Search
GEM: Generative Engine Marketing
AI assistants and AI-powered search are becoming a new medium where advertising will also have a place, but the rules are still being written. With GEM we get you ready for it, build the test infrastructure and track how your brand is perceived inside AI.
Short answer
GEM, or generative engine marketing, is the marketing discipline that covers using paid or sponsored placements within AI assistants and AI-powered search experiences, and managing the brand perception that forms in these engines. Because the field is still developing, most of the work today is preparation, controlled testing and perception monitoring.
What is GEM, and how does it differ from GEO?
GEM is the marketing work that manages paid visibility within generative engines and the perception of your brand that forms in them. GEO looks after the organic side: your brand being mentioned in AI answers of its own accord and with accurate information. GEM, on the other hand, deals with clearly labelled advertising and sponsored spaces around or within those answers.
The relationship is similar to the one between SEO and search ads in classic search. Organic visibility provides a lasting foundation, while a paid placement buys visibility for a particular query or period. Neither replaces the other, and the healthiest outcome usually comes when they are planned together.
We have explained how the five search disciplines differ in this context in our article on the differences between SEO, AEO, GEO, AIO and GEM. On this page we focus only on paid placement and brand perception.
Where does advertising in generative engines stand today?
Advertising in generative engines is at an early and fragmented stage today. Some platforms are testing, or offering in limited markets, formats such as sponsored content or sponsored follow-up questions beside or below the answers; others do not accept advertising or have not yet firmed up their plans.
On Google's side the picture is a little more concrete: ad placements linked to AI summaries are offered in some markets through existing Google Ads campaigns. Which markets, languages and campaign settings are eligible changes over time, so we check the current situation at the start of every engagement.
On this page we do not speculate about when a particular platform's ad product will launch in Türkiye or how it will be priced. Any date or format information given without a confirmed announcement could steer your budget decisions in the wrong direction.
| Area | What we see today | What is still unclear |
|---|---|---|
| Placement format | Formats such as spaces labelled as ads beside or below the answer, and suggested questions, are being tested | Which formats will last, and whether a common standard will emerge across platforms |
| Targeting | Placements rely largely on the context of the question and the conversation | How keyword, topic or audience targeting will be combined |
| Measurement | Clicks and conversions can be measured with tools similar to those of classic digital advertising | How impact that happens without a click will be reported |
| Availability | Many formats are limited to certain markets or certain advertisers | Timing of launches in Türkiye and other markets |
| Policy | Platforms stress that ads are separate from the organic answer and labelled | The detail of sector-specific restrictions and content rules |
What should you prepare before the medium opens up?
Preparation means building, now, the infrastructure that will let you run a measurable test from day one when the medium opens up. The early period of any new ad inventory is valuable for learning, but that value can only be used if the data, creative and measurement are ready.
- Clean product and service data: your product catalogue or list of services is up to date, complete and consistent. In shopping-focused answers, product data feeds are becoming increasingly decisive.
- Solid conversion measurement: form, call and sale events are defined correctly, and server-side measurement and, where relevant, a CRM connection are in place.
- Conversational creative: short copy and descriptions written in a question-and-answer voice that respond to the user's question.
- Landing pages that follow the question: greeting users arriving from an ad with a page that is the natural continuation of the question they asked.
- A decision on the testing budget: setting aside a separately managed testing allowance for new channels that does not put the main campaigns at risk.
- Brand safety and regulation: writing down in advance the contexts in which you do not want to appear and the advertising restrictions that apply to your sector.
In regulated sectors such as health, finance and law, ad copy is already subject to strict rules. We assume these rules will also apply to placements in generative engines, and prepare copy that complies with regulation and makes no medical or financial promises.
How is the test infrastructure built?
The test infrastructure is a measurement set-up that can tell whether spend in a new channel is genuinely producing an additional result. At its heart is the idea of incrementality: what an ad adds on top of the conversions that would have happened anyway without it.
- Separating campaigns in the new channel from other channels with distinct naming and distinct tracking parameters.
- Defining the success criteria in writing before the test starts, and making clear which outcome means carrying on and which means stopping.
- Using a comparison group where possible: not showing the ad in one region, period or audience segment and measuring the difference.
- Tracking indirect effects such as branded searches and direct traffic throughout the test period.
- Assessing the results not with a channel report but with a shared report that shows the entire marketing mix.
We build the measurement side together with our CRO and analytics team. That way, when a new inventory opens, we are not starting from scratch but adding a new channel to the existing measurement set-up.
How is an ad inside a conversation written?
An ad inside a conversation is written to offer a useful continuation of the question the user is asking at that moment. A user talking to an assistant is expecting an answer; a message that interrupts that flow, has nothing to do with the question or exaggerates will neither get noticed nor do anything but damage trust in the brand.
In classic search advertising, short headlines and keyword matching come to the fore. In placements within generative engines, context is decisive: the same product should be offered with a different rationale to a user asking about budget and to one asking about quality. So rather than a single piece of copy, we build creative preparation around question scenarios.
- An opening sentence that makes clear which of the user's questions is being answered.
- Concrete, verifiable information: the scope of the service, delivery conditions, the area served.
- Language free of exaggeration, claims of superiority and promises that cannot be verified.
- A clear next step that takes the user to a page that naturally continues their question.
How is brand perception in AI tracked?
Brand perception in AI is tracked by putting the same set of questions to different engines at regular intervals and recording how the brand is described. Where GEO measurement focuses on whether the brand is mentioned and cited as a source, perception tracking looks at the content of the description.
- Tone: is your brand described in positive, neutral or hesitant language?
- Positioning: with which features, as suitable for which audience and alongside which competitors is it mentioned?
- Accuracy: is the information on products, services, price range and location current and correct?
- Recommendation context: under what conditions is it recommended, and under what conditions are users steered to other options?
- Recurring objections: if a complaint or shortcoming keeps coming up in the answers, where does it come from?
When a problem with perception is spotted, the fix is usually not advertising but correcting the sources: updating old information, filling in missing pages, and passing on a genuine problem in the customer experience. A paid placement cannot correct a message that contradicts the organic description; at best it sits beside it.
Who should already have GEM on the agenda?
GEM should already be on the agenda for brands that set aside a regular budget for search ads and whose customers research at length before deciding. In comparison-heavy categories such as e-commerce, B2B services, education, real estate and health tourism, some user questions are shifting to AI assistants.
For businesses that are new to digital marketing or whose basic measurement is not yet in place, the priority is different: SEO, search ads and the measurement set-up come first. Today GEM should appear in the marketing plan as an experimental line; it should take not the place of the main budget but part of the learning allowance.
How do we work with uncertainty?
In an uncertain field, the most valuable thing is honest information. We follow platform announcements and official documentation, and do not make recommendations based on unconfirmed rumours. If a format is only being tested in certain markets, we say so, and we do not ask for budget on the promise of early access.
In these channels, what we commit to is thorough preparation, tests run against criteria written down in advance, and open reporting of what has been learned. We agree each test's target metric and budget limit together before it starts.
Where should you start with GEM? Our recommendation
If you set aside a regular budget for search ads, start GEM with preparation, not spend; if your basic measurement is not yet in place, SEO, search ads and measurement come first. Running a measurable test from day one when the medium opens up depends on preparation done today. This is the order we recommend:
- Record the perception: Prepare a question set specific to your category and put the same questions to different AI assistants; note the tone used about your brand, how accurate it is and which competitors it appears with.
- Fix the sources first: If you see old or wrong information in the answers, find where it comes from, update old pages and fill in missing ones; a paid placement cannot correct a message that contradicts the organic description.
- Get data and measurement in order: Make your product catalogue or list of services up to date, complete and consistent; make sure form, call and sale events are defined correctly.
- Write the test rules: Set a separate testing allowance that keeps the main campaigns safe; write down the success criteria and stopping point before the test starts, and give test campaigns distinct naming and tracking parameters.
- Follow the medium and verify: Keep up with platform announcements and official documentation; do not commit budget until you have confirmed that a placement is genuinely open in your market and sector.
You can follow this order with your own team. If you are in a regulated sector such as health, finance or law, or you advertise across several channels and want to isolate the incremental effect of the new medium, we recommend running the test design and measurement together.
GEM, SEO, AEO and GEO: how is GEM planned alongside them?
GEM manages how AI describes your brand; the raw material for that description is the content and sources made visible through SEO, AEO and GEO work. Below we cover the combinations GEM is most often asked about together.
SEO and GEM: linking search visibility to perception in AI
SEO and GEM are two ends of the same brand story. SEO determines which searches your brand is found for on Google; GEM manages how AI assistants describe and recommend your brand. Because assistants rely heavily on web pages and external sources when they describe a brand, the content strengthened by SEO is also GEM's raw material. Running the two together lets you measure search traffic and AI-driven demand against the same goals. More detail: SEO.
AEO and GEM: the right answer and the right narrative
AEO makes sure the short answers given to questions about your brand are accurate and based on your own sources; GEM manages the AI narrative those answers add up to and the comparison with competitors. Clear answers to questions about scope, location, way of working and what sets your service apart are the building blocks assistants use when describing a brand. When we run this pair together, we compare frequently asked questions with how the brand is described in AI and close the gaps. More detail: AEO — answer engine optimisation.
SEO, AEO and GEM: from being found to being chosen
Together, SEO, AEO and GEM cover your brand being found in search, standing out as the right answer to questions and being described in AI answers in a way that makes it the preferred choice. In competitive categories where customers compare their options, this sequence works: first visibility, then the answer, then perception. Measurement is brought together in a single report covering organic traffic, presence in answer boxes, how you are described in AI and branded searches.
SEO, GEO and GEM: from being a source to generating demand
Together, SEO, GEO and GEM aim for your brand to rank on Google, be cited as a source by AI assistants, and turn that visibility into quote requests and enquiries. In this trio SEO provides the groundwork, GEO takes on becoming a source, and GEM handles the narrative and conversion. We track visits from AI as a separate channel and regularly check how the brand is described in assistants using a consistent set of questions. More detail: GEO — generative engine optimisation.
How we work
Readiness audit
We review your product and service data, conversion measurement, landing pages and sector advertising restrictions in terms of readiness for placements in generative engines.
Baseline perception measurement
Using a question set specific to your category, we record how your brand is described on different engines, its tone, its accuracy and its position against competitors.
Channel monitoring
We follow the platforms' ad products, market availability and policy changes, and report on the options that become suitable for you.
Test design
We write in advance the campaign structure, creative, success criteria and comparison method to be used when a suitable placement opens up.
Controlled test and report
We run the test with a separate testing allowance and assess the incremental effect and the change in perception in a shared report.
What we deliver
- Readiness audit for placements in generative engines and a list of gaps
- Summary of the current advertising situation on each platform
- Brand perception question set specific to your category and a baseline report
- Recommendations for conversational creative and landing pages
- A written test plan with success criteria and a comparison method
- A measurement structure set up with separate tracking parameters
- Periodic report on perception tracking and developments in the channel
Get a quote
How does the quote process work?
It starts with a message. We prepare the rest, and no work begins until you have seen in writing what you pay for, why, and how much.
Message us
Tell us briefly about your business, your goal and your website, on WhatsApp or by email.
Free initial analysis
We review your search visibility, how AI answers mention you and any ad accounts you run, and prepare a one-page summary.
Strategy call
We go through the summary together and agree on priorities, goals and the metrics we will track.
Written proposal
We send a proposal that spells out scope, deliverables, timeline and fee. We start once you approve it.
Request a quote
Fill in the form; we will review your goal and current position and come back to you with a written proposal.
Service:GEM — Generative Marketing
Frequently Asked Questions
- Can we advertise on ChatGPT?
- Advertising options in generative assistants vary from platform to platform and market to market; some formats are only being tested in certain countries or with a limited number of advertisers. At the start of the work we check the latest announcements to see whether a particular platform accepts advertising in Türkiye, and we do not plan budget on unconfirmed information.
- What is the difference between GEM and GEO?
- GEO aims to get your brand mentioned organically, with accurate information, in AI answers. GEM manages paid or sponsored placements within these engines and the brand's perception inside AI. It resembles the relationship between SEO and search ads in classic search; the two deliver more consistent results when planned together.
- If there are no ads yet, why start on GEM now?
- When a new inventory opens, the early period is valuable for learning, but to make use of that value the data, measurement and creative need to be ready. Brand perception tracking can also be done today, independently of advertising. Preparation and perception monitoring mean you are not starting from scratch when the medium opens up.
- If AI describes our brand negatively, will advertising fix it?
- Usually not. A paid placement does not remove negative information given in the organic answer; at best it sits beside it. First we find which sources feed that description, correct old or wrong information and, if there is a genuine customer experience problem, pass it on to you.
- How much budget should we set aside for GEM?
- We do not give a fixed recommendation. For now, GEM is an experimental line that should run on a separate testing allowance that does not put the main campaigns at risk. The size of that allowance is agreed together, based on your current ad spend, how important AI assistants are in your category and how mature the placements available for testing are.
- How often should brand perception be tracked?
- We set the tracking interval according to how fast your category moves. Because AI answers can shift with model updates, new sources and changes in search results, a one-off measurement can be misleading. We repeat the same question set at regular intervals and under the same conditions, and take extra measurements after a major launch, a crisis or a price change.
- Is advertising in generative engines possible in the health and finance sectors?
- In regulated sectors advertising is subject to extra rules in every channel, and the detail of those rules for new placements is not yet clear. In these sectors we prepare copy that complies with regulation and makes no medical or financial promises, and we do not publish anything without checking each platform's current policy.
Let us measure
your visibility today.
We map your current search visibility and your standing inside generative engines. Free, one page, real data.
