Avinash, OAO and the data contradiction nobody solves
Avinash Kaushik proposes dropping SEO, AEO and GEO and merging everything into OAO, under one scorecard. I agree with the diagnosis and disagree with the alarm: results always counted. What changed is that we lost the data. GA and GSC are the bare minimum, and we can't see what happens inside the LLMs. Instead of building a prompt-tracking system, I move clients from rankings to conversions and attribution. No measurement, no management.
I didn't expect to see Avinash Kaushik in the acronym war. In this week's newsletter, TMAI #530, the man who spent twenty years saying a metric is only worth the decision it unlocks is proposing one more acronym: OAO, Organic Answers Optimization. His idea is to merge SEO and AEO into one thing, because, in his framing, the job is one: influence a machine to recommend your brand and convert that into traffic. He says that, in the end, sales are what count. He's right. But that was always true. That isn't what changed. What changed is that we went blind.
The chair Avinash didn't pull up
Avinash wants to throw out SEO, AEO and GEO and fold everything under OAO, with a single scorecard and three CFO KPIs: non-brand organic search profit, Answer Engine profit, and an AI Brand Score that approximates how often the models recommend you. On the underlying diagnosis I agree, and I've written it before: it's all one machine to influence, SEO feeds AEO and the reverse, and neither the customer nor the CFO cares whether the sale came from old Google or from ChatGPT. But spending the article naming the baby is rearranging the furniture while the house loses its roof.
"Sales are what count": and when did they not?
His point is true and it's old. There was never a decade when sales didn't count. Avinash describes the behaviour shift well: we moved from Search (hunt and pick among ten links) to Answer (ask and get the finished reply) and, for 1 to 2% already, to Complete (let an agent handle the purchase). The shift is real. But presenting "sales are what count" as the revelation of the AI era is the part that doesn't hold. The goal never changed. What changed is that we lost the instruments to read it.
What actually changed: we lost the data
The click decline started before the LLMs. Zero-click was already here, with Google keeping the answer and the click (Avinash himself cites SparkToro: close to two-thirds of US searches end without a click). Then, on top of that hole, a second one opened: we stopped seeing what happens inside the LLMs. It's not one crisis, it's two, stacked. And we only talk about the second because it's new.
GA and GSC are the bare minimum
Let me be blunt: the AI data in Search Console exists so Google can't be accused of having nothing. GA4 follows the same logic. They're the bare minimum. ChatGPT did more by adding a utm_source=chatgpt.com to the link than Google did with its entire AI report in GSC. And even that doesn't last: Avinash himself notes that in mid-October OpenAI changed the referral strings and part of that traffic began landing in Direct, with no source. Building a brand without seeing the LLMs is going back to 2000, when Urchin still sorted everyone's life out (Google bought it in 2005 and turned it into Analytics). The difference is that, in 2000, nobody pretended to have the data.
The contradiction I can't get out of my head
And here's what truly astonishes me. We move giant investments. We train AI that breaks into systems, that does hacking, that solves puzzles stuck for centuries. And those same platforms can't give a brand the basic metrics to manage its assets and decide where to put the money. The machine that deciphers the undecipherable can't tell you where the sale came from. It's a contradiction so large it almost sounds like a provocation. Whoever can do the harder thing doesn't deliver the easier one. Or doesn't want to.
What I do with clients, instead of complaining
Complaining isn't a strategy. With my clients, I'm moving the focus from rankings to conversions and to attribution models that show the consumer journey and, above all, its holes. Rankings became vanity the day half of searches stopped producing a click. Conversion and attribution are what's left of the observable and the honest. This is where my CITAR method lives: in what you measure downstream, not in what you guess upstream. No measurement, no management. With it, even poor, there's a decision.
A prompt-tracking system is the new vanity dashboard
And there's a mistake I see almost everyone about to make: building a system to track prompts. Don't waste your time. An LLM doesn't answer the same thing twice. And note: this isn't me against Avinash, it's Avinash proving my point. His own KPI, the AI Brand Score, he describes as probabilistic, a distribution of answers gathered by repeating prompts, and he concludes that "the monthly search volume of any one prompt is effectively one". If the owner of the metric calls it a distribution, measuring prompt by prompt is measuring noise with decimal places. At BrightonSEO this year I watched the whole room admit, between talks, that AI metrics lie and that there's an attribution crisis. The conclusion I brought back is the same: anchor on the observable, don't buy stage hacks.
Verdict
Avinash's recipe for the future is "more content, more places, more often". It may well be right. The problem is that, without measurement, that's faith, not management, and he knows it better than anyone. While we argue over whether it's SEO, GEO, AEO or OAO, the data still doesn't show up. Whoever waits for the perfect tool will measure nothing for another year. Whoever goes back to the observable, to conversions, to attribution, to logs, decides today with what there is. The hole that matters isn't one of method. It's one of data. And that one is closed neither by OAO nor by CITAR. It's closed by whoever builds the tools that don't yet exist.
Until then, the brand that decides in the dark isn't doing GEO or OAO. It's managing what it can't see, and calling it strategy.