Most companies run these as two budgets managed by two people who never speak: paid search over here, AI visibility over there. That made sense for about eighteen months. It doesn’t any more, and the reason is worth understanding before your next quarter’s budget gets set.
Google has spent the last year rebuilding how ads are selected and written. The short version: ads inside AI-generated results are no longer matched to a keyword you chose and rendered from copy you wrote. They are assembled by the same model that writes the answer, out of the same material it reads to write the answer.
Which means the work that gets you cited organically and the work that makes you eligible for paid placement have quietly become the same work.
What actually changed
Traditional paid search had a clean shape. A person types a query. Ads compete for that string. You measure impressions, clicks, conversions.
AI Mode breaks each step. People ask longer, vaguer questions, refine them across several turns, compare options, and arrive at a decision without ever seeing a conventional results page. There’s no single query to bid on, and often no click to count — Seer Interactive’s analysis of over 25 million impressions found the overwhelming majority of AI Mode queries produce no outbound click at all.
Google’s answer to the bidding problem was to stop asking you for keywords. The newer campaign machinery targets on inferred intent rather than matched strings. Its answer to the creative problem was to generate the ad at serve time — writing custom explainers and conversational responses about your business, on the fly, from your landing pages and product data.
That second change is the important one. The ad is no longer a thing you wrote. It’s a thing the model wrote about you, using what it can find out about you.
The convergence
Now hold that next to how organic AI citation works.
A model deciding which businesses to name in an answer does it by assembling what it has read: your own pages, and — critically — what independent sources say about you. Corroboration from places it already trusts, consistency across those mentions, and clarity on your own site about what you do and for whom.
The inputs to paid selection are now a subset of the inputs to organic selection. Vague service pages, thin descriptions, missing specifics, an unclear category — these used to cost you organic citations while your ads carried on regardless, because your ads were copy you controlled. Now they cost you both. Google has been explicit that incomplete descriptions, missing images and outdated prices reduce your chances of featuring in AI-generated responses.
Put plainly: the quality of your input layer is now an eligibility gate on inventory you are paying for.
Three practical consequences
Your site copy is targeting infrastructure
Not brand expression, not a design deliverable — the raw material a model uses to decide whether to surface you and what to say about you. If you can’t point to the sentence you want quoted back to a buyer, you don’t have one, and something worse will be generated in its place.
Cheap top-funnel traffic is disappearing, and the remaining clicks cost more
As informational queries get answered on the page, budget concentrates on bottom-funnel commercial terms, competition there intensifies, and bid pressure rises. The click you buy is more expensive and more valuable. It also arrives later in the decision, after a conversation you never saw — which means the thing that determined whether you were in that conversation was your presence in the model’s reading, not your bid.
Paid becomes your fastest read on organic
This is the part most people miss. Your ads account is now a live instrument for the same intent surface your organic visibility depends on. Search terms and query data from AI-driven placements tell you the actual language buyers use when they’re reasoning out loud with a machine — not the keyword list someone built in 2023. That’s the highest-quality input available for deciding which questions your corroboration programme should be targeting.
So which one do you fund?
Both, but with the dependency the right way round.
Paid buys you presence now. It’s immediate, controllable, and measurable in a way organic AI citation isn’t yet. It’s also rented, priced against competitors who are all crowding into the same narrowing set of commercial queries, and increasingly gated on inputs you may not have built.
The input layer — corroboration, consistency, legibility — compounds. It doesn’t switch off when the card declines. And it now determines both whether the model names you unprompted and whether it can construct a decent ad about you when you pay it to.
Funding paid without the input layer means paying rising prices for placements you’re only partly eligible for, in a channel where you can’t see most of what happens. Funding the input layer without paid means waiting for compounding to arrive while competitors buy the interim.
The sequencing that works: fix the input layer first because it gates both, then use paid to buy presence and, just as importantly, to tell you what the input layer should be aimed at next.
The test
Ask whoever runs your ads account one question: which of our placements are appearing inside AI-generated answers, and what does the model say about us when it writes them?
If nobody can answer, that isn’t a reporting gap. It’s the whole channel.
We measure both sides. Where you’re named in AI answers, where your competitors are named instead, and whether your paid spend is working with that or against it. Book a consult.
