For many years the search results page kept the same skeleton: a few ads at the top, ten blue links underneath. AI-generated answers are changing that skeleton. The first thing a user sees is no longer a list of links but a block of text answering their question. The consequences of that change on the organic side have been discussed at length. The paid side has barely been written about in Turkish-language sources at all.
This article tries to fill that gap — but it starts with a warning. Product availability in this area changes quickly and differs from market to market. Below there is no claim of the "on this date this feature launched in this country" kind, because while this article was being prepared we had no way of verifying such a claim from a primary source. What is described is not product announcements but structure: when the answer text moves to the centre of the results page, how the position, the function and the measurement of an ad change. Once you understand the structure, which feature arrived when becomes secondary information.
When the answer itself is the results page, where does the ad land
On a classic results page the ad's job was clear: the user was scanning a list, and the ad sat at the top of that list. The competition was for position. When the answer text moves to the centre of the page, that logic breaks, because there is no list being scanned; there is a paragraph being read.
In that situation the ad can land in one of three places: above the answer, below the answer, or inside the answer's flow as a separate block. Each means something different in terms of user behaviour. An ad above the answer appears before the user has read the answer — it draws attention but arrives without context. An ad below the answer appears after the user has read it and has begun thinking about the next step; it meets a user with more mature intent, but that user may already have got the answer to their question. A block placed inside the answer carries the highest visibility and, to the same degree, requires being consistent with the tone of the content.
The advertiser has no way of choosing between these placements individually; placement is the platform's decision. But that does not mean the decision has no effect on you: knowing the context in which your ad copy will be read determines how you write that copy.
What changes is not the rank but the moment the user is in
In classic search, the user the ad met had not yet found an answer to their question. On a page carrying an answer text, the ad meets a user who has already received the answer. That is a different point in the funnel.
The consequence: on informational queries such as "what is", "how do I", "how long does it take", the ad's job gets harder, because the user's need has already been met. On action-oriented queries such as "who do I buy from", "who does this near me", "book an appointment", the ad's job gets easier: the answer text informs the user but leaves them at the point where they have to choose a supplier.
The budget-side equivalent is direct: a click paid for on an informational query becomes more fragile than before in a layout that carries an answer text. Action-oriented queries hold their value, and become more critical still as competition heats up. That split was something an ad account should have had anyway; the new layout makes it impossible to postpone.
We dealt with how to measure demand in an environment where clicks themselves are declining in our article on marketing and demand measurement in zero-click search; this piece is the paid-channel leg of that picture.
The figures circulating about lost clicks, and why they are not repeated here
More than one independent study has been published on AI-generated answers reducing organic clicks, and news outlets relay the figures from those studies. The figures differ markedly from study to study: as the sample, the query type measured, the device mix and the measurement date change, so does the result.
That is why no single percentage is given in this article. Writing a number we did not measure ourselves, without being able to relay its source exactly, gives the reader the impression of verified information. Stating the direction, on the other hand, is honest: independent studies give a consistent signal that clicks to traditional results fall on queries where an answer text appears. The size of the loss depends on what happens in your sector, on your queries, and you can only read that from your own data.
A second observation is also in circulation: that ad click-through rate is shaken less than organic clicks in this layout. That is a plausible hypothesis — the ad block is preserved as its own area on the page, and the user reading the answer text still comes across it. But we do not have enough evidence to present it as a general rule; the way to see whether it holds on your account is the measurement section below.
Ad copy now has to speak the same language as the answer, not the query
The basic rule of classic ad copywriting was simple: repeat the word the user typed in the headline. On a page carrying an answer text that rule falls short, because the user has already read that word.
The approach that works is to fill the gap the answer leaves behind. The answer text usually explains the "what"; what remains are the questions of "who", "in how long" and "on what terms". Aiming ad copy at those gaps turns it from a repetition of the answer into its continuation.
In practice that means three changes. First, taking generic definitional sentences out of the copy: a headline like "What is teeth whitening" is worthless to a user who has read the answer. Second, leading with distinguishing conditions: working hours, coverage area, delivery time, payment options, warranty scope — concrete details the answer text cannot supply. Third, building the landing page on the same logic: the page should open on the assumption that the visitor already knows the subject and move straight to decision information.
This is a logic already applied on the content side, carried over into ad copy. It is the same framework we describe on our GEO agency page for how content is structured for answer generation; the difference is that here the text appears in paid space.
Why separating brand queries from informational queries has become mandatory
In many accounts, searches on the brand name sit in the same campaign as service searches. That was already unhealthy in a layout with no answer text; in the new layout it becomes actively misleading.
Here is why: on brand queries the user is already looking for you, click-through rate is high and cost is low. On informational queries the answer text comes in between and performance drops. When the two groups are collected in one campaign, the good numbers from the brand side mask the bad numbers from the informational side. The average looks fine while the area where money is being lost stays invisible.
A three-way split is enough in practice: brand queries, action-oriented service queries and informational queries. Keeping the three in separate campaigns makes it possible both to divide the budget deliberately and to see which group the new layout is affecting and by how much. Without that split it is not possible to answer the question "did AI answers affect our budget".
The measurement side: how you see this change in your own account
This is the most concrete section of the article, because this is what can be done today without waiting for product announcements. The aim is to test the general claims arriving from outside against your own data.
Three measurements are read together.
The trajectory of impressions and click-through rate by query group. Once you have built the three-way split above, track each group's impression count and click-through rate monthly. If the answer text is having an effect, it will show up earlier in the informational group. A single month's data means nothing; you need a series covering at least a few months.
Reading it alongside the organic side. Pull the organic impression and click data for the same queries from search console. A picture where organic clicks fall while impressions hold is the most legible sign that an answer text has come in between. Putting the paid side's behaviour over the same period next to it makes it possible to discuss which way the budget should shift.
Whether your brand appears in the answer texts. This is not measured by the ad dashboard; it requires asking assistants and answer-generating search interfaces regularly with a fixed question set and recording the results. We wrote the method out as five metrics in our article on how to measure AI visibility.
Read together, the three produce a picture that says more than any single metric could: whether demand has disappeared or merely moved, and how much work the paid channel is doing in its new location.
One page, two channels: where GEO and advertising intersect
The least discussed consequence of the answer text moving to the centre is this: organic visibility and ad performance are converging on the same page.
If an assistant or answer interface uses your page as a source while producing the answer, the user sees your brand name inside the answer. If your ad also appears on the same query, the user has seen the same name twice — once as a neutral source, once explicitly as sponsored. We have no measurement of how that repetition affects click behaviour, and we are not presenting it as a gain; but it is clear that running the two channels independently of each other is no longer defensible.
The practical consequences are three: your service pages have to be good enough to work both as landing pages and as citable sources; the distinguishing conditions you use in ad copy have to appear in writing on the same page; and the reports for both sides have to be read in the same meeting. If they are run by separate teams, they at least need to meet over a shared query list.
We covered the organic reflection of this shift on the query side earlier in our article on how Google AI Overviews affect businesses; the difference here is the same question turned into a budget line item. We built out step by step how to derive the shared query list from ad account data in our article on turning ad data into a content and GEO plan.
What to do today, and what to wait on
The difficulty of deciding in this area is that the product side is changing fast. It helps to keep the distinction clear: some work can be done today and produces value regardless of new features; other work deserves to wait.
What can be done today. Splitting brand, action and informational queries into separate campaigns. Moving ad copy away from repeating the answer and towards distinguishing conditions. Rebuilding landing pages on the assumption that the visitor already knows the subject. Setting up a monthly report where paid and organic data are read together over the same query list. Verifying that the measurement chain is sound — independent of the answer-text discussion, no decision taken on broken measurement is reliable.
What deserves to wait. Moving the bulk of the budget to newly announced campaign types. Building a new campaign structure specific to a placement. Opening a separate team or budget line for a feature you cannot yet verify is active in your own market.
The test for the distinction is simple: would this work be useful to your business even if the feature never arrived? If the answer is yes, do it today. If it is no, waiting until you have verified that the product genuinely appears in your own market is cheaper.
What we do not know
Writing out plainly what this article leaves open is more useful than making it look closed.
We cannot independently verify which placement is active in which market, in which language and on which query type. There is not enough public data to establish a general rule about how ad click-through rate behaves in this layout. The effect of appearing as a source in an answer text on ad performance for the same query has not been measured. And the answers to those three questions will keep changing as the product side changes.
The picture on the ChatGPT side is different: there the ad sits in a separate area below the answer, and we covered separately in our article on advertising on ChatGPT whether a business established in Turkey can open a self-serve account today.
That uncertainty is not a reason for inaction. On the contrary, it explains why the measurement section above is written so concretely: in an environment where general claims arriving from outside cannot be verified, the only data you can rely on is what comes out of your own account. We wrote which items we read your own data against in the scope of our ad management service; if you would like to build the picture together, you can get in touch.
Frequently Asked Questions
Can ads be shown inside Google AI Overviews?
On results pages carrying AI-generated answers, ad slots are structured to sit above the answer, below it, or within its flow. Which placement is active in which market, in which language and on which query type is variable, however, and while this article was being prepared we had no way of verifying that from a primary source. The reliable way to learn your own situation is to track your own account's impression and click data by query group; look at your own data rather than at general claims.
Do I need to open a separate campaign for AI Mode ads?
Opening a separate campaign specific to a placement, without verifying that the placement is active in your own market, does nothing except split the budget. The split that produces value today is this instead: brand queries, action-oriented service queries and informational queries should be kept in separate campaigns. That split is useful independent of new features, and it also prepares you in advance to see which group a new placement affects when one genuinely goes live.
If AI summaries are lowering organic clicks, should the ad budget be increased?
Drawing a direct conclusion is not right, because the size of the loss varies markedly by sector and query type. The data needed for the decision is this: the trajectory of organic impressions and clicks over the same query list, read alongside the trajectory of the paid side's impressions and click-through rate, as a series covering at least a few months. In a picture where organic clicks fall but demand persists, a shift towards the paid channel is defensible; in a picture where demand itself is contracting, raising the budget only increases cost.
What does writing ad copy to fit an AI answer actually mean?
It means being the continuation of the answer rather than a repetition of it. The answer text usually explains what the subject is; the user's remaining questions are who, in how long and on what terms. Ad copy has to aim at that gap: concrete, distinguishing details such as working hours, coverage area, delivery time, payment options or scope, instead of generic definitional sentences. The same logic applies to the landing page; it should open on the assumption that the visitor already knows the subject and move straight to decision information.
How do I measure this change on my own account?
Read three measurements together. First, split your queries into three groups — brand, action and informational — and track each group's impression count and click-through rate monthly. Second, pull the organic impression and click data for the same queries from search console and put it next to the paid side; a picture where organic clicks fall while impressions hold is the most legible sign that an answer text has come in between. Third, ask assistants regularly with a fixed question set and record, with dates, whether your brand appears in the answer texts. A single month's data means nothing; you need a series covering at least a few months.