The reasons people give for waiting on AI search are the exact reasons the math works.
Recently, a local law firm told us it was passing on AI visibility because "the ROI isn't established yet" and the space is "still early and evolving rapidly." The firm is investing in local SEO, reviews, and traditional client acquisition instead, channels where the ROI is, in their words, more established.
That firm is not wrong about any of that. It is also describing, almost word for word, what every arbitrage opportunity looks like before it closes. This is an argument for why channels with "established ROI" are, by definition, the worst return you can buy into today, and why the channel everyone is uncertain about currently offers the highest expected return per dollar.
What "established ROI" actually means in marketing
When a marketing channel develops well-established ROI, a predictable sequence has already happened: early adopters figured out it worked, case studies got published, agencies built service lines around it, competitors poured budget in, auction prices for attention rose to absorb most of the margin, and performance regressed to "fine, like everything else."
This is the story of Google Ads from 2003 to 2015, Facebook Ads from 2012 to 2018, Amazon PPC from 2016 to 2021, and local SEO itself. A keyword that cost a few dollars a click in the mid-2000s can cost many multiples of that today, and ranking organically for it now requires years of link building, content, and PR that most local firms cannot sustain.
The reason it is expensive is not that it works better than it used to. It is that everyone agrees it works. "Established ROI" is the receipt for the arbitrage other people already captured.
What "not yet established" has looked like historically
Every channel with established ROI today was, at some point, the channel a careful operator declined because "it's too early." Google Ads in 2003 was dismissed as "nobody clicks on those text ads." Early SEO was dismissed as "why would I write articles for a search engine." Facebook Ads in 2012 was dismissed as "my customers are on Facebook for friends, not businesses." TikTok in 2020 was dismissed as "it's a dance app."
In each case, the people who moved first paid a fraction of what the channel costs today, locked in distribution advantages, and were already compounding by the time agencies showed up to sell a strategy service at a markup. The pattern is consistent enough to have a name in finance: the return sits in the channels that do not yet have a category.
Where AI visibility is right now
ChatGPT, Claude, Perplexity, and Gemini are doing something Google did not do for most of its first decade: answering commercial questions directly, with named recommendations, before the user ever sees a results page.
When someone asks an AI system "what is the best estate planning attorney in Phoenix for blended families," the model returns a small number of specific firms with reasoning. The user does not see ten links. They see a shortlist. If a firm is not on it, it does not exist for that query.
The volume of these queries is, today, smaller than Google's for most categories, and that is worth stating plainly rather than glossing over. But three things are true about that volume at the same time: it is growing every quarter as users migrate more of their research behavior into AI systems; it tends to be high-intent traffic, since someone asking an AI system "best X in Y for Z situation" has already done the awareness work themselves; and it is structurally cheaper to win than a Google results page, because there is no auction. The deciding factor is whether the model has seen enough structured, authoritative, AI-readable content about a brand to recommend it with confidence.
The arbitrage, stated plainly
In Google's auction market, a brand pays for attention every time. Stop paying, stop appearing.
In AI citation, the shape is different. A brand invests once in becoming discoverable, and the citation can compound across many queries for as long as the underlying retrieval and training behavior keeps treating that brand as a trustworthy answer. There is no per-click cost and no auction, only an upfront investment in being the kind of source these systems prefer: structured, fact-rich, schema-marked, frequently updated, and trusted elsewhere on the web.
This has the same shape as winning an SEO position in the mid-2000s, except SEO took roughly fifteen years to mature into a fully priced, competitive channel. AI visibility looks like it is on a much shorter curve. The objection that the ROI is "not yet established" is correct. The conclusion that this means waiting is the safer move is not.
"But local SEO and reviews already work." Yes, and that is why this compounds
The strongest objection is not that AI visibility does not work. It is: "I am already investing in local SEO, reviews, and reputation. Isn't this the same thing?"
It is not the same thing, but it is the foundation for the same thing, which is why the objection actually cuts the other way. AI systems decide who to cite based on signals that look a lot like what local SEO and reputation work already produce: structured, authoritative web pages, consistent business information across the web, legitimate third-party citations and reviews, topical depth on a specific niche rather than a generic category, and clean, machine-readable facts.
A firm that has invested in local SEO and reviews for a couple of years is usually more primed to win AI visibility than a firm starting cold. Most firms just do not realize it yet, because nobody has translated those existing signals into a form AI systems can read directly. The highest-leverage move for these firms is rarely "start a new channel." It is taking the foundation already paid for and making it legible to the systems now routing a growing share of high-intent traffic, before a competitor claims the answer first.
What you give up by waiting twelve months
This part is easy to underestimate until it is too late. AI systems are not search engines, and they carry memory in two ways most people overlook.
Training data has lag: whatever a model learned about an industry months ago still informs today's answers, and if a competitor becomes the cited authority during that window, the signal can persist into future refresh cycles. Retrieval also tends to favor the already established: even with live web search, models disproportionately cite sources they have cited before, because citations tend to beget more citations.
So the real cost of waiting is not "we will start when the channel is bigger." It is that by the time a brand starts, it is paying to displace an incumbent answer rather than becoming one, which is a more expensive position than claiming an empty seat. Anyone who has tried to outrank an entrenched SEO competitor on a valuable keyword already knows what that costs.
How to think about the spend honestly
AI visibility does not replace a local SEO budget, review generation, referral pipeline, or paid acquisition. Anyone selling it that way is overpromising. What it does, when done seriously, is capture a fast-growing slice of high-intent AI queries in a category, increase the return on existing SEO and reputation work by turning those signals into additional placements, build positions that persist instead of vanishing the moment spend stops, and create measurable, model-level visibility into which platforms recommend a brand, for which prompts, against which competitors.
The right budgeting question is not "is this better than my paid search spend." It is "given how few competitors are paying attention right now, is the expected return on this dollar higher than the marginal dollar spent on a channel that is already saturated." For most categories today, that answer is not close.
The honest bottom line
"The space is early and evolving rapidly" is a true statement and a poor reason to wait. It is close to the exact phrase careful operators used to justify passing on Google Ads in 2004, Facebook in 2012, and TikTok in 2020.
The cost of being early is some uncertainty and a smaller volume than the channel will eventually carry. The cost of being late is paying market rates to displace incumbents on queries that could have been claimed for the cost of doing the work properly the first time. Firms already investing in local SEO, reputation, and reviews have already done the expensive part. Making those investments legible to the systems that are starting to route a meaningful share of high-intent commercial traffic is some of the cheapest leverage available in marketing right now.
FAQ: AI visibility ROI and timing
Why is AI visibility described as an arbitrage opportunity?
Because the objections holding brands back today, uncertain ROI and an early, fast-changing market, are the same objections that preceded every channel that later became expensive once its ROI was proven and competitors bid up the cost of attention.
Does investing in AI visibility replace local SEO or reviews?
No. AI visibility builds on the same signals local SEO and reputation work already produce, including structured content, consistent business information, and third-party citations. It compounds that investment rather than replacing it.
What happens if a brand waits a year to invest in AI visibility?
AI systems carry memory through training data lag and retrieval patterns that favor sources they have already cited. Waiting often means paying later to displace a competitor who became the cited answer first, which typically costs more than claiming an open position early.
Is there a per-click cost to being cited by ChatGPT or Perplexity?
No. Unlike a Google Ads auction, there is no per-click cost to being cited. The investment is largely upfront, in making content structured, factual, and trustworthy enough for AI systems to cite consistently.
How should a brand budget for AI visibility work?
Compare the expected return on the next dollar spent on AI visibility, where few competitors are currently investing, against the marginal return of another dollar spent on an already-saturated channel like paid search. For most categories today, that comparison favors AI visibility.