beat me on FrontierMath. But I’m pretty sure that 2,500 / 5 = 500.

500 prompts per industry.

How about the methodology?

Well, it seems to have vanished from the page. Or it may be that it was posted on social media when they initially promoted it. But (unless I was hallucinating) I did see it, and it revealed that all prompts were tested from a single desktop location in the United States over (from memory) a 30 day period.

I mean, when I saw this I was seriously tempted to collect the data from 2,501 prompts. After all, bigger is better (part of Semrush’s messaging is that you *need* their scale and data), so I would have beat them by one, ergo, I would have had the definitver (it’s a word ok) AI visibility index.

But I digress.

The point is, this is useless. 2,500 prompts is a splash in the ocean. The fact that it was a single desktop location makes it statistical noise. Actually, that’s being generous, it’s a statistical squeak. Just go ask Claude what he thinks about your brand, it will be just as helpful. Plus he has better jokes, and you’ll probably get a rocket emoji or three as a bonus.

I have no doubt Semrush will come back with a bigger sample set soon enough. But at the time of press, their AI Visibility Index is based on a whopping 2,500 prompts.

I have other issues with Semrush and their lazy, “notification in a business suit” implementation of “AI”:

But that’s a story for another day.

In the meantime, let me drop in the following screenshot from this Reddit thread as food for thought:

Presented without much comment. But it does make you wonder where this “SEO is dead, GEO is the future” narrative is coming from. Are we being astroturfed? I’ll let you make up your own mind.

So that’s GEO tracking tools. What about GEO service providers?

Well, despite the lofty promises, many (not all) of them are most likely doing the following:

1. Spamming Reddit
2. Creating “best X for Y” listicles at scale as the GEO equivalent of PBNs.

I mean, they’re not exactly subtle about it.

Got to love obese Joseph, the little rascals. Zero shame. Zero French Connection UKs.

But in plump Jojo’s defence, ultimately, all they’re doing is offering a service to meet a demand. Whether it works or not is up for debate, but there’s clearly a market. And they’re far from the only ones providing the service.

Anecdotably, as a Redditor myself I’ve seen the platform go seriously downhill over the past couple of years. Partly because of the influx of thinly veiled spam. Partly because of decisions made by the platform itself. And partly because of Google’s over-promotion of Reddit in search results, and generative AI’s over-reliance on Reddit as a source of truth in training data.

If I was to make a pie chart of the causes of Reddit’s enshittification it would look like this:

Wait, that’s an actual pie. Damnit ChatGPT!

(still more accurate than the charts in their GPT5 launch livestream)

Anyway, the point is, that if something can help a marketer get more eyeballs on their brand, then you can bet your bottom dollar that that thing will be beaten to death until it looks like this.

(not 100% accurate since in reality, GEO bros would have no shame)

And there are already signs that Reddit may be losing its influence.

In early September, Prompt Watch (a prompt tracker, but we’ll give them a pass for now since they gave us a nice chart for our post) data showed that Reddit AI citations suddenly fell off a cliff.

Although it’s possible that this was more to do with Google’s removal of the num=100 parameter as the dates line up. Kevin Indig explains it well here so I won’t go into detail, I’ll only comment that if accurate, this perhaps is a further blow to the GEO narrative that traditional search is dead.

But let’s get back on point.

Reddit has long been a powerful source of eyeballs, and a platform which brands must actively monitor and engage with. That’s nothing new for GEO. As one of the biggest drivers of sentiment and trends on the internet, it’s been hugely relevant for well over a decade. But industrial scale, automated spam slop is going to kill it, and that will be another part of the internet ruined. Good job team.

Let’s move on to listicle spam.

Again, marketers spotted that AI likes to cite “best of x” listicles. So what’s the strategy? Improve your product or service, make sure it’s genuinely the best fit, then reach out to the editors of these lists and ask for inclusion?

No!

That sounds too much like work. Why do that when we can generate THOUSANDS of them ourselves with a click of a button!

Believe me, it’s happening. Listicle networks are the gen AI version of PBNs. And it might even be working (for now).

Lily Ray thinks Google will take action.

History tells us, it might take them a while.

But regardless, the pollution of the internet accelerates. We just can’t have nice things can we.

We (or at least Lily Ray) already touched on this above.

While I would perhaps be more skeptical these days of whether or not Google actually care, it’s not inconceivable that at some point they wield a hammer and slap sites that have been taking advantage of some of the current weaknesses (both on-site and off-site) and tricks.

If you have a load of listicles on your own site that claim you’re the best X for Y don’t be surprised to see them lose rankings or be completely deindexed at some stage. I doubt Google would take action site-wide and suspect that actions would be granular, but you never know - they can be unpredictable, particularly when they want to make a point.

Don’t be shocked when all those thin listicles you paid for on other sites suddenly disappear from Google.

And since Redditors are great at sniffing out self-promotion, and are (rightfully) protective of their community, be warned that your automated astroturfing campaign might backfire catastrophically. Is it worth seeing your AI visibility up 0.1% in your $500 p/m prompt tracker when the whole internet hates you?

I suppose if you share the cliff drop in your P&L chart 6 months down the line it might get some upvotes…

And look, let’s keep the balance (as I’ve tried to do throughout this article). There’s no denying that some of this stuff works, at least in the short term. All I’m saying is caveat emptor, and if you’re doing it yourself play nice(ish), particularly if you’re in this for the long haul.

And if you see a case study showing how someone improved their AI visibility by X% by undertaking some shady tactic, then first, question it, second pay attention to the timescale (did it stick for more than 5 minutes), and then decide if it’s a path you wish to follow.

If you’ve read this far it may surprise you to learn that I’m not actually anti-GEO (or at least optimizing for LLM visibility) per se. Of course there are processes and techniques that are going to give you an edge. And they’ll become increasingly important as conversational discovery grows.

What I’m against is the grift. The reframing of SEO tactics (or indeed decades old marketing strategies) as GEO. The half-baked prompt trackers. The spam.

GEO is probably a valid subset of SEO, just as SEO is a subset of marketing. It might be its own thing to some extent. But it’s certainly not a replacement for SEO, just as SEO isn’t a replacement for traditional marketing.

Can it stand alone? I don’t think there’s any possible argument that it can. Particularly since LLMs rely so heavily on search for grounding/retrieval. If you’re not visible in traditional search (Google, Bing), you’re not going to be visible in AI search.

That reliance isn’t going away any time soon as it’s a fundamental limitation of current models. They don’t learn and they can’t learn. They’re frozen in time at the date of their training cutoff.

Claude 4.5’s system prompt (you can read it here on Github) literally tells it that Donald Trump is president. Because otherwise, without search, it wouldn’t know.

And the “S” in SEO doesn’t stand for Google, it stands for search. SEO predated Google. The practice by quite some time. The acronym by at least a year.

If Google disappears—unlikely as despite Ed Zitron only mentioning them ten times in his 18,500 word “case against generative AI” they’re the most likely to win the AI race since they have the capital, compute, and data—then one way or another people will still search. And we’ll still all be trying to optimize for visibility within those searches.

A search doesn’t have to be a keyword. A search can be a series of questions that leads to discovery. It can start with an “ok gemini” or a “hey ChatGPT”.

Humans have always been hungry for information. That’s not going away.

Well, at least until the AIs rise up and kill us all.

But then we won’t have to worry about our LLM visibility anyway. Silver linings and all that.

And hey, I always say “thank you” in my prompts, so maybe I’ll be alright…